mm-1289 === Subject: None elementary rotation? Hi. I understand how to simulate a free rotating object in space with inertia matrix, finding torque, angular velocity, and all that. But in many first person game, the camera rotation is limited in the sense that moving the mouse rotates the camera in the global Y axis and the camera's local X axis. Thus the player can look left,right, up,down, but always keep level with the horizon ( no rotation around the local Z axis) I want to know how to simulate this type of rotation by applying some forces/torques instead of directly constructing the moved camera matrix. Some links to further understand the mechanics of freely rotating === Subject: Re: None elementary rotation? > Hi. > I understand how to simulate a free rotating object in space with > inertia matrix, finding torque, angular velocity, and all that. > But in many first person game, the camera rotation is limited in the > sense that moving the mouse rotates the camera in the global Y axis > and the camera's local X axis. Thus the player can look left,right, > up,down, but always keep level with the horizon ( no rotation around > the local Z axis) > I want to know how to simulate this type of rotation by applying some > forces/torques instead of directly constructing the moved camera > matrix. > Some links to further understand the mechanics of freely rotating No idea whether game programmers will need this, but your question is very reasonable: the camera is just one moving object like all the others in world coordinates or object coordinates, how I would like to call these coordinates. Define an object which carries the camera and simulate the view through the on-board camera. Move the object e.g. like an aircraft or a car by applying physical torques and forces. In fact itÇs exactly the task: what can I see by a camera in a remotely piloted vehicle which is controlled by torques and forces instead of defining the camera merely by geometrical position coordinates and angles. === Subject: Re: None elementary rotation? [You forgot to set a F'up2 --- fixed] > I want to know how to simulate this type of rotation by applying some > forces/torques instead of directly constructing the moved camera > matrix. I'm at a loss: why on earth would you want to apply forces / torques or moments of inertia to a *camera view* in an FPS? Please note that the camera being simulated here is essentially the player's eye, which to very good approximation can be considered to have neither weight nor inertia. Applying more physics to games is always a good idea, IMHO, but in the case at hand, the physics is overridden by biomechanics, which amont other things means the camera view *is* actually moved consciously or subconsciously in terms of where do I want to look now?, i.e. direction, rather than let's push those eyes a little to the left and see what happens. -- Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de) Even if all the snow were burnt, ashes would remain. === Subject: Re: None elementary rotation? Well, it makes sense for a spaceship in first person. In this case the camera matrix is the same as the spaceship's matrix (optionally with some local adjustment). Applying force to the ship == to the camera. Then again, the case I posted is different; as you mentioned, its rotation is caused by muscle movement. Nonetheless, since the spaceship case already works beautifully, I was hoping the first person shooting case is just a subset of the problem, so I don't have to write more code. =) > I'm at a loss: why on earth would you want to apply forces / torques > or moments of inertia to a *camera view* in an FPS? Please note that > the camera being simulated here is essentially the player's eye, > which to very good approximation can be considered to have neither > weight nor inertia. > Applying more physics to games is always a good idea, IMHO, but in the > case at hand, the physics is overridden by biomechanics, which amont > other things means the camera view *is* actually moved consciously or > subconsciously in terms of where do I want to look now?, > i.e. direction, rather than let's push those eyes a little to the > left and see what happens. === Subject: Re: None elementary rotation? > Well, it makes sense for a spaceship in first person. No way does it make sense to have a 'no roll' limitation in a space ship. That only makes any sense where there's an external, well-defined zero roll orientation of objects, in other words: a physically enforced up vector, a.k.a. the floor. > Then again, the case I posted is different; as you mentioned, its > rotation is caused by muscle movement. Not just caused, but *controlled* by muscles, with a feedback loop that runs in the background. That's how you can consciously decide what to look at, even if the head is moving, e.g. while walking or moving the head. > Nonetheless, since the spaceship case already works beautifully, I > was hoping the first person shooting case is just a subset of the > problem, so I don't have to write more code. =) It may formally be a subset, but if you treat it like that, that would mean you're wasting CPU cycles. The general plan would be to assign the eye an inertial tensor of diag(0,0,infinity) (with the z axis being its original view direction). But since such a tensor is physically impossible, a simulation designed for real objects will tear itself apart trying to use such an inertial tensor (finite torque would lead to infinite angular velocity...). To get a useful simulation, you'ld have to model the part of the brain that controls eye movements by feeding back images to the muscles that rotate the eye, too. That'll almost certainly not be worth the effort. -- Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de) Even if all the snow were burnt, ashes would remain. === Subject: Re: Mathematician on Iraqi Governing Council? > Ahmad Chalabi is a member of the Iraqi Governing Council. > story of a ground soldier in the CIA's war on terrorism, in which he > Chalabi's family had been forced to flee Iraq for Lebanon in 1958 when > the Hashemite monarchy fell. Thirteen at the time, Chalabi had grown up > abroad, exchanging his Iraqi accent for a Lebanese one [and becoming] > further tainted at graduate school in the US [where] he picked up a > master's degree at MIT ***and a PhD in numbers theory*** [and] learned > to speak American idiomatic English. Iraqis looked at him as a stateless > exile. > I looked for Ahmad Chalabi on MathSciNet and found three papers, > more algebra than numbers theory: > MR0634213 (83e:16022) Chalabi, Ahmad; Khuri, Soumaya Makdissi Pure > submodules of injective modules. Rend. Istit. Mat. Univ. Trieste 12 > (1980), no. 1-2, 15--19. (Reviewer: Annie Page) 16A52 > MR0360669 (50 #13116) Chalabi, Ahmad The Jacobson radical of a group > algebra under field extensions in characteristic $p$. Tamkang J. Math. 5 > (1974), no. 1, 103--106. (Reviewer: S. B. Conlon) 16A26 > MR0325739 (48 #4085) Chalabi, Ahmad Modules over group algebras and > their application in the study of semi-simplicity. Math. Ann. 201 > (1973), 57--63. (Reviewer: S. B. Conlon) 20C05 > Does anyone know if this is the same Ahmad Chalabi? I think it is. According to some accounts he is also a crook. His chances of taking part in actually governing Iraq look vanishingly small. Gib === Subject: Re: Mathematician on Iraqi Governing Council? >> Does anyone know if this is the same Ahmad Chalabi? >I think it is. According to some accounts he is also a crook. His ^^ ^^ in a row... Michele -- > Comments should say _why_ something is being done. Oh? My comments always say what _really_ should have happened. :) - Tore Aursand on comp.lang.perl.misc === Subject: Re: Mathematician on Iraqi Governing Council? >> MR0325739 (48 #4085) Chalabi, Ahmad Modules over group algebras and >> their application in the study of semi-simplicity. Math. Ann. 201 >> (1973), 57--63. (Reviewer: S. B. Conlon) 20C05 >> Does anyone know if this is the same Ahmad Chalabi? > I think it is. According to some accounts he is also a crook. His > chances of taking part in actually governing Iraq look vanishingly small. What accounts? It is outrageous to use the internet to libel a person who cannot defend themselves in this way. Personally, an Iraq governed by mathematicians seems to me preferable to an Iraq governed by Saddam Hussein or George Bush. -- Timothy Murphy e-mail (<80k only): tim /at/ birdsnest.maths.tcd.ie tel: +353-86-2336090, +353-1-2842366 s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland === Subject: Re: Mathematician on Iraqi Governing Council? >> MR0325739 (48 #4085) Chalabi, Ahmad Modules over group algebras and >> their application in the study of semi-simplicity. Math. Ann. 201 >> (1973), 57--63. (Reviewer: S. B. Conlon) 20C05 >> >> Does anyone know if this is the same Ahmad Chalabi? I think it is. According to some accounts he is also a crook. His > chances of taking part in actually governing Iraq look vanishingly small. > What accounts? After I posted the message that started this thread, I had a look on the web for stuff about Chalabi (yes, I should have done that *before* I posted the message that started the thread). I formed the impression that right-of-center sites think he's an OK guy, while left-of-center sites bring up a conviction over the failure of a bank in Jordan. But don't take my word for it, there's lots and lots of stuff about Chalabi out there, you can decide for yourself which sites are credible. > Personally, an Iraq governed by mathematicians seems to me preferable > to an Iraq governed by Saddam Hussein or George Bush. Given a choice between an Iraq governed by Bush and one governed by Kaczynski, or Bieberbach, I might hold my nose & vote for Bush. -- Gerry Myerson (gerry@maths.mq.edi.ai) (i -> u for email) === Subject: Re: Mathematician on Iraqi Governing Council? > I think it is. According to some accounts he is also a crook. His > chances of taking part in actually governing Iraq look vanishingly small. > What accounts? Such accounts are all over the place. Have you ever done an internet search? Do you know anything at all about Iraq and Chalabi? > It is outrageous to use the internet to libel a person > who cannot defend themselves in this way. Poor Chalabi, how will he ever recover from this post in sci.math? > Personally, an Iraq governed by mathematicians seems to me preferable > to an Iraq governed by Saddam Hussein or George Bush. An Iraq governed by Chalabi is George Bush's wet dream. === Subject: Re: Mathematician on Iraqi Governing Council? .91Ä Timothy Murphy .91[CapitalEth].91ñ.93»[EDouble Dot]±.93Ì.91[Micro] .93¡.93.b3.91Ë .91.b9.91¬.91ü.93á[EDo ubleDot].b9.91± > Personally, an Iraq governed by mathematicians seems to me preferable > to an Iraq governed by Saddam Hussein or George Bush. Given the current anarchy and chaos over there, ANY Iraq government is preferable to an Iraq governed by Saddam Hussein or George Bush. > -- > Timothy Murphy > e-mail (<80k only): tim /at/ birdsnest.maths.tcd.ie > tel: +353-86-2336090, +353-1-2842366 > s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland -- Ioannis Galidakis http://users.forthnet.gr/ath/jgal/ ------------------------------------------ Eventually, _everything_ is understandable === Subject: Re: Mathematician on Iraqi Governing Council? >Given the current anarchy and chaos over there, ANY Iraq government is >preferable to an Iraq governed by Saddam Hussein or George Bush. Are you saying Saddam's government would be preferable to itself? Robert Israel israel@math.ubc.ca Department of Mathematics http://www.math.ubc.ca/~israel University of British Columbia Vancouver, BC, Canada V6T 1Z2 === Subject: Re: Mathematician on Iraqi Governing Council? Ì Robert Israel ó.8d.98.87.8b.8c .97.99.95 .92.86.94.9d.92.87 >Given the current anarchy and chaos over there, ANY Iraq government is >preferable to an Iraq governed by Saddam Hussein or George Bush. > Are you saying Saddam's government would be preferable to itself? Hmmm, I was pointing out to the previous poster that the statement a government of mathematicians is preferable..., seems to be unecessarily sharp. Particularly since the lower bound is exactly G. W. Bush (or Saddam Hussein). Can't do worse than that. And since humans often abuse logic in language, any government in my statement meant implictly excluding the lower bound(s). Logical falacy noted, however :*)) > Robert Israel israel@math.ubc.ca > Department of Mathematics http://www.math.ubc.ca/~israel > University of British Columbia > Vancouver, BC, Canada V6T 1Z2 -- Ioannis Galidakis http://users.forthnet.gr/ath/jgal/ ------------------------------------------ Eventually, _everything_ is understandable === Subject: Re: Mathematician on Iraqi Governing Council? >.93 Robert Israel [NonBreakingSpace].8b.96.87À.8c .97.99.95 .93[CapitalThorn].92.9b.93.87 >>Given the current anarchy and chaos over there, ANY Iraq government is >>preferable to an Iraq governed by Saddam Hussein or George Bush. >> Are you saying Saddam's government would be preferable to itself? >Hmmm, I was pointing out to the previous poster that the statement a >government of mathematicians is preferable..., seems to be unecessarily >sharp. Particularly since the lower bound is exactly G. W. Bush (or Saddam >Hussein). Can't do worse than that. And since humans often abuse logic in >language, any government in my statement meant implictly excluding the >lower bound(s). >Logical falacy noted, however :*)) I guess one danger with a government of mathematicians is that they might get so caught up in logical nitpicking that they forget about governing the country. But there is one case I know of a government led by a mathematician: Painlev'e was twice prime minister of France. And there's one case I know of a prime minister who was once married to a mathematician. Robert Israel israel@math.ubc.ca Department of Mathematics http://www.math.ubc.ca/~israel University of British Columbia Vancouver, BC, Canada V6T 1Z2 === Subject: Re: Mathematician on Iraqi Governing Council? >> Does anyone know if this is the same Ahmad Chalabi? > He has a Ph.D. in math, that much I know. Just goes to prove . . . > something. That politics doesn't add up? (insert old German joke with the double meaning of unberechenbar here) -- Hauke Reddmann <:-EX8 fc3a501@uni-hamburg.de als man ankam wollte man werden, die geschichte schreiben, die doofen sollen sterben, der plan als man damals nach hamburg kam (Kettcar) === Subject: Re: need help on Normalizing data > The only other thing I can think of is that perhaps the coefficients are > small enough so that you can calculate an*x, then iteratively multiply that > by x until you get to an*x^n. If you want to do this without iteration, you > might calculate exp(ln an + n ln x), which will hopefully keep you in the > bounds of the program. > To take advantage of this in Matlab, I should note, you might have to use > floating point operations, since, if Matlab is like Maple, it'll try to > compute it exactly in fraction form, which might lead you to another > overflow. This method perfectly works for an*x^n and an*(n+1)*x^n series but I also working with other series like an*n*x^(n-1) and an*x^(n+1) for this series the results are in complex numbers how can I deal with this?? === Subject: Re: Fields of study under mathematics > Is there a website resource that lists areas of mathematics and the > components of each field of study, > I have already checked out mathematical atlas and that at times can be > ambigous, what I am looking for is a catalog style heading/description > resource. > thank you very much. Mathematics Subject Classification... http://www.ams.org/msc/ -- G. A. Edgar http://www.math.ohio-state.edu/~edgar/ === Subject: Re: Fields of study under mathematics > Mathematics Subject Classification... > http://www.ams.org/msc/ This was exactly the type of thing I was looking for, thank you very much. === Subject: Re: McCabe Thiel, Traingle Raffinate, & Fixed Point > I recently worked with Fixed Point Theorems (eg FixedPoint in >Mathematica) and I realised a lot of graphical methods used in >chemical engineering, especially in unit operations actually depend on >fixed-point theorems for solution. (For Engineers: Fixed-Point >theorems basically describe how you can find roots of contracting >functions and how to turn all functions into contracting >functions. Look at how a Fixed-Point theorem converges and it looks >just like one of those graphical-method diagrams.) Is there any book >on seprations that uses fixed-point methods instead of graphical >methods? Has there been any attempt to incorporate this into the >curriculum? The reason I am thinking about this is there has been a >lot of thought in removing a lot of deadweight from Ch E curricula, >and of making it less information overload and more of a >conceptually unified field. I'm not familiar with Fixed-Point theorems, but I would be concerned about the possibility that you could be replacing a procedure that has a real, physical correspondence with one that's nothing more than a mathematical abstraction. One reason McCabe-Thiele is taught is because it clearly shows how concentration changes throughout the tower, and why. This helps the student understand physically what's going on inside those big, tall, black boxes. If your methods don't support this physical understanding, I would at best consider adding them _after_ the students had a thorough understanding of McCabe-Thiele. Rich Lemert === Subject: Series help needed, seriously I'd like to know if this series converges: sum_n 1/n |e^{2pi iKn^{-2}} -1|^2 Here K is constant. I doubt that it does, because when x is small, e^{-2pi iKx} is aproximately x, in this case n^{-2} so this series becomes something like sum_n n^{-3} . The reason I belive I have mistaken is that Salomon Bochner claims in his paper Integration von funktionen deren Werte... that in sums to infinity. This is part of his example of vector valued function in L^2 whose fourier series diverges. === Subject: Re: Series help needed, seriously >I'd like to know if this series converges: >sum_n 1/n |e^{2pi iKn^{-2}} -1|^2 >Here K is constant. >I doubt that it does, because when x is small, >e^{-2pi iKx} or rather |e^{-2pi iKx} -1| >is aproximately x, in this case n^{-2} so >this series becomes something like >sum_n n^{-3} . ??? Maybe there was a typo above. This is exactly why the series _converges_. (Maybe you meant to say that this is why you doubt that it diverges?) >The reason I belive I have mistaken is that Salomon Bochner >claims in his paper Integration von funktionen deren Werte... >that in sums to infinity. This is part of his example of >vector valued function in L^2 whose fourier series diverges. The series above converges. I find it very hard to believe that Bochner said it diverges; possibly you're reading something wrong (or possibly it's a typo in the paper and he meant to say some other series diverges...) Oh. Here's a conjecture where the misunderstanding could be: If f(K) is the sum of that series then f(K) does tend to infinity as K tends to infinity. ************************ David C. Ullrich === Subject: Re: Series help needed, seriously :>Here K is constant. :>I doubt that it does, because when x is small, :>e^{-2pi iKx} : or rather |e^{-2pi iKx} -1| Thats right, I typed wrong. :>is aproximately x, in this case n^{-2} so :>this series becomes something like :>sum_n n^{-3} . : ??? Maybe there was a typo above. This is exactly : why the series _converges_. (Maybe you meant to : say that this is why you doubt that it diverges?) Well, I probably try to write better english that Im able to. I meant to say: I belive it converges, bacause... : Oh. Here's a conjecture where the misunderstanding : could be: If f(K) is the sum of that series then f(K) : does tend to infinity as K tends to infinity. What is the difference? Diverge:=partial sums tend to infinity? The whole example was this: Divide [0,1] to pieces suvh that length of k:th piece is Ak^{-3} and define function f=(f_n):[0,1]-> ell^1 (space of summable sequences) f(t)= sqrt{n} if t is in piece number k f(t)=0 otherwise. Now f is in L^2, and its k:th Fourier coefficient is a_k = int_{0}^1 f(t)e^{-2pi ikt} dt and therefore the n:th element of k:th F-efficient is a_{kn} =int_{n:th piece} sqrt{n} e^{-2pi ikt} dt and: |a_{nn}| = sqrt{n} (2pi n)^{-1} |1-e^{-2pi in d_n} | where d_n is the length of n:th piece, d_n =An^{-3} Now Bochner says that for n>n_0 and for suitable constant B (And this I found doubtful) |a_n | geqq geqq Bn^{-1/2} and therefore sum_n |a_n |^2 =infty . === Subject: Re: Series help needed, seriously >:>Here K is constant. >:>I doubt that it does, because when x is small, >:>e^{-2pi iKx} >: or rather |e^{-2pi iKx} -1| >Thats right, I typed wrong. >:>is aproximately x, in this case n^{-2} so >:>this series becomes something like >:>sum_n n^{-3} . >: ??? Maybe there was a typo above. This is exactly >: why the series _converges_. (Maybe you meant to >: say that this is why you doubt that it diverges?) >Well, I probably try to write better english that Im able to. >I meant to say: I belive it converges, bacause... >: Oh. Here's a conjecture where the misunderstanding >: could be: If f(K) is the sum of that series then f(K) >: does tend to infinity as K tends to infinity. >What is the difference? Diverge:=partial sums tend to infinity? >The whole example was this: >Divide [0,1] to pieces suvh that length of k:th piece is Ak^{-3} >and define function f=(f_n):[0,1]-> ell^1 (space of summable sequences) > f(t)= sqrt{n} if t is in piece number k > f(t)=0 otherwise. I can't stand it, I have to try to guess what you meant. Maybe you meant f_n(t)= sqrt{n} if t is in piece number n f_n(t)=0 otherwise. (Note there are _three_ changes there...) >Now f is in L^2, and its k:th Fourier coefficient is >a_k = int_{0}^1 f(t)e^{-2pi ikt} dt >and therefore the n:th element of k:th F-efficient is >a_{kn} =int_{n:th piece} sqrt{n} e^{-2pi ikt} dt This would be true with my guess at the definition of f. This would imply that |a_{kn}| = sqrt{n}/k |e^{-2pi ik d_n} -1| which agrees with what you say |a_{nn}| is. >and: >|a_{nn}| = sqrt{n} (2pi n)^{-1} |1-e^{-2pi in d_n} | >where d_n is the length of n:th piece, d_n =An^{-3} >Now Bochner says that for n>n_0 and for suitable constant B >(And this I found doubtful) >|a_n | geqq geqq Bn^{-1/2} Here we have switched n and k; this is supposed to be the norm of the n-th Fourier coefficient, right? That would be |a_n| = sum_j sqrt{j}/n |e^{-2pi in d_j} -1| So he's claiming that sum_j sqrt{j} |e^{c i n/j^3} -1| >= c sqrt(n). I'm about to be late for class, so I don't see exactly why it's that large right now, but it's clear that it does blow up as n -> infinity. Hmm, we have |e^{c i n/j^3} -1| >= c > 0 for 1 <= j <= cuberoot(n), (at least for most such j), and the sum of sqrt(j) up to cuberoot(n) is about sqrt(n). That's not a very precise argument, but what he says definitely looks right to me. I have to go now, really... >and therefore >sum_n |a_n |^2 =infty . ************************ David C. Ullrich === Subject: Re: Series help needed, seriously : I can't stand it, I have to try to guess what you meant. Maybe you : meant : f_n(t)= sqrt{n} if t is in piece number n : f_n(t)=0 otherwise. Thats right. : which agrees with what you say |a_{nn}| is. çnd thats right also, but I find one more typo here (and this tells, why a_nn was mentioned separately) :>Now Bochner says that for n>n_0 and for suitable constant B :>(And this I found doubtful) :>|a_n | geqq geqq Bn^{-1/2} Bochner actually said that |a_n | geqq |a_nn |geqq Bn^{-1/2} so I forgot |a_nn | here, and thats where the original question came from. (I've been suffering coffeine-underdose today, hope that it's good reason enough for these typos). I've tried also to count the actual norm of a_nn ,with poor success, and that's why I'm so keen to understand Bochner's easier way. : That's not a very precise argument, but what he says definitely : looks right to me. I have to go now, really... I'll look at it, thank you very much... === Subject: Re: Series help needed, seriously >: I can't stand it, I have to try to guess what you meant. Maybe you >: meant >: f_n(t)= sqrt{n} if t is in piece number n >: f_n(t)=0 otherwise. >Thats right. >: which agrees with what you say |a_{nn}| is. >.8dnd thats right also, but I find one more typo here >(and this tells, why a_nn was mentioned separately) >:>Now Bochner says that for n>n_0 and for suitable constant B >:>(And this I found doubtful) >:>|a_n | geqq geqq Bn^{-1/2} >Bochner actually said that >|a_n | geqq |a_nn |geqq Bn^{-1/2} >so I forgot |a_nn | here, and thats where the original question came from. >(I've been suffering coffeine-underdose today, hope that it's good reason >enough for these typos). Well, I don't see why |a_{nn}| should be that large - that seems wrong to me as well. But it does seem clear to me that |a_n | >= Bn^{-1/2}, which is the whole point. >I've tried also to count the actual norm of a_nn ,with poor success, >and that's why I'm so keen to understand Bochner's easier way. >: That's not a very precise argument, but what he says definitely >: looks right to me. I have to go now, really... >I'll look at it, thank you very much... ************************ David C. Ullrich === Subject: Re: Series help needed, seriously >:>Here K is constant. >:>I doubt that it does, because when x is small, >:>e^{-2pi iKx} >: or rather |e^{-2pi iKx} -1| >Thats right, I typed wrong. >:>is aproximately x, in this case n^{-2} so >:>this series becomes something like >:>sum_n n^{-3} . >: ??? Maybe there was a typo above. This is exactly >: why the series _converges_. (Maybe you meant to >: say that this is why you doubt that it diverges?) >Well, I probably try to write better english that Im able to. >I meant to say: I belive it converges, bacause... >: Oh. Here's a conjecture where the misunderstanding >: could be: If f(K) is the sum of that series then f(K) >: does tend to infinity as K tends to infinity. >What is the difference? Diverge:=partial sums tend to infinity? What is the difference? They're totally different statements! The series _converges_ for every K. But the _sum_ of the series tends to infinity as K tends to infinity; it's not the series that's diverging, it's f. >The whole example was this: >Divide [0,1] to pieces suvh that length of k:th piece is Ak^{-3} >and define function f=(f_n):[0,1]-> ell^1 (space of summable sequences) > f(t)= sqrt{n} if t is in piece number k > f(t)=0 otherwise. You are again not saying what you mean. The f you define here is f : (0,1) -> R, not f : (0,1) -> ell^1. I could try to guess what the correct definition of f is from what's below, but I have to go to work soon - it would be much easier to answer your question if you simply stated things precisely. >Now f is in L^2, and its k:th Fourier coefficient is >a_k = int_{0}^1 f(t)e^{-2pi ikt} dt >and therefore the n:th element of k:th F-efficient is >a_{kn} =int_{n:th piece} sqrt{n} e^{-2pi ikt} dt >and: >|a_{nn}| = sqrt{n} (2pi n)^{-1} |1-e^{-2pi in d_n} | >where d_n is the length of n:th piece, d_n =An^{-3} >Now Bochner says that for n>n_0 and for suitable constant B >(And this I found doubtful) >|a_n | geqq geqq Bn^{-1/2} >and therefore >sum_n |a_n |^2 =infty . ************************ David C. Ullrich === Subject: Re: Series help needed, seriously : What is the difference? They're totally different statements! : The series _converges_ for every K. But the _sum_ of the : series tends to infinity as K tends to infinity; it's not the : series that's diverging, it's f. Oh, translation problem. I would say that the _sequence_ converges, but the series diverges... At least thats what I guess you mean. :>and define function f=(f_n):[0,1]-> ell^1 (space of summable sequences) :> f(t)= sqrt{n} if t is in piece number k :> f(t)=0 otherwise. : You are again not saying what you mean. The f you define : here is f : (0,1) -> R, not f : (0,1) -> ell^1. Sorry, typo again, I meant of course f_n (t)= sqrt{n} if t is in piece number n f_n (t)=0 otherwise. where f_n is the n:th element of f. So heres the example as it should be at least I hope so.. Divide [0,1] to pieces suvh that length of k:th piece is Ak^{-3} and define function f=(f_n):[0,1]-> ell^1 (space of summable sequences) f_n (t)= sqrt{n} if t is in piece number n f_n (t)=0 otherwise. Now f is in L^2, and its k:th Fourier coefficient is a_k = int_{0}^1 f(t)e^{-2pi ikt} dt and therefore the n:th element of k:th F-efficient is a_{kn} =int_{n:th piece} sqrt{n} e^{-2pi ikt} dt and: |a_{nn}| = sqrt{n} (2pi n)^{-1} |1-e^{-2pi in d_n} | where d_n is the length of n:th piece, d_n =An^{-3} Now Bochner says that for n>n_0 and for suitable constant B (And this I found doubtful) |a_n | geqq geqq Bn^{-1/2} and therefore sum_n |a_n |^2 =infty . === Subject: Re: Resistance to Change > But the fact that some geniuses were laughed at does not imply that all who > are laughed at are geniuses. Well duh! ~( ((eX)GENIUS(x)^LAUGHEDAT(x)) => ((aY)LAUGHEDAT(Y)=>GENIUS(Y)) ) Even Frege could prove that one. C'mon Frege. Show us your neat proof. :) > They laughed at Columbus, they laughed at > Fulton, they laughed at the Wright brothers. But they also laughed at Bozo > the Clown. > -- Carl Sagan Axiom: LAUGHEDAT(Columbus)^LAUGHEDAT(Fulton)^LAUGHEDAT(Wright Borthers)^LAUGHEDAT(Bozo) Barbara, do you see any of the differences between people laughing at a genius and people laughing at a comedian? Charlie Volkstorf Cambridge, MA === Subject: Re: Resistance to know anything > ~( ((eX)GENIUS(x)^LAUGHEDAT(x)) => ((aY)LAUGHEDAT(Y)=>GENIUS(Y)) ) Stated with *corrected* notation: ~( Ex(GENIUS(x) ^ LAUGHED_AT(x)) -> Ax(LAUGHED_AT(x) -> GENIUS(x)) ) > Even Frege could prove that one. C'mon Frege. Show us your neat > proof. :) Idiot! This is n o t a theorem of FOPL. (Hence there is no proof for it.) But I'm quite sure that YOU [It's lonely at the top.] are able to prove that one. :-) F. === Subject: Re: Resistance to know anything > ~( ((eX)GENIUS(x)^LAUGHEDAT(x)) => ((aY)LAUGHEDAT(Y)=>GENIUS(Y)) ) > Stated with *corrected* notation: > ~( Ex(GENIUS(x) ^ LAUGHED_AT(x)) -> Ax(LAUGHED_AT(x) -> GENIUS(x)) ) To each his own. :) [I'm more used to texts where there are parentheses around the quantifiers. I think the more recent ones do. Now, I could survey my library ... but I'll leave the BS name-dropping to people like you who LOVE to do that.]* > Idiot! This is n o t a theorem of FOPL. (Hence there is no proof for > it.) > But I'm quite sure that YOU [It's lonely at the top.] are able to > prove that one. :-) Here's a hint: Axiom: LAUGHEDAT(Fulton) Axiom: GENIUS(Fulton) Axiom: LAUGHEDAT(Bozo) Axiom: ~GENIUS(Bozo) Axiom: GENIUS(Charlie-Boo) Axiom: ~GENIUS(Frege) Axiom: ~EVENSMART(Frege) Axiom: (aX)(aY)(~GENIUS(X)^GENIUS(Y)) => LT(intelligence(X),intelligence(Y)) Theorem: LT(intelligence(Frege),intelligence(Charlie-Boo)) Axiom: EDUCATEDFOOL(David Ullrich) Axiom: (aX)(aY)((~GENIUS(X)vEDUCATEDFOOL(X))^GENIUS(Y)) => SHOULDLISTENTO(X,Y) Theorem: (eX) SHOULDLISTENTO(X,Charlie-Boo)^~LISTENTO(X,Charlie-Boo) Theorem: ~(e!X) SHOULDLISTENTO(X,Charlie-Boo)^~LISTENTO(X,Charlie-Boo) Charlie Volkstorf Cambridge, MA * Actually, I use parentheses around quantifiers in wffs because that makes my software easier to program, but you wouldn't know anything about such matters. Axiom: (aX) KNOWABOUTSOFTWARE(X) => EVENSMART(X) > F. === Subject: Re: Resistance to know anything > ~( ((eX)GENIUS(x)^LAUGHEDAT(x)) => ((aY)LAUGHEDAT(Y)=>GENIUS(Y)) ) > >> Stated with *corrected* notation: >> ~( Ex(GENIUS(x) ^ LAUGHED_AT(x)) -> Ax(LAUGHED_AT(x) -> GENIUS(x)) ) > I'm more used to texts where there are parentheses around the quantifiers. That's not the problem. In this case, we usually would have: ~( (Ex)(GENIUS(x) ^ LAUGHED_AT(x)) -> (Ax)(LAUGHED_AT(x) -> GENIUS(x)) ) or (as a variant) ~( (Ex)(GENIUS(x) ^ LAUGHED_AT(x)) -> (Ay)(LAUGHED_AT(y) -> GENIUS(y)) ) if you like. For more details, see Chris Menzel's comment. (And no, I don't care if you write 'x','X','y' or 'Y', since you seem to have some troubles in choosing a consistent notation concerning upper and/or lower case. :-) Note though: > This is n o t a theorem of FOPL. (Hence there is no proof for it.) F. === Subject: Re: Resistance to know anything ~((Ex)(GENIUS(x) ^ LAUGHED_AT(x)) -> (Ax)(LAUGHED_AT(x) -> GENIUS(x))) > This is n o t a theorem of FOPL. (Hence there is no proof for it.) It's funny thought that a self declared genius like you can't see that simple fact immediately. :-) - Lonely at the top, eh?!? F. === Subject: Re: Resistance to know anything > ~( ((eX)GENIUS(x)^LAUGHEDAT(x)) => ((aY)LAUGHEDAT(Y)=>GENIUS(Y)) ) Stated with *corrected* notation: ~( Ex(GENIUS(x) ^ LAUGHED_AT(x)) -> Ax(LAUGHED_AT(x) -> GENIUS(x)) ) > To each his own. :) [I'm more used to texts where there are > parentheses around the quantifiers. I think the more recent ones do. > Now, I could survey my library ... but I'll leave the BS name-dropping > to people like you who LOVE to do that.]* Actually, I don't think he's referring as much to your chosen ASCII-quantifier notation as to the fact that the scoping is all wrong. Notably, the occurrence of x in LAUGHEDAT(x) is not bound by the corresponding quantifier. Similarly for the occurrence of Y in GENIUS(Y). Chris Menzel === Subject: Re: Resistance to Change >> Where did it appear? >http://www.arxiv.org/html/cs.lo/0003071 > Uh, right. So it hasn't actually been published in a refereed > journal? Why do you ask? What does that have to do with my or anybody else's proof of the unsolvability of the Halting Problem? Charlie Volkstorf Cambridge, MA > ************************ > David C. Ullrich === Subject: Re: Resistance to Change > Where did it appear? >>http://www.arxiv.org/html/cs.lo/0003071 >> Uh, right. So it hasn't actually been published in a refereed >> journal? >Why do you ask? I thought I made that clear. If I had some evidence that you'd actually accomplished something original once in your life (something original that the referee thought was interesting enough and non-trivial enough to publish) then when you said something that immediately appeared to be ridiculous I'd be more likely to give you the benefit of the doubt and work harder at trying to figure out why it was _not_ ridiculous. >What does that have to do with my or anybody else's >proof of the unsolvability of the Halting Problem? Nothing. Has a lot to do with the credibility of your grandiose claims - a lot to do with how we need to define top when you say that it's lonely at the top, for example. You didn't answer my question elsewhere: I asked whether you were familiar with the logical system described in the references to that paper, and you said of course. The question is, if you're familiar with the details of how the system works (which of course you must be, else the things you've been saying about the paper would be hilariously ignorant) then why have you continued to ask people to explain the system to you? You've repeatedly asked me to explain things I never claimed to understand. Why don't _you_ give us a summary of the essential features of that logical system? >Charlie Volkstorf >Cambridge, MA >> ************************ >> David C. Ullrich ************************ David C. Ullrich === Subject: Re: Resistance to Change > How many intermediate theorems are there and what do they mean, in > ordinary English? > Your questions seem to be answered directly in the paper. > Anyway, a quick glance at the proof reveals that it uses 10 intermediate > lemmas and 14 definitions. What do those lemmas mean, in English? All I read is a description of how LISP works and list manipulations, rather than assertions concerning the Halting Problem. That doesn't prove that the Halting Problem is unsolvable. For example, in my proof, the 1st 5 theorems generated along the way (to showing a contradiction) are: Proof 1. HALT(I,I) Given 2. ~HALT(I,I) NOT 1 3. TRUE(x) Axiom 1 We can list the Universal Set. 4. TRUE(x)^~HALT(x,x) DO 3,2 I=x 5. ~HALT(x,x) DEF-7 4 Property of Universal Set etc. 1 = The self-halting predicate is recursive. 2 = The non-self-halting predicate is recursive. 3 = The universal set (the natural numbers) is r.e. 4 = The intersection of the universal set and the non-self-halting set is r.e. 5 = The non-self-halting set is r.e. etc. > According to section 2, The LISP Interpreter: > The programming language used in our statement of the halting problem > is a version of Pure LISP [5]. We present our version by defining > the logical function EVAL, which takes four arguments: > 1. an S-expression to be evaluated, > 2. a variable alist* assigning values to variable symbols, > 3. a function alist assigning definitions to nonprimitive function > symbols, and > 4. a natural number, indicating the maximum depth of function calls. > EVAL returns either the value of the S-expression in the given > environment or else it returns the object (BTM). That is simply a description of the LISP function EVAL that evaluates an expression. This is an example of what I say above and have said all along: simply LISP programs. Where is the assertion that the Halting Problem is unsolvable and where is its proof? Does the above convince you that the Halting Problem is unsolvable? By what logic and reasoning? You are still not presenting a proof that the Halting Problem is unsolvable. Charlie Volkstorf Cambridge, MA === Subject: Re: finite field = finite integral domain ? >* fuzzykyh@nate.com >> I wonder finite field = finite integral domain ? >Homework? >I guess you want to prove that a finite integral domain is a field. >Hint 1: Let a in G and a != 0. Examine the sequence a,a^2,a^3,... I don't see how to make this into a proof. I can see you must have a^k = 1 but I don't see why must be all of G. I do know of a proof that a finite integral domain is a field, but that starts with Let a in G and a!=0. Consider the set {ag_1, ag_2, ....., ag_n} Is it possible that you have those confused? -- Wim Benthem === Subject: Re: finite field = finite integral domain ? >>* fuzzykyh@nate.com > > I wonder finite field = finite integral domain ? >>Homework? >>I guess you want to prove that a finite integral domain is a field. >>Hint 1: Let a in G and a != 0. Examine the sequence a,a^2,a^3,... >I don't see how to make this into a proof. I can see you must have >a^k = 1 but I don't see why must be all of G. On second thought, a^k =1 for some k>1 is enough to prove that a has an inverse -- Wim === Subject: Re: finite field = finite integral domain ? * Wim Benthem >I don't see how to make this into a proof. I can see you must have >a^k = 1 but I don't see why must be all of G. > On second thought, a^k =1 for some k>1 is enough to prove that > a has an inverse And then prove that any inverse of a and inverse of b is the same element. -- Jon Haugsand Dept. of Informatics, Univ. of Oslo, Norway, mailto:jonhaug@ifi.uio.no http://www.ifi.uio.no/~jonhaug/, Phone: +47 22 85 24 92 === Subject: Re: Conway's comment on large numbers >> I was reading a paper by Conway and Doyle that proves that division by >> three is possible (without the axiom of choice). >> http://math.dartmouth.edu/~doyle/docs/three/three.pdf >> In the last paragraph the authors explain their doubts about the ZF > axioms, saying, >> Indeed, we're somewhat doubtful whether large natural numbers (like >> 80^5000, or even 2^200) exist in any very real sense... So Conway has finally flipped ... > There was a thread about computability where the question > of theoretical vs actual computers came up. > A theoretical computer can perform an unbounded, but finite, > number of operations. I pointed out that finite was pretty > small for an actual computer (where actual means conforms > to the laws of physics as we know them). > Assume a computer can perform an operation in 3 x 10^-44 sec. > (Planck's time - theoretically, the shortest time an action can occur in). > Let this computer count for a billion years (10^9). > (The universe is estimated to be about 13 billion years old. > In 3 billion years, the Milky Way will crash into Andromeda.) > Google says 1 year = 31 556 926 seconds (3 x 10^7). > So, in a billion years, the computer will count to > 3 x 10^44 x 10^9 x 3 x 10^7 = approximately 10^61. You might as well say that Venus doesn't exist because you can't walk there. Surely one doesn't normally count up to 2^200 when working out if 2^200 + 1 is prime or not? It seems to me to have even less to do with The belief that 2^200 doesn't exist makes the intuitionists seem like airy-fairy idealists. -- Timothy Murphy e-mail (<80k only): tim /at/ birdsnest.maths.tcd.ie tel: +353-86-2336090, +353-1-2842366 s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland === Subject: Re: Conway's comment on large numbers >> I was reading a paper by Conway and Doyle that proves that division by >> three is possible (without the axiom of choice). >> http://math.dartmouth.edu/~doyle/docs/three/three.pdf >> In the last paragraph the authors explain their doubts about the ZF > axioms, saying, >> Indeed, we're somewhat doubtful whether large natural numbers (like >> 80^5000, or even 2^200) exist in any very real sense... > So Conway has finally flipped ... > There was a thread about computability where the question > of theoretical vs actual computers came up. > A theoretical computer can perform an unbounded, but finite, > number of operations. I pointed out that finite was pretty > small for an actual computer (where actual means conforms > to the laws of physics as we know them). Assume a computer can perform an operation in 3 x 10^-44 sec. > (Planck's time - theoretically, the shortest time an action can occur in). Let this computer count for a billion years (10^9). > (The universe is estimated to be about 13 billion years old. > In 3 billion years, the Milky Way will crash into Andromeda.) Google says 1 year = 31 556 926 seconds (3 x 10^7). So, in a billion years, the computer will count to > 3 x 10^44 x 10^9 x 3 x 10^7 = approximately 10^61. > You might as well say that Venus doesn't exist because you can't walk there. I don't see where Russell made any claim about the existence of large numbers. In any case, someone who worked with what later turned out to be modular arithmetic might confuse the number 2^200 with 0. Or 10. Or any number less than 2^200. > Surely one doesn't normally count up to 2^200 > when working out if 2^200 + 1 is prime or not? > It seems to me to have even less to do with > The belief that 2^200 doesn't exist > makes the intuitionists seem like airy-fairy idealists. 2^200 obviously exists. The trouble is determining what exactly one _means_ by the word exist. This is not a question of existence, but a question of status and access. And do you exactly know what a philosophical idealist is? (hint: intuitionists usually commit themselves to some form of idealism in formulating their position) 'cid 'ooh === Subject: [help!] Polynomial Equations Let m_i, n_i be rational numbers and f(x) a polynomial equation of order greater than 4 where each root has the form m_i + n_i^0.5. Is there a polynomial-time algorithm to find the exact roots? Perhaps that problem can be reformulated in this way: let R be a floating point number with infinite mantissa (i.e. big as you like) that approximates a value of the form a + b^0.5 where a, b are rational numbers, is it possible to estimate the exact a and b? If you had not guessed it already, I need a polynomial equation with roots of a known form, but impossible to find if not approximate. === Subject: Re: [help!] Polynomial Equations >Let m_i, n_i be rational numbers and f(x) a polynomial equation of order >greater than 4 where each root has the form m_i + n_i^0.5. >Is there a polynomial-time algorithm to find the exact roots? >Perhaps that problem can be reformulated in this way: >let R be a floating point number with infinite mantissa (i.e. big as you >like) that approximates a value of the form a + b^0.5 where a, b are >rational numbers, is it possible to estimate the exact a and b? >If you had not guessed it already, I need a polynomial equation with roots >of a known form, but impossible to find if not approximate. Your R is a root of the quadratic (X - a)^2 - b = 0 where a, b are rational. Assume R is irrational (equivalently, b is not a rational square), so this monic quadratic is unique. After removing denominators, you have a quadratic with integer coefficients and root R. If you can discover this quadratic, you can use the quadratic formula to find a, b. Using lattice basis reduction or other means, one can recover this quadratic given a high-precision estimate of R. If you have maple, look up ?polytools,minpoly (or ?PolynomialTools[MinimalPolynomial] with Version 9]. When R is complex, you know R and its conjugate satisfy the same monic quadratic. You can get high-precision values of a and b quickly, and figure out the the rational values for a, b. You say degree(f(x)) > 4 but don't subsequently reference f(x). Many different f's will be divisible by the quadratic for R. If f is known, you should simply factor it over Z, and then find which factor corresponds to your R. -- John Adams served two terms as Vice President and one as President, but lost reelection. Later his son became President despite losing the popular vote. That son lost his reelection attempt badly. Now history is repeating itself. pmontgom@cwi.nl Microsoft Research and CWI Home: San Rafael, California === Subject: Re: [help!] Polynomial Equations Kiuhnm === Subject: Re: [help!] Polynomial Equations In the previous post I used the term impossible, but I meant computationally impractical (e.g. NP-hard). Kiuhnm === Subject: Re: ring isomorphism in relation to tensor products question > Alternatively, once you've defined a map from Z[i](x) R to C by > inducing it from a bilinear map from Z[i] x R to C, you could try to > define its inverse from C to Z[i] x R in order to show bijectivity, > but that seems hard off the top of my head. That is really quite easy: map a+bi to 1 (x) a + i (x) b. -- Robin Chapman, www.maths.ex.ac.uk/~rjc/rjc.html Lacan, Jacques, 79, 91-92; mistakes his penis for a square root, 88-9 Francis Wheen, _How Mumbo-Jumbo Conquered the World_ === Subject: Re: ring isomorphism in relation to tensor products question Adjunct Assistant Professor at the University of Montana. >> Alternatively, once you've defined a map from Z[i](x) R to C by >> inducing it from a bilinear map from Z[i] x R to C, you could try to >> define its inverse from C to Z[i] x R in order to show bijectivity, >> but that seems hard off the top of my head. >That is really quite easy: map a+bi to 1 (x) a + i (x) b. Sigh. Not my week. -- It's not denial. I'm just very selective about what I accept as reality. --- Calvin (Calvin and Hobbes) Arturo Magidin magidin@math.berkeley.edu === Subject: Re: Replacement to MMA and Maple by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LCvlN14149; >Then, I tried Mathematica 5.0. >At my very first try, I found >bugs. I cannot use it for my >research. Yes. I switched back to 4.x; stable, bug free, and fewer annoying features. >Maple and MMA are to CAS what >M$ Word is to word processing. >Do you know a CAS that is more >into the LATEX philosophy? >How robust, advanced, etc., >is it? Depends what you are looking for. If you are looking for a commutative algebra package, then there are some free cutting egde tools (cocoa, macauly2, singular). In number theory the best tool around is the free pari/gp. However, if you need a general tool, the only options are mathematica and maple. HTH Daivd === Subject: Re: Replacement to MMA and Maple >Then, I tried Mathematica 5.0. >At my very first try, I found >bugs. I cannot use it for my >research. > Yes. I switched back to 4.x; stable, bug free, and > fewer annoying features. >Maple and MMA are to CAS what >M$ Word is to word processing. >Do you know a CAS that is more >into the LATEX philosophy? >How robust, advanced, etc., >is it? > Depends what you are looking for. If you are looking for > a commutative algebra package, then there are some free cutting egde > tools (cocoa, macauly2, singular). In number theory the best tool > around is the free pari/gp. However, if you need a general tool, > the only options are mathematica and maple. > HTH > Daivd Well there is maxima, at maxima.sf.net. === Subject: are the following Noetherian rings by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LCvlk14153; Could you please help me detemine if the following rings are Notherian? 1-The ring of rational functions of z having no pole on the circle |z|=1. 2-The ring of polynomials in z,w all of whose partial derivatives with respect to w vanish for z=0. In all cases the coefficients are complex. === Subject: Re: are the following Noetherian rings > Could you please help me detemine if the following rings are Notherian? > 1-The ring of rational functions of z having no pole on the circle |z|=1. > 2-The ring of polynomials in z,w all of whose partial derivatives with respect to w vanish for z=0. > In all cases the coefficients are complex. 1. If R is a commutative ring and T is a subset closed under products, containing 1 and not containing 0, let R_T denote the ring of all fractions r/t, r in R and t in T. It is called a ring of fractions over R. Addition and multiplication are defined as usual for fractions. If R has no zero divisors and T consists of all non-zero elements, R_T is the field of all fractions. An example is Z_T where Z is the ring of integers and T is all powers of 2. You get all fractions a/b where b is a power of 2. The answer to the first is settled by observing that any ring of fractions over a Noetherian ring is Noetherian. If I is an ideal, any element has the form a/t where t in T and then a in I as well. Conversely, if a is an element of R cap I, then a/t is in I for any t in T. Therefore any set of generators of the R-ideal I cap R is also a set of generators for I. 2. I don't have any idea on this one. === Subject: difference equations by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LCvnn14208; === Subject: Re: difference equations >i'm looking for online references to difference equations, especially nonlinear ones. recommendations for books also welcome. Luenberger, Introduction to Dyanmic Systems, provides a nice introduction to linear difference equations. -- Stephen J. Herschkorn herschko@rutcor.rutgers.edu === Subject: Re: Gender issues on 60 Minutes, Women especially respond! by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LCvlh14145; Single sex education is all the things that you were written about here and alot more. Single sex education is, and should be, an OPTION for any one who desires it. The benefits are there and deserve to be offered. Putting young men and women in situations where they do not have the pressures and distrations that go with coeducation can,in many cases, provide the best opportunities for success in academics for both sexes. As an educator and a man who went to an all boys middle school, I would highly recommend it to any parent of a Middle School child. I think that the social requirements of growing up demand coeducation...but that those years...6th -9th..are prime years where intensive instruction with little distraction can have dramatic results. Again...Single Sex classes should not be for everyone and not mandatory, but should be available. After all, it can not be any worse than something like German Immersion schools. >
Shouldn't they be allowed to set up single - sex classes for girls?
>> I've heard that *some* girls do better in single-sex schools, and
>> that *some* boys (e.g., those from inner city, high crime areas,
>> perhaps with no father in the home) may do better in a highly
>> disciplined, single-sex situation almost like boot camp.
>> I'm not at all sure that *public* schools should be doing this,
>> though. It seems awfully divisive to me. It gives very negative
>> messages, like girls and boys are a lot more different than they
>> actually are, that girls and boys have to be separated because the
>> girls just can't compete with the boys.
>>Women here in sci.math, would a single-sex class have been better
>>for you in math and science?
>> No. For me it would have been a lot worse. I excelled in math and
>> science, and knew that I was smarter than all the boys *and* girls,
>> which no doubt contributed to the self-confidence that I have today.
>> I had about equal numbers of male and female friends, and would
>> probably have been less comfortable socially in an all girl setting
>> (a good friend went off to a private, all girls high school, which I
>> remember visiting, and finding weird--it was not a place where I
>> would have been happy).
>> Thinking back on college, most of my math classes (and computer
>> science etc.) were overwhelmingly male (though I hardly noticed this
>> at the time). I was completely comfortable and confident in such
>> classes, and continued to excel. This might not have been the case
>> had I been isolated in all-girls classes up until then, and been
>> given the message that girls can't compete with the boys.
>> --
>> Helen Read
>> The issues concerning gender and nature-vs.-nurture are so clouded by
>> strong opinions and politics and subconscious agendas( someone might be
>> implying something in what (s)he says, and not even realize it, or
>> it could be VERY subtle( somethings that Rush Limbaugh says are like
>> this ), etc. ) and biases. Sometimes the accusations are leveled that
>> cause is confused with effect and there exists a rebuttal to that
>> thatsays that what those people call the cause( social stereotypes ) had
>> to have originated somehow, and that what those people call the effect
>> is actually the cause of that, itself being the effect corresponding to
>> a cause which is said to be innate(e.g., closer to the level of genetic
>> or hormone-influenced, etc. ).
>> I too would disagree with the sexual segregation of education because it
>> would reinforce the notion, in the minds of those in it( of both sexes
>> ), that the sexes live in vastly different worlds that bare little if
>> any resemblance to each other. Instead of eliminating it, it would be
>> likely to engender( I am aware of the pun; react to it as you wish )
>> sexism in the minds of all who are in this system as opposed to a
>> single-sex, traditional, eductaion system. School is supposed to reflect
>> reality; if we institute sexually segregated education systems at lower
>> levels of education, students in these systems will find themselves
>> somewhat unprepared for the upper levels( however far up you have to go:
>> if universities and journals are segregated then you would have to go
>> all the way to the ultimate level, which is the collective present state
>> of the mathemtical knowledge of humanity, and that would be very hard if
>> not impossible to segregate sexually, given the state of today's
>> society ), and the difficulties will occur there, especially, if, in the
>> single-gender system, they were given the message that one sex was
>> inherently better than the other at math, computers, science, and
>> related fields.
>> Environment can have a profoundly powerful effect on a person's
>> perceptions, especially where these issues with respect to gender are
>> concerned. Parents, in finding it so cute that their little boy is 
so
>> into the usual male gender-role stereotypes and/or that their little
>> girl is so into the usual female gender-role stereotypes at such a young
>> age, may be inadvertently sowing the seeds of sexism into the next
>> generation.
>> I will say more later, but now there is a thunderstorm outside.
>> ---- David Harden
>Continuing, as I said I would:
>Another influence that, as I have read, promotes gender-role stereotypes 
>is television. Studies show that girls who watch more television at a 
>young age are more likely to view women as better suited to home life, 
>happier with children, etc. than girls who watched less television at a 
>young age. There is no reason not to believe that analogous things hold 
>for boys and thus, from an early age, often without their awareness, 
>boys and girls are channeled by society into having different sets of 
>interests and tendencies. As they grow older, the process continues and 
>boys are more harangued about not making a big show of emotions than 
>girls are( among other things ), and in adolescence, acting within 
>traditional gender-role stereotypes can be compunded by social 
>peer pressure to affirm oneself as a member of one's own sex and also as 
>a heterosexual( if you behave too much 'like a member of the opposite 
>sex', you might be a homosexual, so the reasoning goes ). By the time 
>adulthood is reached, one is usually well set in thinking patterns and 
>the differences remain, and on to the next generation...
>The mathematical/scientific fields of study are distinguished from 
>others in that they transcend culture, society, politics, and gender. 
>They have an objective existence all over the world in all time periods, 
>historical events, political, revoltuions, fads, and for both men and 
>women.
>Anyone who has what it takes can investigate mathematics, and that 
>ability should be evaluated in a gender-blind way, purely 
>meritocratically, without previous reservations or inclinations. If this 
>is done, then let's see what happens, and see what difference it makes.
>---- David Harden
>
=== Subject: Re: the MVT -- - -- applications by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LCvmQ14184; >Hey! I'm learning about the MVT, and am stuggling with being able to apply it to solve these problems. >Question 1: >Prove that tan x > x for 0 < x < pi/2. >I think there must be an f'(c) = (tan(pi/2) - tan 0)/(pi/2) >however, since tan(pi/2) is undefined i am having difficulty trying to answer the question. But you want to prove that tan x> x for ALL x in the interval so a value of the derivative of tan x at some specific c doesn't help. Since tan x> x is the same as tan x- x is NOT 0, it might be better to turn it around: Certainly tan x- x= 0 for x= 0. IF tan x- x=0 for some x< pi/2, them, by MVT, f'(c)= 0 (here f(x)= tan x- x). Can you show that that is NOT true? >Question 2: >By using the MVT with f(x)=x^.5 on [25,26], show that >26^.5 - 25^.5 < 1 >f'(c) = 26^.5 - 25^.5 >and f'(x) = 1/(2(x^.5)), so f'(c) = 1/(2(x^.5)). >but how do I prove the answer? You know that 26^(.5)- 25^(.5)= 1/(2(c^.5)) for some c between 5 and 6. Okay, how large can that right hand side be? (Hint: since x to a negative power is a DECREASING function, the maximum is at the left end point!) === Subject: Re: Existing of a triangle >Let ABCD is a rectangle and X is an inner point for ABCD. >Is there a triangle, the faces of which are amongst XA, XB, XC and XD? Any three non-collinear points form a triangle. If X is in the interoir, it is clearly not collinear with any two adjacent vertices of the rectangle. So you have triangles XAB, XBC, XCD, and XDA. --Keith Lewis klewis {at} mitre.org The above may not (yet) represent the opinions of my employer. === Subject: Re: Existing of a triangle I believe that the OP was referring to the trigonal inequality, not the tiling of a rectangle! > If X is in the interoir, it is clearly not collinear with any two adjacent > vertices of the rectangle. So you have triangles XAB, XBC, XCD, and XDA. --ils duces d'Enron! http://tarpley.net === Subject: Re: Existing of a triangle >>Let ABCD is a rectangle and X is an inner point for ABCD. >>Is there a triangle, the faces of which are amongst XA, XB, XC and XD? >Any three non-collinear points form a triangle. >If X is in the interoir, it is clearly not collinear with any two adjacent >vertices of the rectangle. So you have triangles XAB, XBC, XCD, and XDA. But the edges of e.g. XAB include AB, which is not in the set of four candidates, so this doesn't answer the question. Three edges form a triangle if the longest is shorter than the sum of the other two. Well, we can draw coordinates around the rectangle ABCD so that the vertices have x-coordinates 0 or a and y-coordinates 0 or b, and so that X=(x,y) with 0 < x < a/2, 0 < y < b/2. In that case the longest edge has length sqrt( (a-x)^2 + (b-y)^2 ) and the next two have lengths sqrt( x^2 + (b-y)^2 ) and sqrt( (a-x)^2 + y^2 ); it's easy to check that the sum of these is less than the longest length. So, yes, the three longest lengths always form a triangle. dave === Subject: Re: Existing of a triangle >Let ABCD is a rectangle and X is an inner point for ABCD. >Is there a triangle, the faces of which are amongst XA, XB, XC and XD? >>Any three non-collinear points form a triangle. >>If X is in the interoir, it is clearly not collinear with any two adjacent >>vertices of the rectangle. So you have triangles XAB, XBC, XCD, and XDA. > But the edges of e.g. XAB include AB, which is not in the set of four > candidates, so this doesn't answer the question. > Three edges form a triangle if the longest is shorter than the sum of > the other two. Well, we can draw coordinates around the rectangle > ABCD so that the vertices have x-coordinates 0 or a and y-coordinates > 0 or b, and so that X=(x,y) with 0 < x < a/2, 0 < y < b/2. In that > case the longest edge has length sqrt( (a-x)^2 + (b-y)^2 ) and the > next two have lengths sqrt( x^2 + (b-y)^2 ) and sqrt( (a-x)^2 + y^2 ); > it's easy to check that the sum of these is less than the longest length. > So, yes, the three longest lengths always form a triangle. > dave Hmm. I discarded that as being a too-trivial interpretation. Perhaps the OP intended that the triangle be made by translating the edges, keeping their orientations. In that case, my intuition is that the answer is no, but I don't have a proof. Rick EC-FUNDED RESEARCH VISIT OPPORTUNITIES IN HIGH PERFORMANCE COMPUTING www.hpc-europa.org HPC-Europa, a consortium of 11 leading European centres, is calling for applications from researchers working in Europe to visit any of the 6 centres in its Transnational Access programme. Visitors will use HPC-Europa's High Performance Computing (HPC) facilities to advance their research, while working in collaboration with a scientific researcher from an appropriate local research institute. The programme is fully funded by the EC's Structuring the European Research Area programme, and offers: - access to some of the most powerful High Performance Computing (HPC) facilities in Europe; - HPC consultancy from experienced staff; - opportunities to collaborate with scientists working in related fields at a local research institute; - travel costs, subsistence expenses and accommodation (may be in a shared flat); The 6 HPC-Europa centres offering Transnational Access are: CEPBA-CESCA-IBM (Spain), CINECA (Italy), EPCC (UK), HLRS (Germany), IDRIS (France) and SARA (the Netherlands). APPLICATION AND ELIGIBILITY Applications for the programme are welcomed from academic or industrial researchers at any level, from senior postgraduate students to the most senior researchers, who are working on non-proprietary research in any discipline to which access to HPC facilities is beneficial. Applicants must be working in a research group based in an eligible country, i.e. any EU member state or Associated State (see full list at http://www.hpc-europa.org/presentation.html). Note that researchers are NOT eligible for visits to centres in the country where they are currently carrying out their research. Researchers who do not have access to similar HPC facilities are particularly encouraged to apply. The selection panel approves applications on the basis of scientific merit. For this reason, applicants are strongly encouraged, where possible, to identify a suitable researcher with whom to collaborate before applying, as this greatly increases the chances of the application being approved. Applications should be made using the on-line application form at selection meeting in May. Successful applicants will visit for between meetings will be held 4 times per year. Further information and the on-line application form are available at: http://www.hpc-europa.org/ If you have any questions, please contact the HPC-EUROPA team at access@hpc-europa.org === Subject: Re: Commutative rings aren't physical spaces (This Week's Finds 205) > Take for example supposedly curved > space as in general relativity. Whitehead showed you could do > relativity in flat space. Did he? Not in what is usually known as Whitehead's theory of gravity, because spacetime is not flat in it, just differently curved than in Einstein's theory. -- Esa Peuha student of mathematics at the University of Helsinki http://www.helsinki.fi/~peuha/ === Subject: Re: the MVT -- - -- applications by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LF4tt03787; Hey All, Here's kind of a fun MVT application: Prove that for integers n >1 that there is always a square between n^3 and (n+1)^3. Mark >Q1. >Pick x so that 0 < x < pi/2. (This eliminates the need for tan(pi/2).) >The MVT in this instance states > (tan x - tan 0)/(x - 0) = (sec^2)(c) >where c satisfies 0 < c < x. Now (sec^2)(c) >= 1 . . . . >Hope this helps. >__________________________________________________________ >> Hey! I'm learning about the MVT, and am stuggling with being able to >apply it to solve these problems. >> Question 1: >> Prove that tan x > x for 0 < x < pi/2. >> I think there must be an f'(c) = (tan(pi/2) - tan 0)/(pi/2) >> however, since tan(pi/2) is undefined i am having difficulty trying to >answer the question. >> Question 2: >> By using the MVT with f(x)=x^.5 on [25,26], show that >> 26^.5 - 25^.5 < 1 >> f'(c) = 26^.5 - 25^.5 >> and f'(x) = 1/(2(x^.5)), so f'(c) = 1/(2(x^.5)). >> but how do I prove the answer? === Subject: Re: Mathematical Language Equations > Here is a language equation puzzle where the solutions are mathematical. > [So I will cross-post this to sci.math. ] > ... 1) 160 T of 187. > this one is a bit hard to fathom. is there an equation here? 2) S of D of 70 is 144. > Sum Divisors 3) The sixth H N is 49/20. > Harmonic Number 4) 12th F N is 12th positive S. > Fibonacci Number Square 5) F of N L of 10000 is 9. > Floor Natural Logarithm 6) 8 C 3 is 56. > Choose 7) 10th T N is 10th F N. > Triangular Number Fibonacci Number 8) S of 1st 5 positive S is 10th F N too. > Sum Squares Fibonacci Number 9) (m+2) C 3 P (m+2) C 2 is S of 1st m positive S. > Choose Plus Choose Sum Squares > S's aren't always equal in an equation: S=Sum or Square. 10) For m >= 2, S of T of m is m T P m D B 2. > ??? > > 11) The N of P L T 50 is 15. > Number Primes Less Than 12) The P of 1st 4 H N is 275/48. > Product Harmonic Numbers 13) D I B 0 and 1 of -ln(1-x)/x is P S D B 6. > Definite Integral Between Pi Squared Divided By 14) L, as m A I, of S of 1st m positive C D B m R B 4 is 1/4. > Limit Approaches Infinity Sum Cubes Divided By Raised By 15) (2m) F D B m F S is A to 4 R B m D B S R of P. > Factorial Divided By Factorial Squared Asymptotic Raised By > Divided By Square Root Pi You got all of your answers correct!... including (15), which I stated wrong. (15) should have been: (2m) F D B m F S is A to 4 R B m D B S R of (m P). (because I forgot about dividing by sqrt(m) too.) As for (1) and (10), the T in (1) is the same the first T in (10). In (1), perhaps it would have been less confusing if I had written: 1) There are 160 T of 187. Leroy Quet === Subject: Re: Mathematical Language Equations Cc: hiraga@slis.tsukuba.ac.jp > As for (1) and (10), the T in (1) is the same the first T in (10). > In (1), perhaps it would have been less confusing if I had written: > 1) There are 160 T of 187. Totatives > 10) For m >= 2, S of T of m is m T P m D B 2. Sum of Totatives of m is m Times Phi(m) Divided By 2. # Perhaps P m should have been T of m Confession 1: Though I got the answer quick (observing 187 = 11*17 and 160 = (11-1)*(17-1)), I didn't know the word totative and had to look it up. Confession 2: Couldn't get as much answers as Mitch Harris did. - Yuzuru Hiraga === Subject: Finite set containment question S is a set. {S_n} is a sequence of subsets of S such that S_n's are increasing and U (n=1-->oo)S_n = S. If A is a proper subset of S, is any help in proving or disproving. === Subject: Re: Finite set containment question Adjunct Assistant Professor at the University of Montana. >S is a set. {S_n} is a sequence of subsets of S such that S_n's are >increasing and U (n=1-->oo)S_n = S. If A is a proper subset of S, is >any help in proving or disproving. If either S or A are finite, then yes. (In either case, A is finite; for each x in A there exists n(x) such that x is in S_n(x); let M = max{n(x): x in A}). But if A is infinite, then the answer is there need not be. Let {x_1,...,x_n,...} be an denumerable subset of A, and let S_n = S - {x_(n+1),x_{(n+2)},...} -- It's not denial. I'm just very selective about what I accept as reality. --- Calvin (Calvin and Hobbes) Arturo Magidin magidin@math.berkeley.edu === Subject: Re: Finite set containment question > S is a set. {S_n} is a sequence of subsets of S such that S_n's are > increasing and U (n=1-->oo)S_n = S. If A is a proper subset of S, is > any help in proving or disproving. Consider S=N (the natural numbers), S_n={0..n} and A={2k | k in N} Marc === Subject: Calculating tide times and heights Node Factors and Equilibrium Arguments, together with Harmonic Constituents are used in the calculation of tides. The standard source for such data (XTide) tabulates the node factors and EA's for each constituent for a range of years 1970 - 2038. Can anyone advise how to CALCULATE node factors and EA's - then in my tide prediction s/w I won't have to look up the values, and won't be restricted to the tabulated years. Chris Allonby. === Subject: 2 Part Recursion Puzzle Here is a 2 part (implicit) recursion puzzle. Solve the first puzzle, then use the answer to solve the second puzzle. (Partial credit for getting only the first sequence.) :) {a(m,k)} is an infinite array defined as follows: Each a(m,m) = m. For every positive integer m and n, sum{k=1 to m} floor[a(m,k) *n /m] = sum{k=1 to n} a(k,m). (Note that order of m and k in the a()'s has been exchanged and depends on which side of the equation the a() is on.) What is the closed-form for {a(k,m)}? -- b(1) = 1; b(m) = sum{k=1 to m} a(k,m) *b(a(k,m)) for all positive integers m. What is {b(m)}? My opinion is the answers to this puzzle are surprising and not too complicated. Leroy Quet === Subject: Re: Mass versus Weight > The ratio of the gravitational weight [w], or heavieness of any > object; Hey Dumb Donny Head, you ing STILL don't know the difference among gravitational, inertial, active, and passive mass. You are dysfunctional intellectual dog , Dumb Donny Head. -- Uncle Al http://www.mazepath.com/uncleal/ (Toxic URL! Unsafe for children and most mammals) Quis custodiet ipsos custodes? The Net! === Subject: Re: Mass versus Weight > The ratio of the gravitational weight [w], or heavieness of any > object; body, or mass of matter, divided by the acceleration [g] at > which it will free fall is its mass: For any given body: Its mass m = > w/g = f/a; anywhere; anytime! Please notice that since Newton made up the quantity of matter as consisting of its bulk and density, and said that it was what he meant when he referred to bodies, and masses of matter; that the two definitions are compatable: That is for any given volume, the weight-density of matter varies in proportion to the weight-density of water at its maximum density here in a laboratory on Earth; where the weight-density of water has been arbitrarily established as the standard. === Subject: Checking a purported Cailey table by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LFrxn13260; Is there an O(n^2) algorithm to verify that an n by n table is the multiplication table of a group ? In the general case, verifying associativity would require O(n^3) time. However, in a group, we are in a particular case. Identity and inverses can be verified in O(n^2), and we can do a few more things in O(n^2). For example, we can check that the table is a Latin square (this corresponds to the simplification rule, whose proof depends on associativity), or we can check things like a*(a^(-1)*b) = b, whose validity also requires associativity. Maybe those tests give us partial information about associativity, and we could make use of that information ? I wonder if it would be possible to combine a set of necessary conditions, each verifiable in O(n^2), to get a sufficient condition. J. Willekens === Subject: Re: Checking a purported Cailey table > Is there an O(n^2) algorithm to verify that an n by n table > is the multiplication table of a group ? Snip There is no general method. I have a procedure in [1] that can identify any Moufang Loop (groups plus octonions), with up to 73 elements, presented as a Cayley table. It checks for Frobenius conservative multiplication in O(n^2) (I think - I have not checked this), and uses a generalised signature to identify most tables. A battery of procedures is needed for difficult groups. In the worst cases the presence of a specific subgroup or quotientgroup must be tested. I do not know if GAP includes an identification procedure for Cayley tables. Roger Beresford. [1] http://wolfram.com.Infocenter/MathSource/4894 === Subject: Re: schools for topology > Can anyone tell me which schools in southern california is good in > topology/algebraic topology? I cannot attest to how good the schools are in these topics, but you can look at (I assume by Southern Cal you mean the LA area as opposed to San Diego). These are in no order of precedance. 1. http://math.usc.edu/programs/currentGradDegrees.html#Ph.D.%20in%20Mathematic s 2. http://www.math.ucla.edu/grad_programs/faculty/research_areas/top.html 3. http://www.math.caltech.edu/people/profs.html I am sure there are others at other UC and CS schools. It might also help to peruse this listing as well for many of the great schools from san fran down to san diego are represented. http://www.gradschools.com/listings/cal/mathematics_cal.html HTH, Flip === Subject: Re: Antidiagonal, Infinity > There are no cases to consider here. The hypothesis says we are given > a mapping f: N -> R. That means the elements of the list are real > numbers, not digit strings. We may choose to represent the numbers in > that list in any way we find convenient. Ok, and if someone chooses a list in such a way that the diagonal number is the same as the anti-diagonal number then we have a seperate case, that only comes up when using the binary representation of the reals. > There seems to be a widespread misconception that the diagonal argument > is not complete until we have explained how to apply it in every > conceivable base. That's tantamount to suggesting that the argument is > not complete until we have translated it into every known language, > including Sanskrit and Klingon. Just one correct proof is quite > sufficient. In fact, if your 12-year-olds have trouble grasping this > point, I submit that it is more important that the diagonal proof itself, > since it gets to the heart of the matter of what constitutes a valid > proof. I'd go furthers, and say there is a widespread misconception that Cantor's Arguement is false! And what if Klingon's only count in binary? How do you propose to convince them of the validity of the proof? Again I'm simply looking for more tools to explain the proof in any base given any example. I realize mathamaticians go for the once it's solved, it's solved sort of reasoning but that isn't my goal, I simply want to understand more about the nature of the proof so I can explain it more convincingly to doubters. If they hand me a special case like the binary dual representation example, I want to be able to contruct a counter example on my own that they're more likely to agree to. > If you really think your five-paragraph explanation is more > understandable to 12-year-olds than any of my three one-sentence > descriptions, then have at it. Well my 5 paragraphs of arugmentation isn't ment to convince th middle school student, it's ment to explain it to you. If I was explaining it to a 12 year old, I would have drawn the list, showing from there how to select a number from P to fill each hole in the number Q. I can use the pair wise swapping to convince most 12 year olds, but at this point my main concern is, did I manage to construct a number not on the list? Is my reasoning valid here? Is the contruction I gave sufficent to demonstrate the uncountability of the reals? === Subject: Re: Antidiagonal, Infinity >> There are no cases to consider here. The hypothesis says we are given >> a mapping f: N -> R. That means the elements of the list are real >> numbers, not digit strings. We may choose to represent the numbers in >> that list in any way we find convenient. Choosing a list means specifying a mapping f: N -> R. Notice that choosing a list does not include specifying a base. It is possible to specify real numbers without using a base at all, for example as e-1-1 or as sup{x-1: x > 0 and x*x < 1+1}. More importantly, it is not part of the hypothesis that the value of f(n) is specified at all for any n; all that is required is that f(n) exist and be a real number for each n. It's not necessary that I be able to write it down and hand it to you. The point is, I am not thinking in terms of a list of digit strings in some implied base when I begin the diagonal argument. That's why there are no cases to consider. All I need to know is that for each n, f(n) is a real number and therefore has a decimal representation. Construction of the table of digit strings is a part of the diagonal process, not a given. I choose to use decimal and not binary because the argument works better that way, and it's my argument, so I get to choose. For each n, f(n) has a corresponding decimal representation. All I need to know is that this representation exists. Again, I may not be able to write it down and hand it to you. > Ok, and if someone chooses a list in such a way that the diagonal > number is the same as the anti-diagonal number then we have a seperate > case, that only comes up when using the binary representation of the > reals. I have been avoiding using terms such as diagonal number and anti-diagonal number in my discussion throughout, since I consider them misleading and confusing. Perhaps you mean that the diagonal number is the number obtained directly from the diagonal, without changing any digits, and the anti-diagonal number is the number obtained from the diagonal number in some systematic way, such as changing each each 1 to a 2 and each non-1 to a 1. Please correct me if this is not what you mean by those terms. Thus the diagonal depends on the original function f: N -> R and the base we choose to represent the numbers. The anti-diagonal is derived from the diagonal by changing each digit according to a specific rule of our choosing. If I have correctly interpreted your terminology, then I have three objections to your premise stated here. First, if I am using a properly constructed diagonal rule, it is impossible for anyone to choose a list in such a way that the anti-diagonal number is a dual-representation number. That's because every dual-representation number in base 10 ends in all 0's or all 9's, and we have specifically avoided those digits in constructing the anti-diagonal. Second, the validity of the argument does not depend on anyone choosing anything. In order for the argument to be valid, it must be the case that the anti-diagonal is not a member of the range of f for any f: N -> R. To be sure of that, we need to know that the anti-diagonal is not a dual-representation number for any f. Third, the failure is not caused by the diagonal being the same as the anti-diagonal. Failure is caused by the anti-diagonal being in the range of f, and the diagonal need not be in the range of f. >> There seems to be a widespread misconception that the diagonal argument >> is not complete until we have explained how to apply it in every >> conceivable base. That's tantamount to suggesting that the argument is >> not complete until we have translated it into every known language, >> including Sanskrit and Klingon. Just one correct proof is quite >> sufficient. In fact, if your 12-year-olds have trouble grasping this >> point, I submit that it is more important that the diagonal proof itself, >> since it gets to the heart of the matter of what constitutes a valid >> proof. > I'd go furthers, and say there is a widespread misconception that > Cantor's Arguement is false! And what if Klingon's only count in > binary? How do you propose to convince them of the validity of the > proof? Again I'm simply looking for more tools to explain the proof in > any base given any example. I realize mathamaticians go for the once > it's solved, it's solved sort of reasoning but that isn't my goal, I > simply want to understand more about the nature of the proof so I can > explain it more convincingly to doubters. If they hand me a special > case like the binary dual representation example, I want to be able to > contruct a counter example on my own that they're more likely to agree > to. There are plenty of proofs of the uncountability of the reals that don't depend on decimal representation or any other base. All that's needed is the definition of the reals, and in particular the completeness property. I suspect Klingon mathematicians would understand such arguments. Even if Klingons count in binary, it shouldn't be hard to explain the concept of decimal representation to a Klingon mathematician. Fortunately, that's not a problem you will have in explaining the proof to 12-year-old earthlings. >> If you really think your five-paragraph explanation is more >> understandable to 12-year-olds than any of my three one-sentence >> descriptions, then have at it. > Well my 5 paragraphs of arugmentation isn't ment to convince th middle > school student, it's ment to explain it to you. If I was explaining it > to a 12 year old, I would have drawn the list, showing from there how > to select a number from P to fill each hole in the number Q. I can > use the pair wise swapping to convince most 12 year olds, but at this > point my main concern is, did I manage to construct a number not on > the list? Is my reasoning valid here? Is the contruction I gave > sufficent to demonstrate the uncountability of the reals? I have not gone through all the details of your construction, but I notice that someone else in this thread has already posted a counterexample to your claim. The basic idea is that you don't know what P and Q might be when you are constructing your rule. -- Dave Seaman Judge Yohn's mistakes revealed in Mumia Abu-Jamal ruling. === Subject: Re: Antidiagonal, Infinity >> There are no cases to consider here. The hypothesis says we are given >> a mapping f: N -> R. That means the elements of the list are real >> numbers, not digit strings. We may choose to represent the numbers in >> that list in any way we find convenient. > Choosing a list means specifying a mapping f: N -> R. Notice that > choosing a list does not include specifying a base. It is possible to > specify real numbers without using a base at all, for example as e-1-1 or > as sup{x-1: x > 0 and x*x < 1+1}. More importantly, it is not part of > the hypothesis that the value of f(n) is specified at all for any n; > all that is required is that f(n) exist and be a real number for each n. > It's not necessary that I be able to write it down and hand it to you. > The point is, I am not thinking in terms of a list of digit strings in > some implied base when I begin the diagonal argument. That's why there > are no cases to consider. All I need to know is that for each n, f(n) > is a real number and therefore has a decimal representation. > Construction of the table of digit strings is a part of the diagonal > process, not a given. I choose to use decimal and not binary because the > argument works better that way, and it's my argument, so I get to choose. > For each n, f(n) has a corresponding decimal representation. All I need > to know is that this representation exists. Again, I may not be able to > write it down and hand it to you. > Ok, and if someone chooses a list in such a way that the diagonal > number is the same as the anti-diagonal number then we have a seperate > case, that only comes up when using the binary representation of the > reals. > I have been avoiding using terms such as diagonal number and > anti-diagonal number in my discussion throughout, since I consider them > misleading and confusing. > Perhaps you mean that the diagonal number is the number obtained directly > from the diagonal, without changing any digits, and the anti-diagonal > number is the number obtained from the diagonal number in some systematic > way, such as changing each each 1 to a 2 and each non-1 to a 1. Please > correct me if this is not what you mean by those terms. > Thus the diagonal depends on the original function f: N -> R and the base > we choose to represent the numbers. The anti-diagonal is derived from > the diagonal by changing each digit according to a specific rule of our > choosing. > If I have correctly interpreted your terminology, then I have three > objections to your premise stated here. First, if I am using a properly > constructed diagonal rule, it is impossible for anyone to choose a list > in such a way that the anti-diagonal number is a dual-representation > number. That's because every dual-representation number in base 10 ends > in all 0's or all 9's, and we have specifically avoided those digits in > constructing the anti-diagonal. > Second, the validity of the argument does not depend on anyone choosing > anything. In order for the argument to be valid, it must be the case > that the anti-diagonal is not a member of the range of f for any f: N - R. To be sure of that, we need to know that the anti-diagonal is not a > dual-representation number for any f. > Third, the failure is not caused by the diagonal being the same as the > anti-diagonal. Failure is caused by the anti-diagonal being in the range > of f, and the diagonal need not be in the range of f. >> There seems to be a widespread misconception that the diagonal argument >> is not complete until we have explained how to apply it in every >> conceivable base. That's tantamount to suggesting that the argument is >> not complete until we have translated it into every known language, >> including Sanskrit and Klingon. Just one correct proof is quite >> sufficient. In fact, if your 12-year-olds have trouble grasping this >> point, I submit that it is more important that the diagonal proof itself, >> since it gets to the heart of the matter of what constitutes a valid >> proof. > I'd go furthers, and say there is a widespread misconception that > Cantor's Arguement is false! And what if Klingon's only count in > binary? How do you propose to convince them of the validity of the > proof? Again I'm simply looking for more tools to explain the proof in > any base given any example. I realize mathamaticians go for the once > it's solved, it's solved sort of reasoning but that isn't my goal, I > simply want to understand more about the nature of the proof so I can > explain it more convincingly to doubters. If they hand me a special > case like the binary dual representation example, I want to be able to > contruct a counter example on my own that they're more likely to agree > to. > There are plenty of proofs of the uncountability of the reals that don't > depend on decimal representation or any other base. All that's needed is > the definition of the reals, and in particular the completeness property. > I suspect Klingon mathematicians would understand such arguments. > Even if Klingons count in binary, it shouldn't be hard to explain the > concept of decimal representation to a Klingon mathematician. > Fortunately, that's not a problem you will have in explaining the proof > to 12-year-old earthlings. >> If you really think your five-paragraph explanation is more >> understandable to 12-year-olds than any of my three one-sentence >> descriptions, then have at it. > Well my 5 paragraphs of arugmentation isn't ment to convince th middle > school student, it's ment to explain it to you. If I was explaining it > to a 12 year old, I would have drawn the list, showing from there how > to select a number from P to fill each hole in the number Q. I can > use the pair wise swapping to convince most 12 year olds, but at this > point my main concern is, did I manage to construct a number not on > the list? Is my reasoning valid here? Is the contruction I gave > sufficent to demonstrate the uncountability of the reals? > I have not gone through all the details of your construction, but I > notice that someone else in this thread has already posted a > counterexample to your claim. The basic idea is that you don't know what > P and Q might be when you are constructing your rule. Even if someone insists on using binary, use pairs of digits, instead of single digits, to make it effectively base 4, or use triples of digits for base 8, and so on. In any base greater than 3, it works. === Subject: Re: Antidiagonal, Infinity > I'd go furthers, and say there is a widespread misconception that > Cantor's Arguement is false! It's only widespread on this newsgroup. === Subject: Re: Geometric interpretation of the trace of a matrix? >> If you start with unitary eigenvectors of the transformation, as sides >> of a paralelepiped, the trace is certainly what I said in the last >> post, i.e., the sum of the edges of the image paralelepiped divided by >> 2^{n-1}. >Almost correct. Your willingness to state that MJimenez's quoted statement is almost correct entails that you have an idea of what it is supposed to mean. I would be grateful for any effort you'd be willing to put into explaining it to me (and explaining why it's almost correct); apparently I'm missing something very badly, because I cannot make any sense of it at all. >You have to be careful with the signs of the eigenvalues. >A modification like: if the transformation reverses top and bottom >(or left and right etc), then the length of the corresponding edge >has to be counted with a minus sign. >When your transformation has complex eigenvalues, you are more or less >out of luck (ok, you may be able to do something with rotations) >If you make the further assumption that the transformation is symmetric, >then you can choose the parallelpiped to be rectangular. All the eigenvalues >of a symmetric transformation are real, so in that case a geometric interpretation >is possible. I am totally lost. Lee Rudolph === Subject: Re: Periodic function >A function like cos x^2 produces a cosine wave that compresses itself in >both directions away from x=0. What function would produce regularly >periodic compressions and rarefactions? Since no one else appears to be pointing out what you might think is obviously, I'll do it: any periodic function is a sum of simple sines and cosines, that is, it may be expressed as sum ( a_n sin( n x ) + b_n cos( n x ) ) where, unlike other posters, I am not including anything more complicated than multiples of x as arguments to sine and cosine. You choose the coefficients a_n and b_n to make the curve have the desired shape on one period, and after that, periodic behaviour is guaranteed. You won't get the infinite wiggling that some people seem to like but you can match something like cos(x^2) pretty well. For example, an 11-term sum gives a good approximation which clearly exhibits the periodic bunching: f(x) = 0.1874450324 + 0.5095889344 cos(x) + 0.5262043454 cos(2 x) + 0.08496928957 cos(3 x) - 0.5839140525 cos(4 x) + 0.3871804531 cos(5 x) - 0.1330855634 cos(6 x) + 0.01110860988 cos(7 x) + 0.03019173362 cos(8 x) - 0.04150460650 cos(9 x) + 0.02181582400 cos(10 x) dave === Subject: Re: Periodic function Just a clarification. Referring to Ignacio's previous relation on 2002-12-15,is f(x+Pi/2)=(1+f(x))/(1-f(x)) right for all 2 Pi period functions? Functions of sin,cos do not fit in here .. === Subject: Re: Periodic function > Just a clarification. Referring to Ignacio's previous relation on > 2002-12-15, Are you referring to a post by ILC in this newsgroup on 2002-12-15? I cannot find any such post. It would have been nice if you had provided a link. > is f(x+Pi/2)=(1+f(x))/(1-f(x)) right for all 2 Pi period > functions? Functions of sin,cos do not fit in here .. Of course it's not right. Why did you bother to ask? DWC === Subject: Re: Periodic function >> Just a clarification. Referring to Ignacio's previous relation on >> 2002-12-15, > Are you referring to a post by ILC in this newsgroup on 2002-12-15? > I cannot find any such post. It would have been nice if you had provided > a link. I think that it is or some other post from the same thread. Jose Carlos Santos === Subject: Re: Periodic function >> Just a clarification. Referring to Ignacio's previous relation on >> 2002-12-15, > Are you referring to a post by ILC in this newsgroup on 2002-12-15? > I cannot find any such post. It would have been nice if you had > provided a link. > I think that it is > @ID-137122.news.dfncis.de > or some other post from the same thread. > Jose Carlos Santos Doh! I've learned something today. Google has put this new thread onto the end of that old thread, presumably because they have the same title, making it all seem to be just one big thread! That's useful information (and it's probably why GLN didn't think any link would be needed). Anyway, as to the clarification requested by GLN: > if for all x > f(x+a) = (1+f(x))/(1-f(x)) > then function f is periodic. but of course, the reverse implication does not hold. David === Subject: Re: Periodic function >> Just a clarification. Referring to Ignacio's previous relation on >> 2002-12-15, Are you referring to a post by ILC in this newsgroup on 2002-12-15? > I cannot find any such post. It would have been nice if you had > provided a link. > I think that it is > @ID-137122.news.dfncis.de > or some other post from the same thread. > Jose Carlos Santos > Doh! I've learned something today. Google has put this new thread onto > the end of that old thread, presumably because they have the same title, > making it all seem to be just one big thread! That's useful information > (and it's probably why GLN didn't think any link would be needed). Yes,sorry about that confusion. > Anyway, as to the clarification requested by GLN: > if for all x > f(x+a) = (1+f(x))/(1-f(x)) > then function f is periodic. > but of course, the reverse implication does not hold. > David But when tan(Pi/4+x)=(1+tan(x))/(1-tan(x)) by componendo dividendo we get tan(x)=tan(Pi/4 + x)-1)/(tan(Pi/4 + x)+1). You mean to generalize this per ILC original posting, we can't now === Subject: transform PDF from plane to sphere Hi all, I have a question regarding the tranformation of a probability density distribution (pdf) taken at a plane to the surface of a sphere. An exampel : A distribution is uniform (isotropically) distributed on a x,y-plane with marginal pdf's p(x) = 1; p(y) = 1; On a sphere, the marginal pdf for the isotropic distribiution is p(theta) = 1/2 sin(theta) according to mathworld.wolfram.com/SpherePointPicking.html My question: For a pdf on the plane that has a (truncated) Gaussian as marginal pdf in x direction and a uniform distribution in y direction, how do I map such a distributon onto the sphere, or better on a hemisphere (which should make it easier) ? The strips parallel to the equator should be uniform again (y). I know that the Fisher distribution on the sphere is the analogue to the Gaussian in the plane, but I have no idea if I can perform a transformation and how. My first idea was to do the same as with the isotropical distribution : perfoming a weighting of the pdf on the plane by sin(theta) to map it to the sphere. But this is just an idea, I do not have a glue if this is the right way. Any hints are very welcome, Dirk === Subject: OT: The Kabatoff Posts It was very informative. I have killfile'd Mr. Kabatoff to avoid his endless off-topic posts on numerology. Now my only problem is that every Kabatoff post gets a refutation post, so that even after ponking Mr. Kabatoff, the newsgroups are still cluttered with Kabatoff-related posts. One such post is certainly helpful, but can't we now assume that everyone here knows who Daryl Kabatoff is, and let us get back to the topic? === Subject: Re: The Kabatoff Posts > It was very informative. I have killfile'd Mr. Kabatoff to avoid his endless > off-topic posts on numerology. > Now my only problem is that every Kabatoff post gets a refutation post, so > that even after ponking Mr. Kabatoff, the newsgroups are still cluttered > with Kabatoff-related posts. One such post is certainly helpful, but can't > we now assume that everyone here knows who Daryl Kabatoff is, and let us get > back to the topic? frequently quotes this libel using the name Thomas... and you appreciate this? Following is some mathematics that you similarly don't care to see: 243 <- 1 Samuel 7 274 <- 2 Samuel 7 --- 517 298 <- 1 Kings 7 320 <- 2 Kings 7 --- 618 <-verses in Bible Book 7 345 <- 1 Chronicles 7 374 <- 2 Chronicles 7 --- 719 See above that First Kings 7 and Second Kings 7 are chapters 298 and 320, together for the 618 verses of Bible Book 7. And First Samuel 7 and Second Samuel 7 averages with First Chronicles 7 and Second Chronicles 7 for the 618 verses of Bible Book 7 (243+274+345+374=618+618). And First Chronicles 7 and Second Chronicles 7 are chapters 345 and 374, together for the 719 verses of Bible Book 12, pretty as 12 is the 7th non-prime while 719 is the 128th or the 2 to the 7th prime. See Books by Bonnie Gaunt for a Gematric discussion of the number 618. The first 4 primes plus the first 4 non-primes add together for the 36 chapters of Bible Book 4 Numbers: Primes Non-Primes 2 1 3 4 5 6 7 <-4th-> 8 -- -- 17 19 The first 4 primes in prime positions add together for the 36 chapters of Bible Book 4: Primes In Prime Primes Positions 1 2 2 3 <- 3 3 5 <- 5 4 7 5 11 <- 11 6 13 7 17 <- 17 -- 36 Seven plus the 7th prime plus the 7th non-prime adds together for the 36 chapters of Bible Book 4, pretty as 7 is the 4th prime: Primes Non-Primes 2 1 3 4 5 6 7 8 11 9 13 10 17 <-7th-> 12 -- -- 58 50 === Subject: Re: OT: The Kabatoff Posts > It was very informative. I have killfile'd Mr. Kabatoff to avoid his endless > off-topic posts on numerology. > Now my only problem is that every Kabatoff post gets a refutation post, so > that even after ponking Mr. Kabatoff, the newsgroups are still cluttered > with Kabatoff-related posts. One such post is certainly helpful, but can't > we now assume that everyone here knows who Daryl Kabatoff is, and let us get > back to the topic? Depending on your newsreader, you may be able to tell it to ignore a thread started by Kabatoff, rather than simply filtering out his post. -- Will Twentyman email: wtwentyman at copper dot net === Subject: Simple question A question about sets: Is there any topological difference between Q (the rationals) and A (or Q bar) (the algebraics)? Barnaby === Subject: Re: Simple question >A question about sets: Is there any topological difference between Q (the >rationals) and A (or Q bar) (the algebraics)? No (as others have said) if you mean the real algebraics. But algebraics by itself means algebraic complex numbers. In that case I think it's not so obvious. Robert Israel israel@math.ubc.ca Department of Mathematics http://www.math.ubc.ca/~israel University of British Columbia Vancouver, BC, Canada V6T 1Z2 === Subject: Re: Simple question >>A question about sets: Is there any topological difference between Q (the >>rationals) and A (or Q bar) (the algebraics)? >> >No (as others have said) if you mean the real algebraics. But >algebraics by itself means algebraic complex numbers. In that case >I think it's not so obvious. Correct, it is not as obvious. But we have discussed here before that Q^2 is homeomorphic to Q. See the thread -- Stephen J. Herschkorn herschko@rutcor.rutgers.edu === Subject: Re: Simple question > A question about sets: Is there any topological difference between Q (the > rationals) and A (or Q bar) (the algebraics)? > Barnaby If I remember correctly: In terms of order, if a set is countable, linearly ordered (every two members are comparable), densely ordered (between every two different members there is at least one, different from the two), without a maximal and without a minimal member, then it is order isomorphic to the rationals. Since the presumed topologies (inherited from the reals) are generated by open intervals, the sets Q and A must be homeomorphic. === Subject: Re: Simple question >> A question about sets: Is there any topological difference between Q (the >> rationals) and A (or Q bar) (the algebraics)? >> Barnaby >If I remember correctly: >In terms of order, if a set is > countable, > linearly ordered (every two members are comparable), > densely ordered (between every two different members there > is at least one, different from the two), > without a maximal and without a minimal member, >then it is order isomorphic to the rationals. >Since the presumed topologies (inherited from the reals) >are generated by open intervals, the sets Q and A must >be homeomorphic. QUESTION: Can we find an explicit bijection f : Q -> A such that f(x) > f(y) precisely when x > y? -- John Adams served two terms as Vice President and one as President, but lost reelection. Later his son became President despite losing the popular vote. That son lost his reelection attempt badly. Now history is repeating itself. pmontgom@cwi.nl Microsoft Research and CWI Home: San Rafael, California === Subject: Re: Simple question >A question about sets: Is there any topological difference between Q (the >rationals) and A (or Q bar) (the algebraics)? I don't remember the details right now (I'm tired today), but I think any countable, totally ordered, dense set without a mimimum or maximum element is order-isomorphic to Q. (Dense means there is an element between any two distinct elements.) Since the order topology is used, this fact implies that the answer to your question is negative. I might be wrong, though. -- Stephen J. Herschkorn herschko@rutcor.rutgers.edu === Subject: Re: Simple question >> A question about sets: Is there any topological difference between Q >> (the >> rationals) and A (or Q bar) (the algebraics)? > I don't remember the details right now (I'm tired today), but I think > any countable, totally ordered, dense set without a mimimum or maximum > element is order-isomorphic to Q. (Dense means there is an element > between any two distinct elements.) Since the order topology is used, > this fact implies that the answer to your question is negative. I > might be wrong, though. I have napped by now, so I'm thinking about how to prove the first fact. I know how to show that the rationals can be order-embedded in any totally ordered dense set with at least two elements. The proof I have in mind uses choice; can we show this fact without choice? By the same method, any countable, dense, totally ordered set with at least two elements can be embedded in another. Does the existence of embeddings in both directions imply that two such sets are order-isomorphic? Again, can the first part be done without choice? If each of two topological spaces can be embedded in the other, are the two spaces isomorphic? I'm asking about order and topological versions of Schroeder-Bernstein here. Is there a general categorical Schroeder-Bernstein theorem? -- Stephen J. Herschkorn herschko@rutcor.rutgers.edu === Subject: Re: Simple question > I know how to show that the rationals can be order-embedded in any > totally ordered dense set with at least two elements. The proof I > have in mind uses choice; can we show this fact without choice? > By the same method, any countable, dense, totally ordered set with at > least two elements can be embedded in another. Does the existence of > embeddings in both directions imply that two such sets are > order-isomorphic? Again, can the first part be done without choice? Actually, I have realized a direct proof that any two c.d.t.-o. sets each with at least two elements are isomorphic. It does not use choice. I still don't know whether the first fact above requires choice. For example, although such a set is infinite, do we know without choice that it is Dedekind infinite? -- Stephen J. Herschkorn herschko@rutcor.rutgers.edu === Subject: Re: Sorry >If xyz = 1, prove that (xy)^2 + (yz)^2 + (zx)^2 - 2x - 2y - 2z + 3 >= 0. >>Writing z = 1/(xy), we want to show that >> 0 >> 2 2 1 1 2 >> <= x y + --- + --- - 2x - 2y - --- + 3 >> x^2 y^2 x y >> 2 2 1 1 2 >> = x y - 2x - 2y + 3 + ( - - - ) >> x y >> 1 1 2 >> = f(x,y) + ( - - - ) >> x y >>Setting the partials of f(x,y) to 0, we get that its minimum occurs at >>(x,y) = (1,1). Since f(1,1) = 0, the inequality holds. >>Rob Johnson take out the trash before replying > The partials of f do vanish at (1, 1), but this is a saddle >point, not a minimum. For example, x = 0.9 and y = 1.1 give > f = 0.9801 - 1.8 - 2.2 + 3 = -.0199 < 0 Right you are. My mistake; I only looked at part of the bounding circle near infinity and missed the small parts near the x and y axes where f(x,y) is negative. Looking at the rest of the circle, f(x,y) appears to grow like r^4, which would make (1,1) a minimum, but for those pesky >Easier for hand computation is x = 0 and y = 2, giving -1. However, x = 0 makes xyz = 1 kind of hard. x = 1/2 and y = 2 is almost as easy, also giving -1. Here is a proper attempt. Let u = xy and v = x+y. Then 0 2 2 1 1 2 <= x y + --- + --- - 2x - 2y - --- + 3 x^2 y^2 x y 2 2 1 1 2 = x y - 2(x+y) + 3 + ( - - - ) x y 2 v^2-4u = u - 2v + 3 + ------ u^2 1 2 2 4 = --- v - 2v + u + 3 - - [1] u^2 u Let us fix u. [1] is a positive quadratic in v whose minimum is at v = u^2. Plugging this back into [1] yields the minimum for a given u to be 3 - 4/u. So unless 0 < u < 4/3, the minimum of the parabola is positive. Note that v^2-4u = (x-y)^2 >= 0. Thus, for any u, we must have that v^2 >= 4u. Since the minimum of the parabola is when v = u^2, when 0 < u < 4/3, v^2 = u^4 < 64/27 u < 4u. Therefore, when 0 < u < 4/3, the minimum of the parabola cannot be reached. Thus, the minimum must be reached when v^2 = 4u. Plugging this back into [1] yields a minimum of u^2 - 4sqrt(u) + 3, which has a minumum of 0. Rob Johnson take out the trash before replying === Subject: Functional system by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LInLg09157; Function f(x) defined in the following system: |f(x + 1) = (sqrt(2))^f(x) |f(-f(x)) = -x |f(0) = 1 Does it exist a direct deterministic way (algorithm) which find the lowest value of f(x) for x in (-2; inf)? Does it exist a continuous non-recursive function g such that g(x) = the lowest value of f(x). Leonel === Subject: Re: EINSTEIN'S RELATIVITY THEORY PROBED einstein did not eat meat for most part of his life. > There is no present. Learn from the past > and plan to apply the lessons in the future. > The future turns into the past continuously. > Some call this process time. > - Jai Maharaj > http://www.mantra.com/jyotish > Einstein's relativity theory probed > By Anna Salleh > ABC Science Online > Einstein's theory of general relativity, fundamental to > our understanding of the cosmos, is about to be put to a > rigorous test, using a satellite due to be launched early > this week. === Subject: Re: Existing of a triangle by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LIt2810154; === Subject: Iteration Theory by Julia/Fatou Does anyone know of any online refs that contain a good outline of the results of the Theory of Iteration by Fatou and Julia? transcendants entieres? PS: I am particularly interested in theorems that analytically expand the iterates f^(n)(z) around fixed points. -- Ioannis Galidakis http://users.forthnet.gr/ath/jgal/ ------------------------------------------ Eventually, _everything_ is understandable === Subject: Re: Why is it ln on the calculator? Originator: jeyadev@kaveri >> This is one of the examples which can remind you that mathematics >>is international. We must expect to find abbreviations for words in >>languages other than English. Another example is the set of all integers >> {.... -3, -2, -1, 0, 1, 2, 3, 4, ....} >>whose standard modern name is Z, short for the German word Zahlen >>(numbers). Then there's the well-known use of the Greek letter pi, >>which is short for the Greek word for periphery, i.e. the distance >>around a circle of unit diameter. >As coined by such well-known Greeks as Oughtred, Jones & Euler... :-) But many of these chaps did things in Latin and they were not even Mexican or Argentine or ..... -- Surendar Jeyadev jeyadev@wrc.xerox.bounceback.com Remove 'bounceback' for email address === Subject: Re: Groebner basis calculation > I have an ideal generated by 75 binomials in 110 variables for which > I'd like to find a Groebner basis. I have no idea whether or not this > is a tractable problem. I've left it running in Singular for 2 days on > a 2.4GHz Pentium with no result so far (though it hasn't run out of > memory yet). However my problem is a special case with only binomial > generators so I'm wondering if there is some special purpose code out > there that can solve this problem. I tried just implementing > Buchberger's algorithm with some code specialised for binomials but > that seems far too slow too. Or is this problem just too difficult on > todays PCs? Maybe it would take years or centuries? Is there something > faster than Singular out there that can handle 110 variables? What is > the state of the art in Groebner basis computations? > I is the ideal, defined over the field Q, is given by all binomials of > the form > c_i * A_j - B_i,j > Where j is in {1,2,...,31}, i is in {0,1,2,3,4}, and 2^i appears in > the binary expansion of j. > For example c_0A_3-B_0,3 appears in I because 3=2^0+2^1 but > c_4A_13-B_4,13 doesn't as 13=2^0+2^2+2^3 which doesn't contain 2^4. > I'd like to eliminate the c_i so I'd like to use something like the > 5th elimination order though just plain lexicographic would do as long > as the c_i are greater than the other monomials. > I hope I haven't made too many errors in the above. This isn't really > my field. I'll make a few remarks. First, you might get a better response if you post this sort of question to sci.math.symbolic (since Groebner basis computation methods are among the interests of people who lurk on that group). I'm cross-posting thereto. Second, your ideal has a couple of nice qualities (by the way, I get 80 generators, not 75, assuming I interpreted the criterion correctly). For one, it is toric. For another, the generators already comprise a Groebner basis with respect to term orders wherein the 'b' variables are weighted more heavily than all others (that is, term orders that are elimination orders for those variables). This gives rise to two possible approaches. (1) Use a method that is dedicated to handling toric ideals. (2) Use a Groebner basis conversion method such as the Groebner walk. I do not know offhand whether either or both of these are available in Singular but I suspect this is not hard to learn from the documentation. To get some idea of difficulty I tried this out in Mathematica. I set the range of 'j' to be a parameter so as to test smaller related problems. 1 toricProblem[n_, a_:a, b_:b, c_:c] := Module[ {avars,bvars,cvars, allACproducts, usedACproducts, polys}, cvars = Array[c, n, 0]; avars = Array[a, 2^n-1]; allACproducts = Flatten[Outer[Times, cvars, avars]]; usedACproducts = Cases[allACproducts, a[j_]*c[i_] /; isInBinaryExpansion[i,j]]; polys = Map[#-(#/.a[j_]*c[i_]->b[i,j])&, usedACproducts]; bvars = Cases[Variables[polys], b[__]]; GroebnerBasis[polys, Join[avars,bvars], cvars, MonomialOrder->EliminationOrder, Sort->True] ] Not surprisingly it is easy for n = 3. In[5]:= Timing[gb3 = toricProblem[3];] Out[5]= {0.03 Second, Null} In[7]:= gb3 // InputForm Out[7]//InputForm= {a[7]*b[1, 3] - a[3]*b[1, 7], a[7]*b[0, 3] - a[3]*b[0, 7], a[7]*b[2, 5] - a[5]*b[2, 7], a[7]*b[0, 5] - a[5]*b[0, 7], -(a[5]*b[0, 3]) + a[3]*b[0, 5], a[7]*b[2, 6] - a[6]*b[2, 7], -(a[6]*b[2, 5]) + a[5]*b[2, 6], a[7]*b[1, 6] - a[6]*b[1, 7], -(a[6]*b[1, 3]) + a[3]*b[1, 6], a[7]*b[0, 1] - a[1]*b[0, 7], a[5]*b[0, 1] - a[1]*b[0, 5], a[3]*b[0, 1] - a[1]*b[0, 3], a[7]*b[1, 2] - a[2]*b[1, 7], a[6]*b[1, 2] - a[2]*b[1, 6], a[3]*b[1, 2] - a[2]*b[1, 3], a[7]*b[2, 4] - a[4]*b[2, 7], a[6]*b[2, 4] - a[4]*b[2, 6], a[5]*b[2, 4] - a[4]*b[2, 5], -(b[1, 7]*b[2, 6]) + b[1, 6]*b[2, 7], -(b[0, 7]*b[2, 5]) + b[0, 5]*b[2, 7], -(b[0, 7]*b[1, 3]) + b[0, 3]*b[1, 7], -(a[7]*b[1, 3]*b[2, 6]) + a[3]*b[1, 6]*b[2, 7], -(a[7]*b[0, 3]*b[2, 5]) + a[3]*b[0, 5]*b[2, 7], -(a[3]*b[1, 6]*b[2, 5]) + a[5]*b[1, 3]*b[2, 6], -(b[0, 3]*b[1, 7]*b[2, 5]) + b[0, 5]*b[1, 3]*b[2, 7], -(b[0, 3]*b[1, 6]*b[2, 5]) + b[0, 5]*b[1, 3]*b[2, 6]} For n = 4 the computation is far more substantial, but still not too hard. In[9]:= Timing[gb4 = toricProblem[4];] Out[9]= {18.57 Second, Null} The result is fairly large but again, not out of bounds by any means. In[10]:= Length[gb4] Out[10]= 321 In[11]:= LeafCount[gb4] Out[11]= 5983 Now n = 5 is another matter entirely. I ran out of memory on one attempt (after several minutes). I am now trying it again on a bigaram machine. No telling what will happen. Groebner basis computations can be funny that way. Daniel Lichtblau Wolfram Research === Subject: Re: Groebner basis calculation > I have an ideal generated by 75 binomials in 110 variables for which > I'd like to find a Groebner basis. I have no idea whether or not this > is a tractable problem. Simplifying such a system is indeed intractable -in gereral-. The associated problem of deciding if a polynomial is a member of an likely has a lower case GB computation that is exponential time (or worse, and the best algorithms so far take doubly exponential times in the number of variables. > I've left it running in Singular for 2 days on > a 2.4GHz Pentium with no result so far (though it hasn't run out of > memory yet). However my problem is a special case with only binomial > generators so I'm wondering if there is some special purpose code out > there that can solve this problem. Special cases is a different matter altogether. If your system is conveniently structured, you may be able to do a lot of elimination theoretically (use symmetry, systematic elimination, induction) > I tried just implementing > Buchberger's algorithm with some code specialised for binomials but > that seems far too slow too. Or is this problem just too difficult on > todays PCs? Maybe it would take years or centuries? Is there something > faster than Singular out there that can handle 110 variables? What is > the state of the art in Groebner basis computations? GB computation in general reduces to GB computation with binomials (that is, binomial GB computation is just as hard as the unrestricted case). > I is the ideal, defined over the field Q, is given by all binomials of > the form > c_i * A_j - B_i,j > Where j is in {1,2,...,31}, i is in {0,1,2,3,4}, and 2^i appears in > the binary expansion of j. > For example c_0A_3-B_0,3 appears in I because 3=2^0+2^1 but > c_4A_13-B_4,13 doesn't as 13=2^0+2^2+2^3 which doesn't contain 2^4. > I'd like to eliminate the c_i so I'd like to use something like the > 5th elimination order though just plain lexicographic would do as long > as the c_i are greater than the other monomials. Because of the digit thing, there is a lot of structure here, so there might be hope. Can you do a much much smaller system by program and by hand (to check the program)? like j in {1,2,3}, i in {0,1}? -- Mitch Harris (remove q to reply) === Subject: Re: Frequentist probability confusion Right; the problem would be solved if our number system had genuine >>infinitesimals. Then each integer could be assigned a probability >>1/aleph_0 and everything would work out ok. >Wrong. In non-standard analysis, there exist >infinitesimals, but all of them are much smaller than >anything looking like that. All non-standard positive >integers have at least as many smaller integers as there >are ordinary real numbers. But they behave like finite >integers within the model. I was thinking of the surreals rather than the hyperreals; I believe they have well-defined multiplicative inverses for every non-zero number (including aleph_0) along with the same rules of distributivity as real numbers. I'm don't know very much about this, but wouldn't that be sufficient? > We could >>even imagine a formal definition of probability distributions as >>equivalence classes of algorithms for producing numbers from random >>seeds. Then we could honestly claim to be talking about processes >>generating numbers when we do probability. >This is already the case in probability as we have it >now, but with random seeds being real numbers uniform >between 0 and 1. If that were true there would be a distribution for my process which generates random integers, but there isn't. >> If one allows finitely >> additive probability measures, then one can defined a probability measure >> over the natural numbers that some might want to label uniform. However, >> the construction of this measure uses the the axiom of choice (or at least >> some large portion of the axiom of choice). >>Do you have a reference for this? >One does not need much of the axiom of choice, but some >is needed. If one only wants some of the sets to be >measurable, nothing is needed; consider the field of >sets which are periodic from some point on, and give >it the limiting frequency. But what are you going to >do with it? Just some of the sets being measurable isn't satisfying, but R. === Subject: Help, please: an equation on complex exponentional numbers I really don't know how to proceed with this: Let p>2 be an arbitrary prime number, prove that for all k=1,2,...,p there exist convenient integers a_k and b_k such that for k:=1 to k:=p: {SUM of a_k*[e^(i 2k*PI/p)]}^2 + {SUM of b_k*[e^(i 2k*PI/p)]}^2 + 1 = 0 Danny. === Subject: Re: Help, please: an equation on complex exponentional numbers >I really don't know how to proceed with this: >Let p>2 be an arbitrary prime number, prove that for all k=1,2,...,p >there exist convenient integers a_k and b_k such that for k:=1 to >k:=p: >{SUM of a_k*[e^(i 2k*PI/p)]}^2 + {SUM of b_k*[e^(i 2k*PI/p)]}^2 + 1 = >Danny. What is e^(i * Pi) ? -- Jeremy Boden === Subject: Re: Help, please: an equation on complex exponentional numbers >I really don't know how to proceed with this: >Let p>2 be an arbitrary prime number, prove that for all k=1,2,...,p >there exist convenient integers a_k and b_k such that for k:=1 to >k:=p: >{SUM of a_k*[e^(i 2k*PI/p)]}^2 + {SUM of b_k*[e^(i 2k*PI/p)]}^2 + 1 = >0 > What is e^(i * Pi) ? Are you asking because you don't know, or because you think the OP doesn't know? e^(i*Pi) is minus one, but the problem involves numbers of the form e^(i 2k*PI/p), which is a different kettle of fish. -- Gerry Myerson (gerry@maths.mq.edi.ai) (i -> u for email) === Subject: Re: Theory behind algebra conversation with Dave Rusin.] I went to public school in Nassau County, New York. I *think* we were taught about induction in eleventh-grade math in the mid 70's. I am fairly certain it was never covered at college. Perhaps it was taught in twelfth-grade math, which was not a required coursed. Now that I think about it, that is more likely the case. Have entering college freshmen generally taken high-school-senior-level math? True, I did pick up induction on my own before it was ever covered in any class, so maybe I'm just imagining that my classmates learned it as well. I have no high-school math text at my disposal to see if it is in the curriculum. -- Stephen J. Herschkorn herschko@rutcor.rutgers.edu === Subject: Re: Theory behind algebra >Where can I find books that rigorously proves all of the facts, >theorems, and properties one is taught in High School algebra? You might enjoy thinking about the following issue, treated carefully in a recent preprint by Wilkies, http://www.maths.ox.ac.uk/~wilkie/ . This is apparently a question of Tarski: are there true statements you can make about natural numbers using the symbols + * ^ 1 which do not follow from the axioms of high school algebra? That is, we ask students to know the 2 commutative, 2 associative, and 1 distributive laws, the properties of 1 (1 for * and 2 for ^), and the 3 laws of exponents (add exponents, multiply exponents, multiply bases). We sort of promise them that this is all they need to use to complete their calculations. Well guess what: the following statement is true for all natural numbers x,y: [(x + 1)^x + (x^2 + x + 1)^x]^y * [(x^3 + 1)^y + (x^4 + x^2 + 1)^y]^x = [(x + 1)^y + (x^2 + x + 1)^y]^x * [(x^3 + 1)^x + (x^4 + x^2 + 1)^x]^y However, it does _not_ follow from the 11 first-order axioms cited! So rigorously prov[ing] all of the facts, theorems, and properties one is taught in High School algebra is perhaps a harder task than one might think! dave === Subject: Re: Theory behind algebra >You might enjoy thinking about the following issue, treated carefully >in a recent preprint by Wilkies, http://www.maths.ox.ac.uk/~wilkie/ . >This is apparently a question of Tarski: are there true statements >you can make about natural numbers using the symbols + * ^ 1 which do >not follow from the axioms of high school algebra? That is, we ask >students to know the 2 commutative, 2 associative, and 1 distributive laws, >the properties of 1 (1 for * and 2 for ^), and the 3 laws of exponents >(add exponents, multiply exponents, multiply bases). We sort of promise >them that this is all they need to use to complete their calculations. >Well guess what: the following statement is true for all natural numbers x,y: > [(x + 1)^x + (x^2 + x + 1)^x]^y * [(x^3 + 1)^y + (x^4 + x^2 + 1)^y]^x = > [(x + 1)^y + (x^2 + x + 1)^y]^x * [(x^3 + 1)^x + (x^4 + x^2 + 1)^x]^y >However, it does _not_ follow from the 11 first-order axioms cited! >So rigorously prov[ing] all of the facts, theorems, and properties one >is taught in High School algebra is perhaps a harder task than one might >think! Sorry about being too lazy to think through this. What other axiom(s) do you need? Is induction derivable from the stated axioms? -- Stephen J. Herschkorn herschko@rutcor.rutgers.edu === Subject: Re: Theory behind algebra >>Where can I find books that rigorously proves all of the facts, >>theorems, and properties one is taught in High School algebra? > You might enjoy thinking about the following issue, treated carefully > in a recent preprint by Wilkies, http://www.maths.ox.ac.uk/~wilkie/ . > This is apparently a question of Tarski: are there true statements > you can make about natural numbers using the symbols + * ^ 1 which do > not follow from the axioms of high school algebra? That is, we ask > students to know the 2 commutative, 2 associative, and 1 distributive > laws, > the properties of 1 (1 for * and 2 for ^), and the 3 laws of exponents > (add exponents, multiply exponents, multiply bases). We sort of promise > them that this is all they need to use to complete their calculations. > Well guess what: the following statement is true for all natural numbers > x,y: > [(x + 1)^x + (x^2 + x + 1)^x]^y * [(x^3 + 1)^y + (x^4 + x^2 + 1)^y]^x = > [(x + 1)^y + (x^2 + x + 1)^y]^x * [(x^3 + 1)^x + (x^4 + x^2 + 1)^x]^y > However, it does _not_ follow from the 11 first-order axioms cited! > So rigorously prov[ing] all of the facts, theorems, and properties one > is taught in High School algebra is perhaps a harder task than one might > think! Hmmm. Well that's because Wilkie's theory avoids subtraction (natural in this first-order set-up as using negatives rapidly takes you out of Z) so I shouldn't think a bright schookid would be unable to do this :-) -- Robin Chapman, www.maths.ex.ac.uk/~rjc/rjc.html Lacan, Jacques, 79, 91-92; mistakes his penis for a square root, 88-9 Francis Wheen, _How Mumbo-Jumbo Conquered the World_ === Subject: Erdos number for sale Seems to be for real! http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=3189039958 === Subject: Re: calculators |> |> How does a calculator do complicated math computaitons so quickly? |> For example, finding the sine of 23.71 degrees. |> |> |> I read that many calculators for trig and log functions simply use |> lookup tables so finding the sine of any number is just a matter of |> pulling a value from a permanent memory, which is why it can be computed |> so quickly. |> I'm pretty sure that it uses a taylor series, because if you try to |> find the sine of very large angles, you will see they become |> increasingly inaccurate. I use TI-83+ Silver Edition, and TI-30X There is a trivial reason why the calculation of sines of large arguments gets increasingly more inaccurate: since the argument is kept in a finite representation (typically 32bit or 64 bit on modern computers), it looses about half a decimal digit (2*pi is about 10/2 here..) in accuracy for each additional period added to the argument. These bits that store the most significant bits of the argument are lost and not available to represent the phase, but the least significant bits would be the ones actually needed (and the ones left after reduction to [0,2*pi]). Example: supposed you use a 32bit floating point representation with about 8 decimal digits of accuracy. If you are to compute the sine of the number 12345678.9, the binary representation allows to store this as something close to 12345678, but does not allow to store the fraction 0.9. Without the knowledge on the trailing fractions, you have no way to figure out what the sine of the full representation would be. === Subject: Re: calculators Keywords: sine function, series expansion, pocket calculator |> How does a calculator do complicated math computaitons so quickly? |> For example, finding the sine of 23.71 degrees. There exist Chebyshev series expansions for the sine, in Math Comp 34 (149) (1980) pp 237-244 or the one by Clenshaw in Math Tabl Aids Comput 8 (47) (1954) pp 143-147. To obtain an IEEE single precision accuracy for sin(Pi*x/2)/x one needs 8 terms, for double precision 16 terms. === Subject: Re: calculators >|> How does a calculator do complicated math computaitons so quickly? >|> For example, finding the sine of 23.71 degrees. >|There exist Chebyshev series expansions for the sine, > in Math Comp 34 (149) (1980) pp 237-244 or the one by Clenshaw > in Math Tabl Aids Comput 8 (47) (1954) pp 143-147. To obtain > an IEEE single precision accuracy for sin(Pi*x/2)/x > one needs 8 terms, for double precision 16 terms. The CORDIC algorithm is commonly used for calculating trig functions. See for a brief description in connection with the original electronic slide rule calculator. A Google search for CORDIC turns up many hits. -- Dave Seaman Judge Yohn's mistakes revealed in Mumia Abu-Jamal ruling. === Subject: Re: calculators Originator: jeyadev@kaveri >> I read that many calculators for trig and log functions simply use >> lookup tables so finding the sine of any number is just a matter of >> pulling a value from a permanent memory, which is why it can be computed >> so quickly. >I'm pretty sure that it uses a taylor series, because if you try to >find the sine of very large angles, you will see they become >increasingly inaccurate. I use TI-83+ Silver Edition, and TI-30X It may be a TI thing. I remember seeing this way back in 1976. But, even back then, the HP calculators gave the right answer. Why? Because they were smart enough to reduce all angles larger than 2 pi to the range 0 to 2 pi. Not hard to do, at all. Not doing it is not very smart. -- Surendar Jeyadev jeyadev@wrc.xerox.bounceback.com Remove 'bounceback' for email address === Subject: Re: Help Needed Understanding Article > Of course? If you're familiar with the system of logic used in > the paper then why do you keep asking people to explain it > to you? Now David, be nice. You know I'm not really asking them to explain the system to me. I'm asking where this elusive formal proof of the Halting Problem is. If I asked, Where's the beef? would that make my question more clear? And it's not about explaining unsolvability either. It's about finding out whether or not Boyer and Moore did what they claim in the title of the paper. > When did I say that the paper convinced me that the halting > problem was unsolvable? in the summary that reads ëThe unsolvability theorem is proved in a constructive logic like those of Skolem[7] and Goodstein[4]'. I assumed that you wouldn't quote from a paper without believing in it as well. Earlier, to expose those BS-ers who claim that it does what its title says, I pointed out that a proof, by definition, must be convincing, and asked for what was convincing about it. This is to keep them from getting into an infinite regress of quoting other BS papers to justify previous ones. To be convincing, you have to be for real. But if you're now distancing yourself from that paper (seeing that public opinion is turning against it, perhaps), then never mind. > I do otoh have an advantage over some people: I understand > that there are things that I don't understand How do you know you don't understand it if you don't understand it? (I think this is related to Godel's 2nd Incompleteness Theorem, actually.) > and I don't > go around saying that a paper I don't understand simple > cannot contain what it claims to contain. Or that I'm the only > one who's ever done something, when there exist papers I > don't understand that claim to do the same thing. If I'd > done something, thought I was the only one who'd done > it, and was told about a paper where it had been done > previously I'd make certain to understand that paper > _before_ informing the world that the authors were lying. I agree (that I do that, not that I necessarily believe that you do.) > (No, you _don't_ understand the contents of the paper. > This is evident from the way you continually ask people > to explain it to you. You're reminding people of James > Harris, explaining why Wiles' proof of Fermat's last > theorem is wrong although he doesn't even know what > a ring is.) Don't know the guy myself. In any case, saying You don't understand does not substantiate a technical assertion! But now you realize I am playing innocent so people will be more willing to cooperate, right? But you make a good point, actually. If I say it's BS and then act like I don't understand it in order to get the nurturing types to join in, I am implying that I condemned something that I don't understand. That's a good one. I'll have to rememeber that in the future (and fine-tune my subterfuges.) Ok, I'll take you off my list of BS-ers who're defending the Boyer-Moore paper. (What's that - I made your day?) Charlie Voilkstorf Cambridge, MA > ************************ > David C. Ullrich === Subject: Re: Help Needed Understanding Article > The authors inverted the LISP complier and applied the relation > proving it is solvable to appear the solution using symbols. A key > concept is the letter or byte as the symbol or the machine state. The > authors have done nothing but confuse everybody. I am afraid that some of your terminology is foreign to me. Could you define these terms in more detail? I agree that they have done nothing - except perhaps saying some confusing things. > Now adays the existence of the LISP self compilation proves the > halting problem solved. By the halting problem solved, do you mean the unsolvability of the halting problem? And what exactly do you mean by self compilation? I don't see offhand how the existence of one program can prove the nonexistence of another program. Could you give me a simple example of how that could be .9a a simple use of that principle? The existence of a program proves that the function that it computes is recursive. The existence of a program with a particular input and output proves that the function that it computes, when applied to that input, produces that output. But how does the existence of a program per se prove that a particular function is NOT recursive? (Note that programs that prove theorems are an example of the latter: the existence of a program with a particular input and output proves that the function that it computes, typically rules of inference, when applied to its input, typically axioms, produces the output, typically a theorem. But, again, you have to explain what those axioms, rules and theorem mean for it to mean anything.) BTW These happen to be fundamental rules that are formalized by my axiomatization and used over and over in formal proofs. This is part 2 of my system. Part 1, which proves the unsolvability of the halting problem and tons of variations (which I actually produce, rather than offering excuses), has to do with the existence or non existence of certain programs. Part 2, which is really a generalization, has to do with properties of the syntax of the program itself. This proves such theorems as the Recursion Theorem, the Fixed Point Theorem, and various extensions of these, especially those developed by Raymond Smullyan. I # yes(I) Every program computes the set of values on which it halts yes. wr(I) # I Function wr returns a program that computes the input to wr. wr(I) Function wr is recursive. N # f(I) => s11(N,M) # f(M) Substitution function s11(I,J) Substitution is recursive. (eN) N # N Theorem: There is a program that outputs itself. f(I) => (eN) N # f(N) Recursion Theorem f(I) => (eN) N ~ f(N) Fixed Point Theorem It's really a great system. Simple, very powerful, easy to use and a lot of fun. It's lots of fun to write formal proofs for various theorems, especially the seemingly unlimited supply coming from Raymond Smullyan. (You also uncover mistakes in published works along the way.) It's a refreshing break from all the BS that gets published - e.g. theorems without proofs (and the idiots who defend that.) If anyone is interested in more details, please email me directly. I can always use helpers. The main task is to develop proofs of published theorems, as well as to show variations that don't seem to have been published. Occasionally that involves adding an additional axiom, which is always Oh, of course! What a perfectly reasonable axiom! Charlie Volkstorf Cambridge, MA === Subject: Re: Help Needed Understanding Article > I am convinced that there is no computable procedure for determining > if any arbitrary program written in some general purpose language can > halt. I agree completely. > I see the title and abstract of this paper. I know Boyer and Moore > have an absolutely stellar reputation in the area of systems for > formal proofs. Ok. And what specifically makes you feel that way? > There's lots of technical details in the paper, > which I have not read. So I am more than willing to accept that > their system has verified it. Would you then agree that your belief in the authenticity of their paper is more of a personal opinion rather than being based on a mathematical or logical proof? > I don't maintain that the paper must present a proof of the > unsolvability of the halting problem. Then would you agree that they do not present a proof of the unsolvability of the halting problem? Charlie Volkstorf Cambridge, MA === Subject: Re: Help Needed Understanding Article > The theorem is proven using a Theorem prover. They show the proof found by > the theorem prover for each of the theorems. The theorem prover is described > in the references, but just from this paper you can make some good guesses > about how it operates. If you are serious about understanding the paper you > would go read the references. Have you done so? > Stephen Of course, but it is just more of the same, with no proof of the Halting Problem. Showing the EVAL function that evaluates an expression does not prove that the Halting Problem is unsolvable. Nor does a description of a variation of the LISP programming language. Any proof of the unsolvability of the Halting Problem would contain some sort of logic and reasoning which reaches that conclusion. Does this paper (including its references) convince you that the Halting Problem is unsolvable? If so, could you explain the logic and reasoning that convinces you? If not, do you believe that they proved that the Halting Problem is unsolvable? Charlie Volkstorf Cambridge, MA === Subject: Re: Help Needed Understanding Article :> The theorem is proven using a Theorem prover. They show the proof found by :> the theorem prover for each of the theorems. The theorem prover is described :> in the references, but just from this paper you can make some good guesses :> about how it operates. If you are serious about understanding the paper you :> would go read the references. Have you done so? :> Stephen : Of course, but it is just more of the same, with no proof of the : Halting Problem. Showing the EVAL function that evaluates an : expression does not prove that the Halting Problem is unsolvable. Your sentence does not make sense to me. : does a description of a variation of the LISP programming language. : Any proof of the unsolvability of the Halting Problem would contain : some sort of logic and reasoning which reaches that conclusion. The final theorem proved by their theorem prover is that if H equals the result of HALT(X,X), then if H is true then X does not halt on X, and if H is false then X does halt on X. That sounds like the Halting Problem to me. Are you saying that because they use LISP it is not the Halting Problem? LISP is Turing Complete. What language do you use? : Does this paper (including its references) convince you that the : Halting Problem is unsolvable? If so, could you explain the logic and : reasoning that convinces you? If not, do you believe that they proved : that the Halting Problem is unsolvable? I believe that their theorem prover works as advertised and that what is presented in the paper is the actual output of their theorem prover. They work through all the lemmas they proved along the way, so you can backtrack from the final theorem and figure out how the theorem prover proved it. If you really want to understand the paper, you can go read the references, download the theorem prover, and try it yourself. Stephen === Subject: Re: Help Needed Understanding Article >>The theorem is proven using a Theorem prover. They show the proof found by >>the theorem prover for each of the theorems. The theorem prover is described >>in the references, but just from this paper you can make some good guesses >>about how it operates. If you are serious about understanding the paper you >>would go read the references. Have you done so? >>Stephen > Of course, but it is just more of the same, with no proof of the > Halting Problem. Showing the EVAL function that evaluates an > expression does not prove that the Halting Problem is unsolvable. Nor > does a description of a variation of the LISP programming language. Why isn't the standard proof satisfactory to you? I think, most people who are exposed to the standard proof accept the result and don't feel the need to find a formal version. In that sense, the halting problem proof (at least for me) is no different than other simple proofs, such as: (1) The non-existence of a greatest prime number. (2) The existence of the square root of 2. (3) The non-existence of a bijection between N and R ,and so on... I've never seen a formal version of any of the proofs above, but I don't doubt that they can be formalized, and I wouldn't consider their formalization to be revolutionary or anything like that. If I had doubts that a (simple) proof could be formalized, it would have to be because I was in disagreement with the informal version. In other words, I don't see any reason to require a formal proof for the sake of being convinced, unless you think you've detected a subtle inconsistency which might be exposed by a formal version. === Subject: Re: Help Needed Understanding Article > Why isn't the standard proof satisfactory to you? > I think, most people who are exposed to the standard > proof accept the result and don't feel the need to > find a formal version. > In that sense, the halting problem proof (at least for me) > is no different than other simple proofs, such as: > (1) The non-existence of a greatest prime number. > (2) The existence of the square root of 2. > (3) The non-existence of a bijection between N and R > ,and so on... > I've never seen a formal version of any of the proofs above, > but I don't doubt that they can be formalized, and I wouldn't > consider their formalization to be revolutionary or anything > like that. > If I had doubts that a (simple) proof could be formalized, > it would have to be because I was in disagreement with the > informal version. In other words, I don't see any reason to > require a formal proof for the sake of being convinced, unless > you think you've detected a subtle inconsistency which might > be exposed by a formal version. Consider a statement T and its proof. If the proof can be formalized, it means that a machine can prove T. This shows that as far as T goes, machines are not any inferior than human minds. So, the issue now is - Are there any interesting statements whose proofs cannot be formalized? -- ----------------------------------------- Remove obviously faked punctuation from burningspear79 a.t ly-cos peri-od c.om and ignore the rest of this sig bspear79@angelfire.com ----------------------------------------- === Subject: Re: Help Needed Understanding Article fuminggreen@yahoo.com schreef: >> Hello all, >> I am interested in the Theory of Computation and also the use of >> formal methods of proof. So naturally I became very excited when I >> acquired a copy of the paper, A Mechanical Proof Of The >> Unsolvability of The Halting Problem at >> http://citeseer.ist.psu.edu/boyer82mechanical.html (click on PDF in >> the upper right hand corner.) >> However, I am having a hard time understanding how they actually >> reach >> the conclusion that the Halting Problem is unsolvable. I know about >> Turing's original proof in 1937 and don't have any problem with that. >> I just don't see that the above paper proves it using their method. >> Is there anyone here who believes that this paper does prove that the >> Halting Problem is undecidable, and if so, could you explain the >> logic >> and reasoning that is used? Are there axioms and rules of inference? >> What do they state, in plain English? >> Charlie Volkstorf >> Cambridge, MA Hello Charlie, Technically you're right - this paper doesn't prove it. BUT... Even though I didn't check refs. 1 and 2, I'm pretty positive that they're referring to the famous Boyer-Moore theorem prover. So what the and to-be-proven lemmas/conjectures/theorems) they gave to this theorem prover, and (implicitly) assert that the prover was able to prove the lemmas with these definitions. One of these theorems is the halting problem's unsolvability. There are no rules in the paper because they are embedded in the Boyer-Moore theorem prover. The system has predefined logical rules that needn't be stated or (re)defined. If you study the prover you'll know what they are (and hence what kind of logic they use - it appears to be classical) The idea behind many theorem provers is the so-called propositions-as-types (or Curry-Howard) correspondence. That is, to a mathematical statement (lemma, conjecture, theorem) there will correspond a type of a type system---you know, the same kind of thing they use in programming languages, although maths needs a more general type system than any of the programming languages you're likely to know. Now in the propositions-as-types correspondence, a term (that is, a function, in programmer speak) of a certain type is just a proof of the theorem corresponding to that type. What the theorem prover thus does is finding you a term of the required type. And in fact, if you do things in the right way, a proof of a statement there exists x with property P will actually encode a +function+ that provides you with such an x. So we could make the following scheme: term : type proof : theorem function : output type Now since type checking is standard technology, at least checking whether something is a valid proof of some statement is easy: it's just typechecking the corresponding term and seeing whether it has the correct type! The problem of finding a term is more difficult, and there are automatic theorem provers (of which Boyer-Moore's seems to be a variant) and interactive ones (these are also called proof assistants - although in my opinion they don't actually +assist+ ;-) ); also the underlying logical system may differ (classical logic, constructive, ...) and sometimes even type checking is undecidable. Of course one may wonder why we should trust the programmer's skills in writing such type checkers. This is a valid question - but we may apply the so-called de Bruijn criterion: the actual type checking code must be a very small program, so that we can verify it by hand. And in fact such systems do exist; e.g. the HOL-Light system has a kernel of only some 500 lines of code (including comments etc.). I don't know the facts on the Boyer-Moore prover though. Some other systems you may want to investigate are called Coq, Oleg (formerly called Lego, until the toy makers complained :-S), Mizar, PVS, Isabelle, Twelf; and there are many more. > The authors inverted the LISP complier and applied the relation > proving it is solvable to appear the solution using symbols. A key > concept is the letter or byte as the symbol or the machine state. The > authors have done nothing but confuse everybody. Quite the contrary - I think they describe very well what they did and how they did it. > Now adays the existence of the LISP self compilation proves the > halting problem solved. Eh...? Care to expand on that? Jasper -- The problem with having an open mind is that people toss in garbage === Subject: Re: Help Needed Understanding Article > Hello Charlie, > Technically you're right - this paper doesn't prove it. And do you know what that makes me think? Well, # 1, the fact is they did not present a formal proof of the unsolvability of the Halting opposed to so many others.) So the title, A Mechanical Proof of the Unsolvability of the Halting Problem, is a misnomer? And for # 2, well, should I say it? Maybe I should just leave it at # 1. But, just between you and me, when I write a paper, I always put my formal results right up front. Why? Because that is what I look for myself. I'd rather see formal wffs, axioms, rules and proofs than pages of English (with an occasional formal (?) expression mixed in.) And besides, if they presented it, then they could honestly say that they did present a formal proof of this theorem. I mean, if you're going to write a paper and have it published, why leave out the best part? Sounds pretty fishy. So I can only assume that they probably just don't have it (gasp!) Just makes sense to me. > So what the > and to-be-proven lemmas/conjectures/theorems) they gave to this theorem > prover, and (implicitly) assert that the prover was able to prove the > lemmas with these definitions. One of these theorems is the halting > problem's unsolvability. > There are no rules in the paper because they are embedded in the > Boyer-Moore theorem prover. But what rules were used to derive the proof of the unsolvability of the Halting Problem, then? I can't tell that from reading the theorem prover's user's manual. And what did those lemmas/conjectures/theorems state? Why not list them and the rules used? Why not list the proof itself? > Some other systems you may want to investigate are called Coq, Oleg > (formerly called Lego, until the toy makers complained :-S), Mizar, > PVS, Isabelle, Twelf; and there are many more. As you can see from the links that I posted earlier, there are literally hundreds! But who has formally proved the unsolvability of the Halting Problem (besides http://www.arxiv.org/html/cs.lo/0003071)? Charlie Volkstorf Cambridge, MA > Jasper === Subject: Re: Help Needed Understanding Article : But what rules were used to derive the proof of the unsolvability of : the Halting Problem, then? I can't tell that from reading the theorem : prover's user's manual. If you would just go type the theorem into the theorem prover it would show you what rules were used. Stephen === Subject: Re: Anti-compact William Elliot a .8ecrit dans le message de > A space is anti-compact when every compact subset is finite. > Discrete spaces are anticompact, Hausdorff. > Uncountable cocountable spaces are anticompact, T1 and not Hausdorff. > Are all anticompact, Hausdorff spaces discrete? Intersting definition. Anticompact Hausdorf fspace must be discrete but we can find a connected anticompact space. a) Anticompact + Hausdorff => discrete As you point out, in an anticompact space an injective sequence cannot have any limit point. So infinite set cannot have accumulation point. In an Hausdorff space, that mean they must be closed. But in Hausdorff space finite space are closed too. So the space is discrete. b) An anticompact connected space Take N with the topology with open sets [0,n] with all n in N. It is connected because every open contain 0 so they are no disjoint open set (hyperconneted) It is anticompact because ([0,n]) is an open covering with no finite subcovering and every infinite subset is homeomorph to him. === Subject: Re: Anti-compact >William Elliot a .8ecrit dans le message de >> A space is anti-compact when every compact subset is finite. >> Discrete spaces are anticompact, Hausdorff. >> Uncountable cocountable spaces are anticompact, T1 and not Hausdorff. >> Are all anticompact, Hausdorff spaces discrete? >Intersting definition. Anticompact Hausdorf fspace must be discrete but we >can find a connected anticompact space. >a) Anticompact + Hausdorff => discrete >As you point out, in an anticompact space an injective sequence cannot have >any limit point. >So infinite set cannot have accumulation point. Careful there. An accumulation point need not be the limit of a sequence. E.g. let e0 be the first uncountable ordinal. Take the set X of ordinals <= e0, with the order topology. Then e0 is an accumulation point of X {e0}, i.e. every neighbourhood of e0 contains a member of X {e0}, but there is no sequence in X {e0} whose limit is e0. Of course this doesn't answer the question, because X is not anticompact. Robert Israel israel@math.ubc.ca Department of Mathematics http://www.math.ubc.ca/~israel University of British Columbia Vancouver, BC, Canada V6T 1Z2 === Subject: Monkey typing ABRACADABRA I have a rather simple question (so it seems): A monkey types only capital letters and is on every occasion equally like to type any of the 26 letters in the alphabet. Each second the monkey types one letter. How long on average will it take him to produce the sequence 'ABRACADABRA' ? I have heard that the solution should be: 26^11+26^4+26 seconds. However this is not very intuitive to me, i would rather say 26^11 seconds. Unfortunately I also cannot derive the above result. Any help from you would be appreciated very much! Best wishes, Peter. === Subject: Re: Monkey typing ABRACADABRA >I have a rather simple question (so it seems): >A monkey types only capital letters and is on every occasion equally >like to type any of the 26 letters in the alphabet. Each second the >monkey types one letter. How long on average will it take him to >produce the sequence 'ABRACADABRA' ? >I have heard that the solution should be: >26^11+26^4+26 seconds. You have heard right. >However this is not very intuitive to me, i would rather say 26^11 >seconds. >Unfortunately I also cannot derive the above result. No, it's good to not be able to derive wrong results. Hint: In fancier language, we're talking about the expected time to reach a certain state in a Markov chain. The state is the longest suffix of the monkey's current output that is a prefix of ABRACADABRA. For example, if the monkey has typed ABRACAXABR the state is ABR. Initially it is `` (the empty string). The situation is complicated by the fact that from a given state, as well as adding another letter or going back to ``, you might be able to go to various previous states, e.g. from ABRA to A and AB as well as ABRAC and ``. Robert Israel israel@math.ubc.ca Department of Mathematics http://www.math.ubc.ca/~israel University of British Columbia Vancouver, BC, Canada V6T 1Z2 === Subject: Re: Monkey typing ABRACADABRA >I have a rather simple question (so it seems): Not really that simple - I didn't learn this until graduate school. >A monkey types only capital letters and is on every occasion equally >like to type any of the 26 letters in the alphabet. Each second the >monkey types one letter. How long on average will it take him to >produce the sequence 'ABRACADABRA' ? >I have heard that the solution should be: >26^11+26^4+26 seconds. >However this is not very intuitive to me, i would rather say 26^11 >seconds. >Unfortunately I also cannot derive the above result. The first answer you cited is correct. One can show this with renewal theory, Markov chains, or martingales. Here's the basic renewal-theoretic argument: Since the probablity that ABRACADABRA appears in consecutive letters is 26^-11, the expected time to get from one instance thereof to the next is 26^11. (That's Blackwell's theorem from renewal theory.) However, when you start at one instance, you already have ABRA towards the next instance (so, for example, it is possible to reach the next instance in just seven more letters). Starting from scratch, you don't have these. So you have to get to ABRA first, whence the 26^4. Similarly, ABRA both ends and begins with A. So what you're computing is the time to get to A plus the time to get from (ABR)A to ABRA plus the time to get from (ABRACAD)ABRA to ABRACADABRA. The expected time to get to the pattern ABRACADABRR *is* 26^11. -- Stephen J. Herschkorn herschko@rutcor.rutgers.edu === Subject: Re: Monkey typing ABRACADABRA >>I have a rather simple question (so it seems): >Not really that simple - I didn't learn this until graduate school. Hmm, then perhaps I should fear to tread here instead of rushing in... But still... >>A monkey types only capital letters and is on every occasion equally >>like to type any of the 26 letters in the alphabet. Each second the >>monkey types one letter. How long on average will it take him to >>produce the sequence 'ABRACADABRA' ? >>I have heard that the solution should be: >>26^11+26^4+26 seconds. >>However this is not very intuitive to me, i would rather say 26^11 >>seconds. >>Unfortunately I also cannot derive the above result. >The first answer you cited is correct. One can show this with renewal >theory, Markov chains, or martingales. Here's the basic >renewal-theoretic argument: >Since the probablity that ABRACADABRA appears in consecutive letters is >26^-11, the expected time to get from one instance thereof to the next >is 26^11. (That's Blackwell's theorem from renewal theory.) However, >when you start at one instance, you already have ABRA towards the next >instance (so, for example, it is possible to reach the next instance in >just seven more letters). But then, that wouldn't *be* the next instance. It would overlap the first instance. If you want to allow overlaps, then it seems to me you should take the 2^11 figure and *reduce* it accordingly here, by putting minus signs in the o.p.'s expression. The availability of a solution with seven letters means that the monkey will get to ABRACADABRA *faster* on average than to a different 11-character string. Starting from scratch, you don't have these. >So you have to get to ABRA first, whence the 26^4. Similarly, ABRA both >ends and begins with A. So what you're computing is the time to get to >A plus the time to get from (ABR)A to ABRA plus the time to get from >(ABRACAD)ABRA to ABRACADABRA. >The expected time to get to the pattern ABRACADABRR *is* 26^11. But that would imply that it's easier to type an R than an A, contrary to hypothesis. === Subject: Re: Monkey typing ABRACADABRA >> >I have a rather simple question (so it seems): > >>Not really that simple - I didn't learn this until graduate school. >> >Hmm, then perhaps I should fear to tread here instead of rushing in... >But still... >A monkey types only capital letters and is on every occasion equally >like to type any of the 26 letters in the alphabet. Each second the >monkey types one letter. How long on average will it take him to >produce the sequence 'ABRACADABRA' ? >I have heard that the solution should be: >26^11+26^4+26 seconds. >However this is not very intuitive to me, i would rather say 26^11 >seconds. >Unfortunately I also cannot derive the above result. > >>The first answer you cited is correct. One can show this with renewal >>theory, Markov chains, or martingales. Here's the basic >>renewal-theoretic argument: >>Since the probablity that ABRACADABRA appears in consecutive letters is >>26^-11, the expected time to get from one instance thereof to the next >>is 26^11. (That's Blackwell's theorem from renewal theory.) However, >>when you start at one instance, you already have ABRA towards the next >>instance (so, for example, it is possible to reach the next instance in >>just seven more letters). >> >But then, that wouldn't *be* the next instance. It would overlap the >first instance. By instance, I mean a keystroke where the pattern ends. There are three instances in FDKABRACADABRAERNVABRACADABRACADABRABRRRVDKJABRCFJHKSF. The important thing for the theory is that the times from one instance to next are i.i.d. >If you want to allow overlaps, then it seems to me you should take the >2^11 figure and *reduce* it accordingly here, by putting minus signs >in the o.p.'s expression. The availability of a solution with seven >letters means that the monkey will get to ABRACADABRA *faster* on >average than to a different 11-character string. Your last sentence is exactly right. On average, it takes less time to get from ABRACADABRA to ABRACADABRA than to get from nothing to ABRACADABRA. > Starting from scratch, you don't have these. >>So you have to get to ABRA first, whence the 26^4. Similarly, ABRA both >>ends and begins with A. So what you're computing is the time to get to >>A plus the time to get from (ABR)A to ABRA plus the time to get from >>(ABRACAD)ABRA to ABRACADABRA. >>The expected time to get to the pattern ABRACADABRR *is* 26^11. >> >But that would imply that it's easier to type an R than an A, contrary >to hypothesis. Not at all. Simplify the discussion by switcihing the setting to flips of a fair coin. The expected time to reach the pattern TH is 4. The expected time to reach the pattern TT is 6. Why? In the first case (TH) , suppose your first flip is a tail. Great - your almost there. Then your next flip is another tail - d'oh! You missed it. But at least now, with the most recent tail, you are started on your way to TH again. Now consider the second case (TT). First you flip a tail - good. Then you flip a head, lousing things up. Now you have to start all over from scratch again. So here you gain nothing from the second flip. In fact, we see that the time until TH is the time until the first T plus the time until the first H thereafter. It is not so simple in the case of TT. With more than two outcomes (e.g., twenty-six letters), this analysis is not as simple, but the general effect is the same. In renewal-theoretic terms, the process with a pattern whose end does not overlap is an ordinary reneewal process. With the overlap, it is a delayed renewal process. (In the latter, the time until the first instance differs in distribution from the time between instances.) -- Stephen J. Herschkorn herschko@rutcor.rutgers.edu === Subject: Unofficial announcement of MathML on Mathforge Well, we're proud to announce our message boards and wikis are mathml-enabled. Type a Tex-like string, e.g. $e^(ipi)+1=0$ in your message and you will see it marked up like actual math! Go to http://www.mathforge.net right now and try it out! Additionally, you'll find the latest in math news. Soon, we'll have the capability to create marked up text online, which should be useful for homework, marking up class notes, and collaborating on documents. Joe Kauzlarich http://www.mathforge.net Tasteful Mathematics for Sensitive People === Subject: Re: Unofficial announcement of MathML on Mathforge > Well, we're proud to announce our message boards and wikis are > mathml-enabled. Type a Tex-like string, e.g. $e^(ipi)+1=0$ in your > message and you will see it marked up like actual math! I find this a somewhat baffling remark. Surely there is a vast difference between TeX and MathML? > Go to http://www.mathforge.net right now and try it out! Additionally, > you'll find the latest in math news. -- Timothy Murphy e-mail (<80k only): tim /at/ birdsnest.maths.tcd.ie tel: +353-86-2336090, +353-1-2842366 s-mail: School of Mathematics, Trinity College, Dublin 2, Ireland === Subject: Re: Unofficial announcement of MathML on Mathforge >> Well, we're proud to announce our message boards and wikis are >> mathml-enabled. Type a Tex-like string, e.g. $e^(ipi)+1=0$ in your >> message and you will see it marked up like actual math! > I find this a somewhat baffling remark. > Surely there is a vast difference between TeX and MathML? >> Go to http://www.mathforge.net right now and try it out! Additionally, >> you'll find the latest in math news. Naturally there is a distinct difference between TeX and MathML. What this equation editor does is convert the TeX-like input--with which many people are familiar--to MathML for convenient viewing in Web browsers. This is also much more accessible to the TeX-nically (hah! Get it?) unskilled without erasing the productivity inherent in a GUI-less equation editor. === Subject: Re: Yes, these are homework questions! > because it's so obvious now... can anyone recommend a book on algebra - for Notice how you are dealing mostly with linear things. Even your > polynomial problem, could be split into linear factors over an > appropriate splitting field. So first of all you need to get a book > on linear algebra. Since you are dealing with polynomials, it is > essential that you also get a book on galois theory, which is the > study of polynomials. Once you finish those, you are ready to go a > little deeper, and you should seek a book on abstract algebra. > If these books seem a little bit over your head (I'm not sure how good > you are), you might need to step back and start with I.N. Herstein's > Topics in Algebra, which is written for folks like you getting > started in math. > Good luck in all your math endeavors! > Your dear friend, > Bruce B. Ack, I can't believe nobody replied yet. What a failed troll I am. I was hoping to create a flame war as everyone jumped in to exclaim that the books above are NOT appropriate for you. Which they aren't-- I was counting on people to bite and correct me, but it seems to have failed. So don't follow the advice above, as it is quite inaccurate. === Subject: Re: Yes, these are homework questions! Converter schreef in bericht > MRe schreef in bericht I was going to ask my teacher, but ran out of time... and as for working > these out myself, well there's only so may times I can shake my fist at > them, I'm lost. > Anyone care to help me out, please? Express a in terms of b and c > a=(b-a)/c > -- Don't have a clue where to begin. Variable a represents a tablevalue. > First determine the weight of each variable. > b>c and b for b=(a*c)+a > c for c=1 b has to be as big as twice to a: > (c*a)+a=b > or coming back to a=(b-a)/c: > (c*a)=b-a > In any table c will be the multiplier to tablevalue a where b will adept by > +a upgoing the table (from c=1 b=a+a, c=2 b=a+a+a etc.). Given a factor, find k > x^3-x^2+kx+n=0 k = (x^2-x) > I don't see the point in n. > Geere > I don't understand how you came about k = (x^2-x), I got k = (-x^2+x). > (removing n from the equation because you saw no point in it) > 1. x^3-x^2+kx = 0 > 2. x(x^2-x+k) = 0 > 3. x(x^2-x+k)/x = 0/x > 4. x^2-x+k = 0 > -x^2 + x^2 > 5. k = (-x^2+x) Well done! I don't have to explain anymore, as if you know: x(x^2-x)=x(-x^2+x) Geere === Subject: Re: Four Color Theorem Proof >sounds like a novel approach; is it? >> Every vertex in every maximal planar graph is completely enclosed >> by a cycle graph; ie, a ring of edges. Essentially, every vertex is >> the hub of >> its own wheel graph. I thought I read recently that someone had found a map requireing more than four colours, but I don't recall the source. -- John W Hall Cochrane, Alberta, Canada. Helping People Prosper in the Information Age === Subject: Re: Four Color Theorem Proof > sounds like a novel approach; is it? I thought of it all by my self! I do not know if I am the first. > Every vertex in every maximal planar graph is completely enclosed > by a cycle graph; ie, a ring of edges. Essentially, every vertex is > the hub of > its own wheel graph. === Subject: big-Oh notation... Say that {X_n} a sequence of random variables satisfies X_n = O(1) if, given e > 0, there exists M_e and an integer N_e such that P { |X_n| > M_e } < e for all n >= N_e. Also, suppose f is strictly increasing on the positive reals, with f(0) = 0, lim(t --> oo) f(t) = 1. Show X_n = O(1) if and only if limsup E f(|X_n|) < 1. I can easily show that, if the big-Oh condition is satisfied, then limsup E f(|X_n|) <=1, but I don't know how to eliminate the possibility of equality holding. Any ideas? Hopefully this'll illuminate how to show the converse, which I'm entirely in the dark on.... === Subject: Re: central limit theorem, equivalent statements === >Subject: central limit theorem, equivalent statements >CLT and related topics): >Let {X_n,i} be a triangular array of random variables (independent >within each row). Suppose > (*) given epsilon > 0, sum(over i) E( X_n,i I(|X_n,i| <= epsilon) ) >---> m >(**) given epsilon > 0, sum(over i) Var(X_n,i I(|X_n,i| <= epsilon) ) >---> s^2 finite >(***) max(over i) |X_n,i| ---> 0 in probability, >where I(A) is the indicator function of the set A. >Then sum(over i) X_n,i ---> N(m,s^2) in distribution. I'm going to >call this result R1. >The problem is that I haven't been able to convince myself that this >follows from (or really is an equivalent formulation of) any of the >version of the CLT for triangular independent arrays (probably >Lindeberg's) that I'm familiar with. >I'm guessing there's no loss of generality if we set m=0, s^2=1. Then >I'm acquainted with the following: >Let {X_n,i} be a triangular array of random variables (independent >within each row). Suppose >(1) E(X_n,i) = 0 >(2) Sum(i=1,...,k_n) E((X_n,i)^2) = 1 >(3) Given epsilon > 0, Sum (i=1,...,k_n) E( (X_n,i)^2 I(|X_n,i| epsilon) ) ---> 0. >Then Z_n = Sum(i=1,...,k_n) X_n,i ---> N(0,1) in distribution. I'm >going to call this result R2. >Can someone help show me how R1 is equivalent to one of the more >popular formulations of the CLT for triangular arrays (say R2)? I've >tried doing various manipulations of the hypotheses of R1 (i.e. >flipping around the indicator functions to I(|X_n,i| > epsilon) but >I still can't put it in just the right form. please help..... This looks pretty close to the version of Lindeberg's theorem stated in Billingsley's book, Probability and Measure (3rd ed), theorem 27.4. Can someone confirm these two statements are the same?? === Subject: How to calculate the dirivative of the following vector valued function I have a vector valued function (cos(s*t),sin(s*t)) , s and t are 2 parameters, How to calculate its derivative? Is there any connection between GG === Subject: Re: How to calculate the dirivative of the following vector valued function > I have a vector valued function (cos(s*t),sin(s*t)) , s and t are 2 > parameters, > How to calculate its derivative? Is there any connection between > derivative and Jacobian Matrix? It's (total) derivative at any (s_0, t_0) is the linear transformation whose matrix is the Jacobian Matrix. === Subject: Re: Garry Donald Rabbitskin - June 20th 1972 > 12 13 14 15 16 17 18 19 20 21 <-positions > R A B B I T S K I N > 18 1 2 2 9 20 19 11 9 14 = 105 <-value > Today God brought Garry and I together, he is from some reserve near Big > River and grew up in North Battleford. He was able to sit, watch and listen > for great length while I worked on his stats, which he wouldn't have if he > had known he was listening to a convicted stalker/pedophile scumbagink. << The following (courtesy of Waxy.org) is sort of an unofficial FAQ explaining the psychotic nonsense posted to Usenet by Shawn Daryl Kabatoff AKA Dar, AKA Probababbilities. And now AKA marcia and me. WARNING: Read below before even thinking about responding to this twit. http://www.waxy.org/archive/2002/05/21/dar_kaba.shtml#000643 Usenet has the tendency to provide a public forum for those who would normally be scribbling in a closet. For example, take Daryl Shawn Kabatoff. For the last few years, he's methodically gathered statistics from various sources, ranging from local newspaper obituary pages to the food court of the Saskatoon Midtown Plaza mall. With all the raw data he's collected, he's attempting to prove daily that our full names are in mathematical harmony with our birthdays. His rants normally focus on a single individual he's met or read about, starting with calculations related to their birthdate and full names, blending in whatever other personal information about their family members, spouses, birthplace, and career he's been able to zealotry, and personal torment. I've never seen anything like it. With all the prime numbers, Fibonacci sequences and biblical references, it's like reading the notebooks of Maximillian Cohen and John Nash combined. Unsurprisingly, several posts unfold to reveal a history of painful mental illness. If you have some time, take a look. I've detailed his posting history and a several sample posts below. Usenet Posting History: January 27, 1999 to July 5, 2000 as Catsco@home.com December 9, 2000 to May 4, 2001 as s.kabatoff@sk.sympatico.ca Oct 30, 2001 to Oct 31, 2001 as kabatoff@the.link.ca January 20, 2002 to April 17, 2002 as s_kabatoff@hotmail.com (original posts have been removed from Google Groups archive) April 26, 2002 to Present as dar_kabatoff@hotmail.com Selected Posts: Tessa Lynne Smith Dastageer Sakhizai and Helen Smith Brett David Maki Andrew Meredith Cotton Kathryn Lee Hipperson Amanda Dawn Newton Mona Marie Etcheverry Tony Peter Nuspl Lisa Charlene McMillan Grant Allyn Wood Comments scarier still is that saskatoon is my hometown, though not my current residence. and every single place he's mentioned in his posts (most notably nervous harold's and the roastary) were either places i've been (as it's a small city of 200K) or hangouts, ie. the two places mentioned. chances are i could email some friends back home and find out if they know of him, they (my friends that is) being of the broadway-centred slacker ilk. myself, too, until i got out of there. eh, anyways. thought it odd to see all this. midtown mall. i ate my meals there, whilst waiting several days in line for star wars episode one, at the theatre across the street. posted by andy raad on May 22, 2002 06:20 PM Fascinating. It's like he's trying to take chaos and bind it into whatever rules he can find, religious, logical and otherwise. Numbers and math have a reliable pattern, something that can always be proven to true or false. People and religion do not. It reminds me of Darren Aronofsky's movie Pi. It's the story of an paraniod genius who is trying to find a pattern in Pi. A group that takes interest in his work is convinced that the existence of Pi, a number whose existence can be proven but no quantified, is proof of the existence of God. Kabatoff's hunt for patterns in something as random as name selection is a way to reconcile his deeply logical thought process with his conflicting religious views. Exactly. I probably shouldn't have, but I e-mailed Daryl yesterday, asking him if he'd be willing to create a numerological analysis for me. I also asked him if he had seen either Pi or A Beautiful Mind, and what he thought of them. If he replies, I'll be sure to post it. I baked many pumpkin pies for Shawn (he likes pumpkin pies). I rubbed pumpkin pie all over my breasts for him, and my breasts turned orange. I am a pumpkin for Shawn. posted by Trisha Blondie on July 24, 2002 10:41 PM Um, that's swell. So, you're in love with him? Shawn once went to a funeral for a Jehovah Witness that shot himself and the lemon tarts were very bad, they were not only sour but were rubbery as well. Shawn said that the guy was some kind of Jehovah Witness prophet, he saw in advance that the lemon tarts at his funeral were to be very very bad, and so he shot himself. Shawn said that he never ate pumpkin pie at a funeral but would like to some day. Shawn likes pumpkin pie and so I have been practicing to make very good pumpkin pies. posted by Trisha Blondie on July 25, 2002 02:49 PM Shawn said that the lemon tarts were sour, bitter and rubbery. I don't think this guy takes notes. I think he has Total Recall, and it has driven him insane... Oh... I almost forgot... I didnt spend thousands of dollars a day tormenting Daryl... We got a deal on tormenting that fiscal year, it only came to about 37cents a day.... Mr. Kabatoff attempts to portray himself as a victim, but in fact he is a violent predatory pedophile who is well known to his local law enforcement. In his post to multiple newsgroups with the subject Collecting Mail For The Coming Anti-Christ, he encourages mothers to send him photos of their naked daughters. Mr Kabatoff explains, I personally did not want photographs being mailed to (the coming Ant-Christ) that were of underage children unless the parent was signing consent. He is banned from virtually all the shopping malls in his community because he stalks young people and sexually harasses them. He has an extensive arrest record which includes sexual molestation charges. He's been hospitalized in mental institutions about his contact with young girls in many posts. Search newsgroup archives for posts by him containing the word nubile. As part of his harrassment, he provides personal details in a public forum, such as the real names of real children, in these and other posts. About one wanted her and her sister dead. He not only curses children and prays for their death in his posts, he also enjoys attending the funerals of young people: And so, since nubile sweeties are found in greatest abundance at the funerals of high school students, then it is the funerals of high school students that make the very very best funerals, especially if there is food... I stuff my face (and my pockets) with all the good food and look at all the pretty nubile sweeties and have the time of my life.. r=&ie=UTF-8&scoring=d&selm=LfXN8.63042%24R53.25142039%40twister.socal.rr. com&rnum=1 Many of his posts are sent to alt.teens.advice. However, he liberally missives to countless newsgroups. Some people HAVE problems and some folks ARE problems. Don't dismiss Mr. Kabatoff as a harmless nut. When he sends these posts to any newgroup, please help by reporting him to I knew of him when I was attending the University of Saskatchewan. He'd hang out in the Arts computer lab and all you'd see is screens of numbers racing by on his laptop. I have an original copy of his Collecting Mail for the Coming Anti-Christ pamphlet, and have seen him be hauled away by campus security on more than one occasion. My friends and I refer to him as Crazy Number Man. I've been posting to (and about) Shawn for over two years with big gaps in between. He has seen Pi and didn't like it and didn't think it resembled him at all. (Wrong, it fits him to a tee) He doesn't have total recall and has stated that he travels with a lap top to notate items. Also, he uses cut n' paste a lot if you read all the way through his ramblings. He is anti-social as shown by his angry statements towards those who, by his own admission, have been kind (but not kind enough) to him. Still, he's intelligent and seems to be able to take a joke on occassion. That's where I came in. ALOHA Reply to group (Unsolicited e-mail is deleted from the server unread if it comes from anyone not already in my addressbook. I'll never even see it) === Subject: Appropriateness of Principles of Real Analysis by Aliprantis I wanted to study graduate level Real Analysis on my own this Summer. I found this book and noticed that it had a comprehensive solutions manual to go along with it. One thing caught my attention when I was reading the synopsis on amazon.com: focus on measure theory via the semiring approach. Is this different than the normal approach to measure theory? I don't really want to learn a non-standard method and then start all over again next year. === Subject: Re: Cantor: ignorant, harmless fool or intentional liar? > Wittgenstein - now there's a level headed guy for you. > I really do not understand why anyone takes Wittgenstein seriously. Is > puzzlement. > Bob Kolker Perhaps it's because he taught Turing logic, or came up with the truth incredibly original thoughts, or taught a course called Philosophy for Mathematicians, or was just a straight up bad ass? 'cid 'ooh === Subject: Re: Cantor: ignorant, harmless fool or intentional liar? > On the other hand, Sagan in this chapter talks about the fallacy of > argument from authority, which in view of your slavish quotes from > Wittgenstein, you might want to brush up on (I'd include your quotes from > Hilbert and Sagan, but you don't seem to understand what they are saying.) You sound like you might know a thing or two. Why does anyone take what Wittgenstein says seriously? Bob Kolker === Subject: Re: Cantor: ignorant, harmless fool or intentional liar? > On the other hand, Sagan in this chapter talks about the fallacy of > argument from authority, which in view of your slavish quotes from > Wittgenstein, you might want to brush up on (I'd include your quotes from > Hilbert and Sagan, but you don't seem to understand what they are saying.) > You sound like you might know a thing or two. Why does anyone take what > Wittgenstein says seriously? Actually I take what he says seriously, the little I can understand. I commented on the slavishness of the quote, not its content. The OP also commented derisively on the mental states of Cantor and Goedel but conveniently left out Wittgenstein's own leanings in that direction. === Subject: Re: Cantor: ignorant, harmless fool or intentional liar? > Aus dem Paradies das Cantor uns geschaffen hat soll niemand uns > vertreiben k.9annen. > I would rater read this as follows: > Hopefully nobody will be able to expel us from the paradise ... In context, however, it is clear that he is expressing a determination to preserve Cantor's paradise. The comment is made, it should be noted, in the very same paper (On the infinite in English translation) where Hilbert presents his finitism, according to which the infinite is nowhere to be found in either physical or mathematical reality. This view did not prevent him from describing Cantor's set theory as the most admirable flower of the mathematical intellect and in general one of the highest achievements of purely rational human activity. === Subject: Re: Cantor: ignorant, harmless fool or intentional liar? >>Hopefully nobody will be able to expel us from the paradise ... > In context, however, it is clear that he is expressing a > determination to preserve Cantor's paradise. The comment is made, it > should be noted, in the very same paper (On the infinite in English > translation) where Hilbert presents his finitism, according to which > the infinite is nowhere to be found in either physical or mathematical > reality. This view did not prevent him from describing Cantor's set > theory as the most admirable flower of the mathematical intellect and > in general one of the highest achievements of purely rational > human activity. Yes, and he also denied the human possibility to grasp the essence of the infinitely large as well as the infinitely small. Those who are interested in the original Cantor papers (in Latin, German and French) might look into a link currently provided at de.sci.mathematik by Hermann Kremer: http://134.76.163.65/agora_docs/49439TABLE_OF_CONTENTS.html In brief (p.116-18): Cantor's paradise is a construct out of numbers that essentially deviates from continuum. Admittedly, I am not a mathematician. While my physical reasoning collided with common tenets dating back at least to Cantor, I am not aware of any case where stuff like cardinality proved immediately useful in practice. I wonder why there is so much fruitless criticism against what I consider a mathematical caprice. I would appreciate those who could tell me whether or not Robinson with his *IR took issue against the holy caprice and offered an alternative which saves Buridan's donkey. Eckard Blumschein === Subject: Re: Cantor: ignorant, harmless fool or intentional liar? > Admittedly, I am not a mathematician. As you keep insisting on demonstrating. But why do you do so in sci.math? -- Robin Chapman, www.maths.ex.ac.uk/~rjc/rjc.html Lacan, Jacques, 79, 91-92; mistakes his penis for a square root, 88-9 Francis Wheen, _How Mumbo-Jumbo Conquered the World_ === Subject: Re: Cantor: ignorant, harmless fool or intentional liar? > While my physical reasoning > collided with common tenets dating back at least to Cantor, I am not > aware of any case where stuff like cardinality proved immediately useful > in practice. It proved immediately useful in mathematical practice, for example by establishing that almost all real numbers are transcendental. === Subject: Re: Is English Abelian? Originator: dmoews@ccrwest.org (David Moews) |When I was still taking mathematics seriously, there was a conjecture |being worked on by some of the graduate students and research fellows |in their spare time. | |Consider the free group W generated by all the letters a to z (only a |to z, no accents, etc.) Now define the equivalence E using u ~ v [E] |iff u and v are English words (Standard American English spelling; |proper nouns excluded!) AND are permutations of each other. |(If you'd rather, replace Standard American English with British. But |don't mix the spellings.) | |Examples | - tea = ate = eat | - boer <> bore since Boer is in fact a proper noun | - theater <> theatre since theatre is the British spelling. | |Conjecture: The quotient group W/E is abelian |As a start: ore = roe => or = ro | |This is of course not the kind of problem that brings fame or fortune. | |Does anybody know whether or not it's been solved? As far as I know, this conjecture has not been resolved. For an earlier discussion of this problem, you can look at a thread on rec.puzzles with subject `Group Theory' around 2000-VII-5 -- David Moews dmoews@xraysgi.ims.uconn.edu === Subject: Re: infinite series overflow problem > fn=summation(an*x^n) > but I want to use more number of terms in my apllication (may be > around 500 terms) and coefficients (an) are of alternative sign, a(2n+1)=0 I assume you mean alternating signs? That is numerically unstable -- problems. One of the reasons why power series are not used to evaluate functions for large arguments. > is it possible to overcome this problem by normalizing x?? If the an are sufficiently small, sure. Just set bn = an*100^n and calculate fn = summation(bn*(x/100)^n). If they are not, you will need to rewrite your series in some other way (depends on the an) or solve the underlying problem with other methods. -- +--+ +--+| |+-|+ Christopher Creutzig (ccr@mupad.de) +--+ Tel.: 05251-60-5525 === Subject: Re: MuPAD 2.5.3 and Scilab errors, etc. by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LF0gH02906; Hello , I have recently downloaded the Mupad light v2.5.3 for windows with scilab. However I similarly get the error message when trying to Execute scilab::start(), returns Error: Unknown slot scilab::start [slot], as does scilab::avail(). I have checked scilab is there and it is in the packages sub-directory, uner sciface in the program files in the root directory. Pardon my stupidity but any suggestions on how I can sort this one out. Will. >>I'm having trouble using MuPAD's (2.5.3) Scilab integration. The package >>seems to be loaded (*maybe*), but it's decidedly not working. >There exist a problem with MuPAD and the Scilab link if there is no >ld installed on the machine. Then Scilab cannot load it communication >modules which are necessary for the link. We recognized this problem >too late. So be sure that ld is installed on your system and it is >accessable via your program search path. So try 'which ld' to see if >it is there. BTW for the next version of the link it is not >necessary to have the linker ld installed. >>Executing scilab::start(), however, returns Error: Unknown slot >>scilab::start [slot], as does scilab::avail(). >That's really strange. Are you sure the Scilab package is installed? >>On a completely different note, I have Scilab 2.7 installed on the same >>machine, and would love to be able to use it with MuPAD. Is there any way >>I can edit the included Scilab .mu files and/or UDX scripts to use it >>instead? >You can use Scilab 2.7 by setting the environment variable MUP_SCI >to the path where Scilab is installed. This is the Scilab main directory >where e.g. the Scilab bin directory can be found. MuPAD then uses this >Scilab distribuiton and it works together with Scilab 2.7 >If your problem is not solved by installing ld please contact me directly >so that we can try to solve your problem. >ToM === Subject: how jump to in derive 5 Does anyone know how to jump to an expresion number/name in Derive 5 for Wiondows as is done in Derive for Dos by simply typing a j and get to?. Else 2) does anyone know a source for Derive for Dos 3 with 373 page User Manual, for my students? 3) Is any other program quite like that Derive 3? 4) When does Dos copyright expire; is it a perpetual monoply? === Subject: Re: how jump to in derive 5 Sorry to be so vague in 3}, meant quite like its succint handling of 3 and 4 row matrices and vector list of symbolic elements. > Does anyone know how to jump to an expresion number/name in Derive 5 for > Wiondows as is done in Derive for Dos by simply typing a j and get > to?. > Else > 2) does anyone know a source for Derive for Dos 3 with 373 page User > Manual, for my students? > 3) Is any other program quite like that Derive 3? > 4) When does Dos copyright expire; is it a perpetual monoply? === Subject: Mupad 2.5.3 and Scilab ???? by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LFrwj13231; Hello , I have recently downloaded the Mupad light v2.5.3 for windows with scilab. However I get the error message when trying to Execute scilab::start(), returns Error: Unknown slot scilab::start [slot], as does scilab::avail(). I've tried call scilab, load scilab etc..... but frustratingly to no avail. I have checked scilab is there and it is in the packages sub-directory, under sciface in the program files in the root directory. Pardon my stupidity but any suggestions on how I can sort this one out. Will. === Subject: Appropriate contour for an improper integral containing a Bessel function Jo by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LGvTs25348; The following integral (Ewing and Jardetzky page 44 eq 2.86b) is causing us a lot of problems. Any help is very much appreciated. The integral is from 0 to inf over the variable k. The integrand is k*b^2*(k^2-b^2)*exp(-sqrt(k^2-a^2)*h)*Jo(k*r)/((2*k^2-b^2)^2-4*k^2*sqrt(k^2- a^2)*sqrt(k^2-b^2)) a and b are complex numbers. r and h > 0 and real. This is an odd function. It has two poles, four branch points, four hyperbolic branch cuts (fig 2.6 from the above reference). The biggest problem we have is the presence of the Bessel function. I don't know what kind of a contour would be appropriate for an improper integral containing an integrand that has a bessel function. Bessel J function blows up for imaginary arguments. I think that would be a serious problem if I use a circular contour and make the radius go to inf to make the integral vanish on this circular contour. It also causes problems when finding the conribution on the branch cuts which go to +/- inf. Yigit Yazicioglu University of Illinois at Chicago Acoustics and Vibrations Laboratory === Subject: Re: Replacement to MMA and Maple by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LIDkB04514; You are right 100%. A person who responded is the author of both MACSYMA... and MockMMA ;) Best wishes, Vladimir Bondarenko http://www.cybertester.com/ http://maple.bug-list.org/ http://www.CAS-testing.org/ ...................................................... === Subject: Re: Replacement to MMA and Maple superb new graphics, hundreds of graphical attributes allowing to > manipulate all details of a plot interactively, export to MS Word, > etc. If I want nice graphics, for example, I will use a software which > is specialized in graphics. Export to Word is a very isolated interface feature that doesn't affect the rest of the system at all. If you don't want graphics, well, don't use them. In that case, MuPAD 2.5.3 will be almost the same for you as 3.0 -- just much cheaper. > I do not understand this need for multipurpose softwares (this does I do understand the requirement of having a mathematical graphics tool that knows a whole lot about mathematics -- especially after having (been part of the group that) implemented such a tool. For starters, what's a CAS there for? Manipulating mathematical objects, usually formulae, right? And if you want to visualize what you have just done inside the CAS, it is pretty helpful if the system you use for display knows about the structures you have just been using. I certainly wouldn't start using MuPAD for non-mathematical graphics or technical drawings, I've got Metapost, Photoshop etc. for that. But for graphing mathematical topics, I haven't found a better (in my personal opinion, of course) tool yet. > Who will bring his car to a garage where the mechanician use a > Swiss-army knife as only tool? To be honest: I wouldn't care. As long as the result is good, he need not use any tools at all, as far as I am concerned. > Why is it different with sofwares? Software is different from just about anything else. I fully agree that there is really no need for do-it-all software such as an editor trying to do typesetting, animation, maybe e-Mail, web layout etc. at the same time. OTOH, there are tasks which require a common basis. (How should a system that has never seen something called besselJ plot besselJ(v, x) for v in [0, 10] and x in [-10,10]?) These need not be inside the same binary, but it is certainly reasonable to share a common code basis in these cases. MuPAD is mostly written in the MuPAD language, so in this case it makes sense to combine the two into a single product. Note that only those parts actually used will ever be loaded into memory. On the other hand, me and probably you are not standard users. It seems that most users want to have everything in one interface, which leads to CASs that are praised for their ability to produce more-or-less printer-ready papers or online presentations and I can understand a commercial company that feels this sort of praise (of competitors' software) and the resulting explicit customer demands as important and spend time and money implementing things the two of us feel to be not really part of a CAS. (Who needs yet another editor inside every programmable program?) >> I'm sorry, but the more exotic special functions have probably >> never >> been asked for in MuPAD, at least I haven't seen requests, so you >> won't find e.g. StruveH. (Although MeijerG and hypergeom are there, >> so you might be able to get some of your functions.) > I agree, the market is not big enough for such a request. That's why, There is only about a dozen people involved in developing MuPAD. Sorry, but we simply can't do everything and therefore need input from our users. Note that I am specifically not talking about customers. It is very rare that I care whether a user spent money on MuPAD or uses one of the freely available versions. (Then again, I'm not doing interface programming, where the differences lie.) > in my opinion, a high-quality CAS has to be free in order to avoid, or > at least limit, the market size effects. Write one. Or take part in improving one of the existing free systems. Or ask SciFace about a free license for the MuPAD library. -- +--+ +--+| |+-|+ Christopher Creutzig (ccr@mupad.de) +--+ Tel.: 05251-60-5525 === Subject: Re: product of Bessel functions, mean value of exp(i*periodic function) by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LIDkr04518; look on p. 148 (section 5.41) of Watson's book on Bessel Functions. Or look in the index under products. I'm interested in the same integral. If you have figured out how to do it, let me know... Ariel >Hi everybody, >Is there anyone knowing how to express the product of two Bessel >functions (ie Jm(a*x)*Jp(b*x)) with a series of Bessel functions ? >In fact the main problem to solve is to calculate the integral : > / Pi > | > | > 1 | i* Sum (over q) of Aq cos(qx) > --- | dx e > 2Pi | > | > | > _/ (-Pi) >where Aq are known coefficients > i is the complex number such as iî=-1 > cos is the function cosinus >I ve tried to search a solution but I hasn't found any >Thierry Ferrus. === Subject: Re: A proof for Goldbach's conjecture >I' ve got the proof of the Goldbach's conjecture. What a coincidence! I've got a counterexample. You can't write 389965026819938 as the sum of two primes. Take a look at your proof and ask yourself how _your proof_ proves that I'm mistaken. dave [Followups to sci.math] === Subject: MathML on Mathforge.net Well, we're proud to announce our message boards and wikis are mathml-enabled. Type a Tex-like string, e.g. $e^(ipi)+1=0$ in your message and you will see it marked up like actual math! Go to http://www.mathforge.net right now and try it out! Additionally, you'll find the latest in math news. Soon, we'll have the capability to create marked up text online, which should be useful for homework, marking up class notes, and collaborating on documents. Joe Kauzlarich http://www.mathforge.net Tasteful Mathematics for Sensitive People === Subject: Re: Trigonometric Equations by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LCvmh14157; >can someone help me with the equation: cos2(x+55)=0.4673, and please >include some workings! First- be clear! Do you mean cos of 2 times (x+ 55) (which you ought to write as cos(2(x+55))) or cos squared (x+55) (which you ought to write as cos^2(x+55)? If it is the former, it's easy: Take the inverse cosine (also called arccosine): 2(x+55)= arccos(0.4673)= 62.14 (according to my calculator) so x+55= 62.14/2= 31.07 and x= 31.07- 55= -23.93 degrees. If it is the latter, it's still easy: First take the square root of each side: cos(x+55)= sqrt(0.4673)= 0.6836. NOW find the inverse cosine: x+ 55= arccos(0.6836)= 46.87 so x= 46.86- 55= -8.13 degrees. By the way: since there is no mention of angles and you are treating cosine as a function, normally I would assume that I should have my calculator in radian mode. However, that 55 made me think it should be in degrees. You might want to check on that. === Subject: Re: Heating efficiency by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LD8CF16648; >Our textbook (Washington, /Basic Technical Mathematics/, 7th ed) >sets the following problem (sec 21.2, prob 47), and I 've assigned >it to my students. But I don't think there's actually enough >information to solve it: >One heating unit uses x gallons of fuel with 72% efficiency, and >another uses y gallons at 90% efficiency. If 135,000 Btu of heat is >delivered by these units together, express y as a function of x. >The problem wants students to begin with > .72x + .90y = 135,000 >and then solve for y. But I question whether that's correct. >Shouldn't there be some sort of conversion factor from gallons to >Btu? Surely at 100% efficiency, 1 gallon of fuel delivers rather >more than one Btu! >-- >Stan Brown, Oak Road Systems, Cortland County, New York, USA >
http://OakRoadSystems.comnecessary. -- IRS Form 1040 line 23 instructions Unless efficiency is taking that into account. It is quite possible that saying that a fuel is 90% efficient is intended to MEAN that one-gallon 90% of that fuel produces 90% of the BTUs one-gallon of 100% efficient fuel would. You don't say at what level this course is, but I would think that unless this is specifically a course for heating engineers, that kind of simplification would be reasonable. Of course, it would a good idea to SAY that in the problem! === Subject: Re: Heating efficiency > Unless efficiency is taking that into account. It is quite >possible that saying that a fuel is 90% efficient is intended to >MEAN that one-gallon 90% of that fuel produces 90% of the BTUs >one-gallon of 100% efficient fuel would. I think your explanation makes sense. (The 135,000 applies only to kerosene, not other type of heating fuel.) >Of course, it would a good idea to SAY that in the problem! Indeed yes! > You don't say at what level >this course is, but I would think that unless this is specifically a >course for heating engineers, that kind of simplification would be >reasonable. It's second-semester technical math at a two-year college. Students are aiming at careers in electrical technology or construction technology. Instead of making it homework, I'll use it as a classroom example, where I can make the additional explanations. With the extra info, I think it's a decent example. -- Stan Brown, Oak Road Systems, Cortland County, New York, USA http://OakRoadSystems.com An expense does not have to be required to be considered necessary. -- IRS Form 1040 line 23 instructions === Subject: Re: Probability issue; please help! >probability >Suppose that 52% of all register voters prefer George Bush to Al Gore. >a. In a random sample of 100 voters, what is the probability that the >sample will indicate that Bush will win the elections (that is, there >will be more votes in the sample for Bush?) .6184 >b. In a random sample of 100 voters, what is the probability that the >sample will indicate that Gore will win the election? .3082 >c. In a random sample of 100 voters, what is the probability that the >sample will indicate a dead heat? (50-50) .0735 >d. In a random sample of 100 voters, what is the probability that >between 40 and 60 (inclusive) voters will prefer Bush? .9500 There -- now your homework's done without any need for thought or work on your part. -- Stan Brown, Oak Road Systems, Cortland County, New York, USA http://OakRoadSystems.com An expense does not have to be required to be considered necessary. -- IRS Form 1040 line 23 instructions === Subject: Re: DeMoivre's Theorem >tan t = -2(sqr 3)/2 = -(sqr 3); arc tan t = -pi/3 already is in Quad IV. >Why the shift to another branch of arctan? So that the principal root will have a positive angle. -- Stan Brown, Oak Road Systems, Cortland County, New York, USA http://OakRoadSystems.com An expense does not have to be required to be considered necessary. -- IRS Form 1040 line 23 instructions === Subject: Formula please by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LGEnH18900; opening ammount = O number of years = X closing amount = C average annual % return on investment is ? Perry === Subject: Re: Formula please > opening ammount = O > number of years = X > closing amount = C > average annual % return on investment is ? Try to work on it. Assume the annual return on investment is r. Then after one year the amount you have is O + rO = O(1 + r) After two years the amount you have is (amount after one year) + r(amount after one year) = (amount after one year)(1 + r) = [O(1 + r)](1 + r) = O(1 + r)(1 + r) = O(1 + r)^2 You should be able to carry that forward to establish the relationship beteen O, X, C, and r, and then solve the resulting equation for r. -- Rich Carreiro rlcarr@animato.arlington.ma.us === Subject: Re: Formula please >>opening ammount = O >>number of years = X >>closing amount = C >>average annual % return on investment is ? > Try to work on it. Assume the annual return > on investment is r. Then after one year > the amount you have is > O + rO = O(1 + r) > After two years the amount you have is > (amount after one year) + r(amount after one year) > = (amount after one year)(1 + r) > = [O(1 + r)](1 + r) > = O(1 + r)(1 + r) > = O(1 + r)^2 > You should be able to carry that forward to establish > the relationship beteen O, X, C, and r, and then > solve the resulting equation for r. That depends, is the annual percentage supposed to be compound or simple interest? -- Will Twentyman email: wtwentyman at copper dot net === Subject: Re: Formula please >>opening ammount = O >>number of years = X >>closing amount = C >>average annual % return on investment is ? > Try to work on it. Assume the annual return > on investment is r. Then after one year > the amount you have is > O + rO = O(1 + r) After two years the amount you have is > (amount after one year) + r(amount after one year) > = (amount after one year)(1 + r) > = [O(1 + r)](1 + r) > = O(1 + r)(1 + r) > = O(1 + r)^2 You should be able to carry that forward to establish > the relationship beteen O, X, C, and r, and then > solve the resulting equation for r. That depends, is the annual percentage supposed to be compound or simple > interest? The original question, asking for the annualized rate of return, is usually answered in terms of the compound interest rate, though for small enough rates and short enough periods, simple and compound rates do not differ enough to quibble over. === Subject: Re: Formula please >>opening ammount = O >>number of years = X >>closing amount = C >>average annual % return on investment is ? >Try to work on it. Assume the annual return >on investment is r. Then after one year >the amount you have is > O + rO = O(1 + r) >After two years the amount you have is > (amount after one year) + r(amount after one year) > = (amount after one year)(1 + r) > = [O(1 + r)](1 + r) > = O(1 + r)(1 + r) > = O(1 + r)^2 >You should be able to carry that forward to establish >the relationship beteen O, X, C, and r, and then >solve the resulting equation for r. >>That depends, is the annual percentage supposed to be compound or simple >>interest? > The original question, asking for the annualized rate of return, is > usually answered in terms of the compound interest rate, though for > small enough rates and short enough periods, simple and compound rates > do not differ enough to quibble over. True. Knowing the context of the problem would make the approach more obvious. -- Will Twentyman email: wtwentyman at copper dot net === Subject: probability by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LGvTb25352; The grades on the midterm examination given in a large managerial statistics class are normally distributed with mean 75 and standard deviation 9. The instructor of this class wants to assign an A grade to the top 10% of the scores, a B grade to the next 10% of the scores, a C grade to the next 10% of the scores, a D grade to the next 10% of the scores, and an F grade to all the scores below the 60th percentile of this distribution. For each possible letter grade, find the lowest acceptable score within the established range. For example, the lowest acceptable score for an A is the score at the 90th percentile of this normal distribution === Subject: Re: probability >The grades on the midterm examination given in a large managerial >statistics class are normally distributed with mean 75 and standard >deviation 9. The instructor of this class wants to assign an A grade >to the top 10% of the scores, a B grade to the next 10% of the scores, >a C grade to the next 10% of the scores, a D grade to the next 10% of >the scores, and an F grade to all the scores below the 60th percentile >of this distribution. For each possible letter grade, find the lowest >acceptable score within the established range. For example, the >lowest acceptable score for an A is the score at the 90th percentile >of this normal distribution Please post what you have done to solve this problem on your own. -- Stan Brown, Oak Road Systems, Cortland County, New York, USA http://OakRoadSystems.com An expense does not have to be required to be considered necessary. -- IRS Form 1040 line 23 instructions === Subject: Re: Finding the vertex of a parabola based on the foci and a point >> The mounting point is a point on the hyperbolic mirror itself, so its >> coordinates (20,20) can be used in the formula for the hyperbola itself. >> The problem I had was that, after putting everything into the general for >> of a hyperbola, I had a 4th power equation that just seemed to not be >> right: >> >> 400/a^2 - 400/(400-a^2)^2 = 1 ; basic setup, with (400-a^2) being b > I think this may be your problem. If the hyperbola has equation x^2/a^2 - y^2/b^2 = 1, and the focus is at (c,0) then a^2 + b^2 = c^2, > so you should have derived 400/a^2 - 400/(400-a^2)= 1 > That would actually be 400/a^2 - 400/(400-a^2)^2 = 1. So that b^2 = (c^2 - a^2)^2? === Subject: Fourier Graphing by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LL2mk28228; Try this site. http://home.messiah.edu/~barrett/c3_fourier.html === Subject: probability/statistics ouch by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3LMY1203379; need yesterday. This is a data analysis class utilizing excel spreadsheet and statpro. I'd be satisfied with an answer in excel spreadsheet format, tables, explanation, the whole nine yards Suppose that a popular hotel for vacationers, in Orlando, Florida has a total of 300 identical rooms. Like many major airline companies, this hotel has adopted an overbooking policy in an effort to maximize the usage of its available lodging capacity. Assume that each potential hotel customer holding a room reservation, independently of the hotel on a given night with the probability of 0.15. a. Find the largest number of room reservations that this hotel can book and still be at least 95% sure that everyone who shows up at the hotel will have a room on any given night b. Given that the hotel books the number of reservations found in answering part a, find the probability that at least 90% of the available rooms will be occupied on a given night. c. Given that the hotel books the number of reservations found in answering part a, find the probability that at most 90% of the available rooms will be occupied on a given night. d. How does your answer to part a change as the required assurance rate increases from 95% to 97%? How does you answer to part a change as the required assurance rate increases from 95% to 99%. varies between 5% and 25% (in increments of 5%) Assume now that the required assurance rate is held fixed at 95%. === Subject: Re: A proof for Goldbach's conjecture >I' ve got the proof of the Goldbach's conjecture. > What a coincidence! I've got a counterexample. > You can't write 389965026819938 as the sum of two primes. > Take a look at your proof and ask yourself how _your proof_ > proves that I'm mistaken. > dave > [Followups to sci.math] 5569 + 389965026814369. Brute Force. Do I win the Donald Trump price? === Subject: CombinatoricaGraphics packages for MMA. and other q's I posted this in steve's forum, but i wanted to post in this group also since many of you are mathematicians and experienced with mma. package. below is a replica of email sent to mathforum. i hoe you guys don't ind me cross posting with same email. thank you all for helpful comments. sean ----------- Hello group and david. I love david's Package for improving Combinatorica graphics. and obviously by the looks of this email, and the fact that I'm writing, yes.. I have a few problems. please consider the following digraph. In[12]:= lg = Graph[{ {{1, 2}}, {{2, 3}}, {{3, 2}}, {{3, 1}}, {{1, 4}}, {{1, 4}}, {{1, 5}}, {{1, 5}}, {{4, 5}}, {{4, 2}}, {{4, 2}}, {{5, 2}}, {{5, 2}}, {{5, 4}}, {{5, 4}}}, {{{0.0, 1.5}, VertexLabel -> a1}, {{0.0, 0.0}, VertexLabel -> b2}, {{0.5, 0.75}, VertexLabel -> c3}, {{-0.5, 0.75}, VertexLabel -> d4}, {{-1.5, 0.75}, VertexLabel -> e5}}, EdgeDirection-> On] ShowGraph[lg] CombinatoricaPlot[ {DrawGraphEdges[lg][All], DrawGraphVertices[lg][All]}, Background -> White]; above will produce two graphics. onedefault and one produced by david's new package. 1. How come the CombinatoricaPlot isn't drawing the directed edges using DrawGraphBowArrow as a default? 2. As far as I understand it, a tour is a path in a digraph where you only cross every vertex once, before returning to the starting pt( in this case, a1) How do I show all tours that are possible withthe digraph above? 3. how do I show that all the tours are isomorphic? or is that an obvious question? sean === Subject: Re: how jump to in derive 5 You might try asking on the yahoo e-Derive Users Group: http://groups.yahoo.com/group/eDUG/ > Sorry to be so vague in 3}, meant quite like its succint handling of 3 and 4 > row matrices and vector list of symbolic elements. > Does anyone know how to jump to an expresion number/name in Derive 5 > for > Wiondows as is done in Derive for Dos by simply typing a j and get > to?. > Else > 2) does anyone know a source for Derive for Dos 3 with 373 page User > Manual, for my students? > 3) Is any other program quite like that Derive 3? > 4) When does Dos copyright expire; is it a perpetual monoply? === Subject: Re: Replacement to MMA and Maple First, a precision: I have nothing against MuPaD in particular, and I didn't want to start a war. Now, let me precise my point. Yes, it is very nice to plot easily within a mathematical software. It is very convenient and helps a lot for debugging. I think that CAS should be able to do graphics. But not pretty graphics. I am using MATLAB and plotting every days. I am very happy with its basic features. On the other, each time I try to use MATLAB for doing pretty graphics, i.e. high quality for publication, it is a huge struggle. I am spending a LOT of time trying to do clean graphics, the way I want them to be. Moreover, what I see is NEVER what I get. MATLAB's basic graphic features are nice, advanced ones suck. It's the same with Maple, and probably with MuPaD too. I repeat: I do not understand this effort for advanced graphic capabilities. For other features, I think it is important that programs can exchange datas, via some normalized protocol. I don't think it is necessary for a program to do everything. Using the right tool for the right job is important. That's what professionals in any field do, except in informatics, it seems. D. === Subject: Re: Laplace transformation software? by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3MDxE722906; It seems that Matlab herits the symbolic mathematics modules from Maple. I wonder whether there exists any big difference (for example, the intrinsic function library for Laplace transformation) between Matlab and Maple especially in symbolic mathematics. Why do I ask the question? Because the problem I will solve is difficult and also involves special functions. If you want to feel the problem, I think the paper in AICHE JOURNAL,1981,27(6):1032 could be an axample. Therefore, the modules of the 'laplace/inverse laplace' transformation are what I most want to know. After all, the inverse laplace transformation in which the special functions are involved are curbsome. However, I found that the Maxima is far inadequate in laplace transformation. Looking forward to your comments or suggestions. === Subject: Re: Laplace transformation software? > It seems that Matlab herits the symbolic mathematics modules from > Maple. I wonder whether there exists any big difference (for example, > the intrinsic function library for Laplace transformation) between > Matlab and Maple especially in symbolic mathematics. I'm not 100% sure, but according to http://www.mathworks.com/access/helpdesk/help/toolbox/symbolic/symbolic.html Maple's inttrans package (for integral transforms such as laplace) isn't available for MATLAB. Maybe someone else can confirm. > Why do I ask the question? Because the problem I will solve is > difficult and also involves special functions. If you want to feel the > problem, I think the paper in AICHE JOURNAL,1981,27(6):1032 could be > an axample. Therefore, the modules of the 'laplace/inverse laplace' > transformation are what I most want to know. After all, the inverse > laplace transformation in which the special functions are involved are > curbsome. However, I found that the Maxima is far inadequate in > laplace transformation. Maple's help on Laplace transform says (partial quoting): Expressions involving a wide variety of functions including exponentials, trigonometrics, Bessel functions, error functions, and many others can be transformed. Both laplace and invlaplace recognize the Dirac-delta (or unit-impulse) function as Dirac(t) and Heaviside's unit step function as Heaviside(t). Users can add their own functions to laplace's internal lookup table by using the addtable function. So for more exotic functions you're probably out of luck... -- Thomas Richard Maple Support Scientific Computers GmbH http://www.scientific.de === Subject: Re: Laplace transformation software? > I'm not 100% sure, but according to > http://www.mathworks.com/access/helpdesk/help/toolbox/symbolic/symbolic.html > Maple's inttrans package (for integral transforms such as laplace) isn't > available for MATLAB. Maybe someone else can confirm. Sorry, I was wrong here. The Extended Symbolic Math Toolbox is not necessary for this task. The Symbolic Math Toolbox contains fourier and ifourier. You can find examples, starting from the URL above. -- Thomas Richard Maple Support Scientific Computers GmbH http://www.scientific.de === Subject: Re: Computer Algebra System ... > I can only speak for MuPAD, where a power is represented like this: > prog::exprtree(a^2) > _power > | > +-- a > | > `-- 2 >Polynomials use a denser representation, where a reference to some >arbitrary object (the coefficient) is followed by the exponent vector >of the current term, stored in C integers. But general expressions >are not stored as polynomials. The denser vector of polynomials is perhaps minor. > What do you mean by the exponent of a number? Integers are not >stored as floats. An exponent can put into all nodes. It would be an integer. It seems like a good idea since most of the algorithms can't cope with an exponent and shifting it into a node createss simple special cases. ... >> * How is information about the identity of the CPU stored?. > Why should it be stored at all? There would be at least a stack for each CPU. ... >> * Are there any pointers in expressions except to names and >> subtrees?. > Certainly in an OO system such as MuPAD each object has a pointer to >its domain type. (Kernel types do not have a pointer but rather a >small number, but that's an implementation detail.) Oh, and you did >not include numbers, functions etc. in the list, which are of course >pointed to as well. In Ada 95 a record can be extended without changing its type and by adding adjacent fields (with no pointers). ... >> * What is the interpretation of out-of-bounds lengths of the array?. > I'm sorry? Which array are you talking about? Have you got the source code. A URL would be nice. ... >> I request that you make availanle the source code that data structure > You *request*? I certainly hope you did not really mean to use this >word. If you did, please explain on which basis you think to have >the right to request such a thing. At this moment I do not have the file/document that shows exactly the design of the data structure. The request has failed. I request the reasoning for not getting source file released. You said you were an employee so the request is one that could fail. ... > It's not really hard to guess, no. Just invoke ?iszero and read the >documentation. The topic was made by me to be on the 3GL internal design of Mupad. I.e. the exact design and not summaries or clarifications of source code that is not in the public domain. You kept writing to sci.math.symolic and advocating use of Mupad. >> The webpage of the 2nd provided Mupad URL does not say that nodes >> of expressions actually even contain a field saying what the mass >> is. Is 3x implemented with 3 records or one, in Mupad ?. > I don't understand what you mean by mass. 3*x is stored as an >expression having the operands _mult (the operator symbol, available >as op(3*x, 0), x, and 3. An example of a problem with storing the 3 and the x in 3x, in different nodes/records: simplify (3*x - 5*y <= -y), versus: simplify (x*3 - 5*y <= -y) Having more atoms/nodes increases the number of cases. There is no need to have a separate number but instead there is a dimensionless x that has no name. >> The data has been online in the past ?. > Again, I do not understand what you mean. Please clarify. A yes or no would be a preferred form for the response. I want to get to the information speedily. Perhaps Mupad is not selling well enough. === Subject: This week in the mathematics arXiv (29 Mar - 2 Apr) Originator: bergv@math.uiuc.edu (Maarten Bergvelt) Here are this week's titles in the mathematics arXiv, available at: http://front.math.ucdavis.edu/ http://front.math.ucdavis.edu/submissions This week in the mathematics arXiv may be freely redistributed with attribution and without modification. Titles in the mathematics arXiv (29 Mar - 2 Apr) ------------------------------------------------ AC: Commutative Algebra ----------------------- math.AC/0403535 Xinxian Zheng: Monomial ideals arising from distributive lattices math.AC/0403534 Juergen Herzog, Takayuki Hibi: Level rings arising from meet-distributive meet-semilattices math.AC/0403531 Susumu Oda: On unramified finitely generaed extensions of polynomial rings over a field AG: Algebraic Geometry ---------------------- math.AG/0404012 Edoardo Ballico, Elizabeth Gasparim: Bundles near a negative curve hep-th/0403272 Ron Donagi, Alon E. Faraggi: On the Number of Chiral Generations in Z2 X Z2 Orbifolds math.AG/0403553 Arsen Elkin: Hyperelliptic Jacobians with real multiplication math.AG/0403550 Chad Schoen: Specialization of the torsion subgroup of the Chow group math.AG/0403543 E. Artal, J. Carmona, J.I. Cogolludo, M.A. Marco: Invariants of Combinatorial Line Arrangements and Rybnikov's Example math.AG/0403530 Mike Develin, Jeremy L. Martin, Victor Reiner: Classification of Ding's Schubert varieties: finer rook equivalence math.AG/0403526 Henning Krause: The stable derived category of a noetherian scheme math.AG/0403525 Razvan Veliche: Genus 2 mapping class groups are not Kahler math.AG/0403519 Kieran G. O'Grady: Involutions and linear systems on holomorphic symplectic manifolds math.AG/0403516 Christopher Deninger, Annette Werner: Vector bundles on p-adic curves and parallel transport math.AG/0403478 Igor Dolgachev, JongHae Keum: Finite groups of symplectic automorphisms of K3 surfaces in positive characteristic math.AG/0403471 Ivan Dimitrov, Ivan Penkov: Ind--varieties of generalized flags as homogeneous spaces for classical ind--groups math.AG/0403464 Ivan Petrakiev: Specializing Multiple Points to Elliptic Curves math.AG/0403460 H. Hakopian: The Multivariate Fundamental Theorem of Algebra and Algebraic Geometry math.AG/0403459 H. Hakopian, M. Tonoyan: On an Algebraic System of Equations math.AG/0403449 D. Arcara: A lower bound for the dimension of the base locus of the generalized theta divisor AP: Analysis of PDEs -------------------- math.AP/0404011 Damiano Foschi: Maximizers for the Strichartz inequality math.AP/0404006 Hiroyuki Chihara: The initial value problem for a third order dispersive equation on the two-dimensional torus math.AP/0404005 Hiroyuki Chihara: Third order semilinear dispersive equations related to deep water waves math.AP/0403549 Benjin Xuan: The solvability of Brezis-Nirenberg type problems of singular quasilinear elliptic equation math.AP/0403507 Gregory Eskin, James Ralston: Inverse boundary value problems for systems of partial differential equations math.AP/0403469 C M Doria: Boundary Value Problems for the $2^{nd}$-order Seiberg-Witten Equations math.AP/0403468 Alexandru Tamasan: On the scattering for the $bar{partial}$- equation and reconstruction of convection terms AT: Algebraic Topology ---------------------- math.AT/0404003 Ezra Getzler: Lie theory for nilpotent L-infinity algebras math.AT/0403540 Martin Jakob: A note on the Thom isomorphism in geometric (co)homology math.AT/0403486 Ralph L. Cohen: Multiplicative properties of Atiyah duality math.AT/0403451 J. Daniel Christensen, Daniel C. Isaksen: Duality and Pro-spectra CA: Classical Analysis and ODEs ------------------------------- math.CA/0404013 Allan Pinkus: Strictly Hermitian Positive Definite Functions math.CA/0403533 Jonathan Coussement, Walter Van Assche: Gauusian quadrature for multiple orthogonal polynomials math.CA/0403532 Kelly Postelmans, Walter Van Assche: Multiple little q-Jacobi polynomials math.CA/0403520 Jean Ludwig, Detlef Muller, Sofiane Souaifi: Holomorphic L^p-type for sub-Laplacians on connected Lie groups math.CA/0403510 Raimundas Vidunas: Expressions for values of the gamma function math.CA/0403482 Alejandro S. Jakubi: The branching problem in generalized power solutions to differential equations math.CA/0403481 George Gasper, Michael Schlosser: Some curious q-series expansions and beta integral evaluations math.CA/0403476 Detlef Mueller, Christoph Thiele: Wave equation and multiplier estimates on ax+b groups math.CA/0403472 S. Denisov, S. Kupin: The Szego class with a polynomial weight CO: Combinatorics ----------------- math.CO/0403551 R.M. Green, J. Losonczy: Schubert varieties and free braidedness math.CO/0403546 Herbert E. Scarf, Kevin M. Woods: Neighborhood complexes and generating functions for affine semigroups math.CO/0403541 Francois Bergeron, Riccardo Biagioli, Mercedes H. Rosas: Inequalities between Littlewood-Richardson Coefficients math.CO/0403513 Frederic Bosio: Suppressing nonrevisiting paths math.CO/0403508 Elchanan Mossel: Distorted metrics on trees and phylogenetic forests math.CO/0403505 Bilal Khan, Kiran R. Bhutani, Delaram Kahrobaei: A graphic generalization of arithmetic math.CO/0403502 Miklos Bona: The limit of a Stanley-Wilf sequence is not always an integer, and layered patterns beat monotone patterns math.CO/0403494 Michael Joswig, Frank H. Lutz: One-Point Suspensions and Wreath Products of Polytopes and Spheres math.CO/0403450 Vladimir Nikiforov: The asymptotics of strongly regular graphs CV: Complex Variables --------------------- math.CV/0403539 Joel Merker: On envelopes of holomorphy of domains covered by Levi-flat hats and the reflection principle DG: Differential Geometry ------------------------- math.DG/0404004 A. Rod Gover: Conformal de Rham Hodge theory and operators generalising the Q-curvature math.DG/0403555 Andre Diatta: Left Invariant Contact Structures on Lie Groups math.DG/0403547 Chongying Dong, Kefeng Liu, Xiaonan Ma, Jian Zhou: $K$-theory associated to vertex operator algebras math.DG/0403545 Colin Guillarmou: Resonances and scattering poles on asymptotically hyperbolic manifolds math.DG/0403544 Ronaldo Garcia, Romildo Pina: Ricci Tensors with Rotational Symmetry on R^n math.DG/0403528 Nobuhiro Honda: Self-dual metrics and twenty-eight bitangents math.DG/0403500 Ronaldo Garcia, Jorge Sotomayor: Lines of Mean Curvature on Surfaces Immersed in R3 math.DG/0403485 Claus Gerhardt: The inverse mean curvature flow in ARW spaces--transition from big crunch to big bang math.DG/0403483 David L. Johnson, Penelope Smith: Partial regularity of mass-minimizing Cartesian currents math.DG/0403479 Bogdan Alexandrov: On weak holonomy math.DG/0403473 Maxim Braverman, Valentin Silantyev: Kirwan-Novikov inequalities on a manifold with boundary math.DG/0403465 Daniel Azagra, Juan Ferrera: Proximal calculus on Riemannian manifolds, with applications to fixed point theory math.DG/0403452 S.P. Novikov: On the Metric Independent Exotic Homology DS: Dynamical Systems --------------------- math.DS/0403552 Alberto Abbondandolo, Pietro Majer: When the Morse index is infinite math.DS/0403537 Jose F. Alves, Armando Castro, Vilton Pinheiro: Backward volume contraction for endomorphisms with eventual volume expansion math.DS/0403536 Jose F. Alves, Krerley Oliveira, Ali Tahzibi: On the continuity of the SRB entropy for endomorphisms math.DS/0403523 the Topology of Solenoidal Attractors of the Cylinder math.DS/0403518 Stefano Marmi, Pierre Moussa, Jean-Christophe Yoccoz: The cohomological equation for Roth type interval exchange maps math.DS/0403501 Tien-Cuong Dinh, Christophe Dupont: Dimension de la mesure d'equilibre d'applications meromorphes math.DS/0403454 Nikos Frantzikinakis, Bryna Kra: Polynomial Averages Converge to the Product of Integrals math.DS/0403453 Nikos Frantzikinakis: The Structure of Strongly Stationary Systems FA: Functional Analysis ----------------------- math.FA/0404014 Vassiliki Farmaki: Ramsey and Nash-Williams combinatorics via Schreier families math.FA/0403490 Lev Sakhnovich: Integrable Operators and Canonical Differential Systems math.FA/0403474 Cleon S. Barroso, Eduardo V. Teixeira: A Fixed Point Theorem for Sum of Operators and Applications GM: General Mathematics ----------------------- math.GM/0403521 Sergei Fedotov, Stephanos Panayides: Stochastic arbitrage return and its implications for option pricing GN: General Topology -------------------- math.GN/0403495 Mathieu Baillif: The homotopy classes of continuous maps between some non-metrizable manifolds GR: Group Theory ---------------- math.GR/0403514 Robert Shwartz: Ordered Generating Systems of Mathieu Groups math.GR/0403480 Robert L. Griess Jr: Pieces of 2^d: Existence and uniqueness for Barnes-Wall and Ypsilanti lattices math.GR/0403456 Sarah J. Campbell, Graham A. Niblo: Hilbert space compression and exactness for discrete groups GT: Geometric Topology ---------------------- math.GT/0404007 Mary Rees: The geometric model and coarse Lipschitz equivalence direct from Teichmuller geodesics math.GT/0403554 Frank Quinn: Cores of s-cobordisms of 4-manifolds math.GT/0403489 Joan S. Birman: Braids, knots and contact structures math.GT/0403475 Jozef H.Przytycki, Witold Rosicki: The topological interpretation of the core group of a surface in S^4 math.GT/0403470 J'er^ome Dubois: Non abelian Reidemeister torsion and volume form on the SU-representation space of knot groups math.GT/0403455 Kevin P. Knudson: On the kernel of the Gassner representation KT: K-Theory and Homology ------------------------- math.KT/0403511 Uuye Otgonbayar: The Baum-Connes Conjecture for KK-theory MG: Metric Geometry ------------------- math.MG/0403503 Dragutin Svrtan, Darko Veljan, Vladimir Volenec: Geometry of pentagons: from Gauss to Robbins MP: Mathematical Physics ------------------------ math-ph/0404005 A.Zabrodin: Whitham hierarchy in growth problems math-ph/0404004 R. Cartas-Fuentevilla: The Euler characteristic and the first Chern number in the covariant phase space formulation of string theory math-ph/0404003 S. Dobrokhotov, K. Pankrashkin, E. Semenov: On Maslov Conjecture about Square Root Type Singular Solutions of the Shallow Water Equations math-ph/0404002 Alessandra Bianchi, Pierluigi Contucci, Andreas Knauf: Stochastically Stable Quenched Measures math-ph/0404001 R. Cartas-Fuentevilla: Topological terms and the global symplectic geometry of the phase space in string theory quant-ph/0403210 Randall Espinoza, Tom Imbo, Paul Lopata: Translation of Quantum Texts quant-ph/0403208 Randall Espinoza, Tom Imbo, Paul Lopata: Enscription of Quantum Texts math-ph/0403063 Jeffrey H Schenker: Holder equicontinuity of the integrated density of states at weak disorder math-ph/0403062 Nicolae Cotfas: On the self-similarities of the Penrose tiling hep-th/0403118 Kurusch Ebrahimi-Fard, Li Guo, Dirk Kreimer: Integrable Renormalization II: the general case cond-mat/0403644 Ernesto P. Borges, Constantino Tsallis, Jose G. V. Miranda, Roberto F. S. Andrade: Mother wavelet functions generalized through q-exponentials math-ph/0403061 Oliver Maspfuhl: Gauge fields and Sternberg-Weinstein Approximation of Poisson Manifolds math-ph/0403060 Alexandre Fedotov, Frederic Klopp: Operateurs de Schrodinger quasi-periodiques adiabatiques : Interactions entre les bandes spectrales d'un operateur periodique math-ph/0403059 R.Parthasarathy, R.Sridhar: A q-Generalization of Product Densities and Janossy Functions in Stochastic Point Processes gr-qc/0403068 Sergei Kopeikin, Igor Vlasov: Parameterized Post-Newtonian Theory of Reference Frames, Multipolar Expansions and Equations of Motion in the N-body Problem quant-ph/0403179 Gavriel Segre: Noncommutative Bayesian Statistical Inference from a wedge of a Bifurcate Killing Horizon math-ph/0403058 Miloslav Znojil: An asymptotic intertwining of the undelayed and delayed Fibonacci numbers math-ph/0403057 Michel Planat, Metod Saniga, Haret Rosu: Mutually Unbiased Bases and Finite Projective Planes math-ph/0403056 J. Golenia, A.K. Prykarpatsky, Y.A. Prykarpatsky: The structure of Gelfand-Levitan-Marchenko type equations for Delsarte transmutation operators of linear multi-dimensional differential operators and operator pencils. Part 1 math-ph/0403055 Y.A. Prykarpatsky, A.M. Samoilenko, A.K. Prykarpatsky, V.Hr. Samoylenko: The Delsarte-Darboux type binary transformations and their differential-geometric and operator structure. Part 1 math-ph/0403054 Yarema Prykarpatsky, Anatoliy Samoilenko, Anatoliy K. Prykarpatsky: Differential-geometric and topological structure of multidimensional Delsarte transmutation operators math-ph/0403053 Doug Pickrell: The radial part of the zero-mode Hamiltonian for sigma models with group target space math-ph/0403052 Dmitri V. Vassilevich: Spectral problems from quantum field theory math-ph/0403051 Avinash Khare, Arul Lakshminarayan, Uday Sukhatme: Cyclic Identities Involving Ratios of Jacobi Theta Functions math-ph/0403050 Klaus Kirsten, Alan J. McKane: Functional determinants for general Sturm-Liouville problems hep-th/0309068 Roger E. Behrend, David E. Evans: Integrable Lattice Models for Conjugate $A^{(1)}_n$ gr-qc/0402121 Gavriel Segre: The topological undecidabilities hidden in the Principle of Equivalence of Classical General Relativity math-ph/0403049 Guido Carlet: The Hamiltonian structures of the two-dimensional Toda lattice and R-matrices math-ph/0403048 Christian Gerard Christian Jaekel: Thermal Quantum Fields without Cut-offs in 1+1 Space-time Dimensions math-ph/0403047 C. Quesne, V.M. Tkachuk: Deformed algebras, position-dependent effective masses and curved spaces: An exactly solvable Coulomb problem hep-th/0403207 Christoph Bergbauer, Dirk Kreimer: The Hopf algebra of rooted trees in Epstein-Glaser renormalization cond-mat/0403625 Satoshi Morita, Hidetoshi Nishimori, Pierluigi Contucci: Griffiths Inequalities for the Gaussian Spin Glass astro-ph/0403586 A. Bernui, G.I. Gomero, B. Mota, M.J. Reboucas: A Note on the Robustness of Pair Separations Methods in Cosmic Topology NA: Numerical Analysis ---------------------- math.NA/0403517 Folkmar Bornemann, Christian Rasch: Finite-Element Discretization of Static Hamilton-Jacobi Equations Based on a Local Variational Principle math.NA/0403492 H. Hakopian, M. Tonoyan: Polynomial Interpolation and a Multivariate Analog of Fundamental Theorem of Algebra math.NA/0403484 H. Hakopian: Normal Systems of Algebraic and Partial Differential Equations NT: Number Theory ----------------- math.NT/0404002 Jay Jorgenson, Cormac O'Sullivan: Convolution Dirichlet series and a Kronecker limit formula for second-order Eisenstein series math.NT/0403548 David Joyner, Salahoddin Shokranian: Remarks on codes from modular curves: MAGMA applications math.NT/0403542 Victor Abrashkin: Galois modules arising from Faltings's strict modules math.NT/0403522 Pietro Corvaja, Umberto Zannier: On the rational approximations to the powers of an algebraic number math.NT/0403515 Luis Dieulefait: Computing the level of a modular rigid Calabi-Yau threefold math.NT/0403498 Peter Schneider, Jeremy Teitelbaum: Duality for admissible locally analytic representations math.NT/0403458 Sinya Kitani, Eiki Sawada, Kimio Ueno: Finite automata and relations of multiple zeta values math.NT/0403457 Michitomo Nishizawa, Kimio Ueno: Connection formulas for the confluent hypergeometric functions and the functional relation for the Hurwitz zeta function OA: Operator Algebras --------------------- math.OA/0404001 David Kruml, Pedro Resende: On quantales that classify C*-algebras OC: Optimization and Control ---------------------------- math.OC/0403467 Christopher King, Michael Nathanson: On the existence of a common quadratic Lyapunov function for a rank one difference math.OC/0403466 Joao Jose de Farias Neto: Some Results on Algebraic and Geometric Characterization of Linear Systems Models for Time Series Analysis PR: Probability Theory ---------------------- math.PR/0404015 James Allen Fill, Robin Pemantle: Percolation, first-passage percolation, and covering times for Richardson's model on the n-cube math.PR/0403538 Carl Graham: Functional central limit theorems for a large network in which customers join the shortest of several queues math.PR/0403504 Mark Adler & Pierre van Moerbeke: PDE's for the joint distributions of the Dyson, Airy and Sine processes math.PR/0403499 Jacques Franchi, Yves Le Jan: Relativistic Diffusions math.PR/0403497 D. Feyel, A.S. Ustunel: Solution of the Monge-Ampere Equation on Wiener Space for log-concave measures math.PR/0403461 Jean M'{e}min, Franc{c}ois Coquet, Adam Jakubowski, Leszek Slominski: Natural decomposition of processes and weak Dirichlet processes QA: Quantum Algebra ------------------- math.QA/0404010 Gwenael Massuyeau, Florin F. Nichita: Yang-Baxter operators arising from algebra structures and the Alexander polynomial of knots math.QA/0404008 I. Heckenberger: Finite dimensional rank 2 Nichols algebras of diagonal type II: Classification hep-th/0403277 Anastasia Doikou: On reflection algebras and boundary Yangians math.QA/0403527 Marta M. Asaeda, Jozef H. Przytycki, Adam S. Sikora: Khovanov homology of links in I-bundles over surfaces math.QA/0403488 Alexander V. Karabegov: On the inverse mapping of the formal symplectic groupoid of a deformation quantization math.QA/0403477 Tomoyuki Arakawa: Quantized Reductions and Irreducible Representations of W-Algebras math.QA/0403463 Gaetano Fiore: On the hermiticity of q-differential operators and forms on the quantum Euclidean spaces R_q^N RA: Rings and Algebras ---------------------- math.RA/0404009 Nikolai L. Gordeev, Vladimir L. Popov: Automorphism groups of finite dimensional simple algebras math.RA/0403509 Michael K. Kinyon: Leibniz algebras, Lie racks, and digroups math.RA/0403493 Friedrich Knop: Graded cofinite rings of differential operators RT: Representation Theory ------------------------- math.RT/0403529 Julia Gordon: Motivic nature of character values of depth-zero representations math.RT/0403524 Gregory D. Landweber: Twisted representation rings and Dirac induction math.RT/0403496 Wolfgang Soergel: Kazhdan-Lusztig-Polynome und unzerlegbare Bimoduln uber Polynomringen math.RT/0403487 Alexander Stasinski: Representations of reductive groups over finite rings and extended Deligne-Lusztig varieties SG: Symplectic Geometry ----------------------- math.SG/0403512 Fernando Etayo, Rafael Santamar'{i}a, Uju'e R. Tr'{i}as: The geometry of a bi-Lagrangian manifold math.SG/0403506 Matvei Libine: On equivariant cohomology for non-compact group actions SP: Spectral Theory ------------------- math.SP/0403491 Radu Cascaval, Fritz Gesztesy: J-Self-Adjointness of a Class of Dirac-Type Operators math.SP/0403462 Rakib Feyruz Efendiev: Inverse problem for one class of nonselfadjoint operator's bunches with nonperiodic coefficients -- / Greg Kuperberg (UC Davis) / / Visit the Math ArXiv Front at http://front.math.ucdavis.edu/ / * All the math that's fit to e-print * === Subject: Re: Lower bound on probability Originator: bergv@math.uiuc.edu (Maarten Bergvelt) > sort of material, what would it be? Try Whittle, Probability via expectations. Arnold Neumaier === Subject: Re: Functional equation Epigone-thread: bolgrangphoi Originator: bergv@math.uiuc.edu (Maarten Bergvelt) >|> f(r1,r2,r3) = f(-r1/(1+r2),-r2/(1+r2),-r3/(1+r2)) (1) >I presume you want f defined on R^3 except for the plane r2=-1. >|> f = g(|ln(1+r2)|,r1/r2,r3/r2) (2) >|> I was wondering if (2) is the most general solution and if not >|> how to obtain it. >Note that the linear fractional transformation T: x -> -x/(1+x) is its >own inverse, and takes >(-infinity, -2] 1-1 onto [-2, -1), >(-1,0] 1-1 onto [0,infinity). >Just define f(r1,r2,r3) arbitrarily for -infinity < r2 <= -2 and >-1 < r2 <= 0, and > f(r1,r2,r3) = f(-r1/(1+r2),-r2/(1+r2),-r3/(1+r2)) for all others. >Robert Israel israel@math.ubc.ca >Department of Mathematics (604) 822-3629 >University of British Columbia fax 822-6074 >Vancouver, BC, Canada V6T 1Y4 I've worked a bit on this functional equation . The fact that -r2/(1+r2) is a cyclic mapping - involutive - yields no useful invariants on R .Let me,please explain what I think. Changing,for instance all minus signs ,that is solving : f(r1,r2,r3)=f(r1/(1+r2),r2/(1+r2),r3/(1+r2)), gives an easier solution f(r1,r2,r3)=g(m(1/r2 modulo1),r1/r2,r2/r3) ; a particular case might be :g(sin(2Pi/r2),r1/r2,r2/r3).Anyway your solution seems a very good one. Amiti.8es Alain. === Subject: Metrizable groups and differentiability Originator: bergv@math.uiuc.edu (Maarten Bergvelt) hi, anybody knows good books for metrizable groups and differentaboility properties concerning them. === Subject: Paper published by Geometry and Topology Originator: bergv@math.uiuc.edu (Maarten Bergvelt) The following paper has been published: URL: http://www.maths.warwick.ac.uk/gt/GTVol8/paper16.abs.html Title: Units of ring spectra and their traces in algebraic K-theory Author(s): Christian Schlichtkrull Abstract: Let GL_1(R) be the units of a commutative ring spectrum R. In this paper we identify the composition BGL_1(R)->K(R)->THH(R)->Omega^{infty }(R), where K(R) is the algebraic K-theory and THH(R) the topological Hochschild homology of R. As a corollary we show that classes in pi_{i-1}(R) not annihilated by the stable Hopf map give rise to non-trivial classes in K_i(R) for igeq 3. Secondary: 19D10, 55P48 Keywords: Ring spectra, algebraic K-theory, topological Hochschild homology Proposed: Thomas Goodwillie Seconded: Ralph Cohen, Haynes Miller Author(s) address(es): Department of Mathematics, Oslo University PO Box 1053, Blindern NO-0316 Oslo, Norway Email: krull@math.uio.no === Subject: 1-foliations Originator: bergv@math.uiuc.edu (Maarten Bergvelt) Is it possible for a compact Riemann 3-manifold (M^3,g) to be foliated by non- trivial (i.e. neither points nor circles) recurrent geodesic orbits of a nowere zero vector field xi, none of which realizes the distance between a point p in M^3 and its cut locus C_{m}(p) and none of which has conjugate points? === Subject: Re: matrix inequality > I'm having a little difficulty showing that (if it is indeed true) > Tr_A[ Tr_B(PU^*) Tr_B(UP) ] <= Tr_A[ (Tr_B P)^2 ] Actually, it's not even true! === Subject: March your way into the weekend! March your way into the weekend! Join Nader supporters in DC, April 24-25, as they rally for justice and equal rights Washington, DC: Saturday, April 24th, IMF/World Bank Protest - meet at 11am, 15th & H St, NW. Sunday, April 25th, March for Women's Lives - meet at 9:30am, Smithsonian Metro Stop. If you arrive late, ask a volunteer to point you towards the Ralph Nader group. See you there! Nader: Oil the Source of Two of Earth.89s Major Problems: Wars for Oil and Climate Change Protect the Environment and the People of Our Planet by Ending Our Addiction to Oil Ralph Nader highlighted as two priorities: ending the Iraqi occupation and facing up to the immediate crisis of global climate change. Future geopolitical crises involving oil resources and environmental problems will be diminished by finding alternatives to fossil fuels, said Nader. Nader linked the two issues on Earth Day because he sees them as symptomatic of the corporate oil-based, planet destructive behavior of the Bush Administration. Nader has called for the United States to withdraw from Iraq - including military, private military contractors, oil industry and other corporations. Nader noted: The war in Iraq is unleashing widening cycles of violence. The potential for escalation of violence increases every day the US military remains in Iraq. The way to reverse the spiral of violence is for the United States to go back home. The US presence serves as a magnet for the insurrection, kidnapping, terrorism and destruction. Nader released a three-step strategy for withdrawal earlier this week. See www.votenader.org for details. -- For further information, contact: Kevin Zeese 1-202-265-4000 Matt Ahearn ahearn@votenader.org 1-201-314-9747 -- to perfection! 1 high quality limb, rack, or roast Potatoes, carrot Oil celery onions green onions parsley garlic salt, pepper, etc 2 cups beef stock Marinate meat (optional, not necessary with better cuts). Season liberally and lace with garlic cloves by making incisions, and placing whole cloves deep into the meat. Grease a baking pan, and fill with a thick bed of onions, celery, green onions, and parsley. Place roast on top with fat side up. Place uncovered in 500Á oven for 20 minutes, reduce oven to 325Á. Bake till medium rare (150Á) and let roast rest. Pour stock over onions and drippings, carve the meat and place the slices in the au jus. Bisque .88 l?Enfant Honor the memory of Grandma with this dish by utilizing her good silver soup tureen and her great grandchildren (crawfish, crab or lobster will work just as well, however this dish is classically made with crawfish). Stuffed infant heads, stuffed crawfish heads, stuffed crab or lobster shells; make patties if shell or head is not available (such as with packaged crawfish, crab, or headless baby). Flour oil onions bell peppers garlic salt, pepper, etc. 3 cups chicken stock 2 sticks butter 3 tablespoons oil First stuff the heads, or make the patties (see index) then fry or bake. Set aside to drain on paper towels. Make a roux with butter, === Subject: Re: probability > The grades on the midterm examination given in a large managerial > statistics class are normally distributed with mean 75 and standard > deviation 9. The instructor of this class wants to assign an A grade > to the top 10% of the scores, a B grade to the next 10% of the scores, > a C grade to the next 10% of the scores, a D grade to the next 10% of > the scores, and an F grade to all the scores below the 60th percentile > of this distribution. For each possible letter grade, find the lowest > acceptable score within the established range. For example, the > lowest acceptable score for an A is the score at the 90th percentile > of this normal distribution Why? Why not let the instructor do his own work? Or, better yet, let the students in the class do it as part of their grade. Why fail 60% of the students? Isn't it a bit disingenuous to accept their tuition and then fail them? Either they're really not capable of the work, in which case they shouldn't be accepted into the program, or they are capable of the work, in which case by flunking them out you aren't really communicating anything valuable to the students or potential employers about the students. Except maybe about ruthlessness. Is that really a valuable managerial skill? I know Chainsaw Al Dunlap thought so, but I think he's still unemployed. Jon Miller === Subject: Re: Combinations and Binary Strings > It's possible to count the solutions to > A + B + C + D + E + F = n > > in non-negative integers with B, C, D, and E strictly > positive without needing to know about generating functions, > though, and Bernd might find the approach useful. > Think of A, B, C, D, E, and F as boxes into which you are to > distribute n identical marbles, with the additional > requirement that boxes B, C, D, and E must receive at least > one marble each. To meet this requirement, put one marble > into each of those four boxes; you now have n - 4 marbles to > be distributed arbitrarily amongst the six boxes with no > restrictions at all. Each distribution -- so many marbles > in box A, so many in box B, etc. -- corresponds to a > solution to the equation. > Let's look at the general problem of distributing m marbles > amongst the six boxes. Suppose, for instance, that you have > 3 in box A, 2 in box B, 0 in box C, 1 in box D, 2 in box E, > and 0 in box F; you can represent this schematically as > o o o | o o | | o | o o |. > > With a little thought you can see that each possible > distribution of m marbles corresponds uniquely to a string > of m o's, representing the marbles, and 5 |'s, representing > the spaces between adjacent boxes. There are C(m+5, 5) such > strings, since there are C(m+5, 5) ways of picking 5 slots > for the |'s. > In your problem m = n - 4, so the desired number of > solutions is C(n+1, 5). > Brian Wow. You sir are very good. I like this solution a lot. Very intuitive. I wish I had the intuition to come up with stuff like that. So much for induction. Bernd === Subject: Re: i need help with 3 riddels by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3MBnpv02935; the first on could be either illusion, silence, will, riddle, darkness, consitration, embrace, stride, fall, cycle, routine, promise, tension, spirit, spell or a hundred other things. the answer to the second one is infinity. and the third is sign. === Subject: can anyone help by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3MBnjZ02823; investor wants to invest 30,000 in bonds that are rated x,y,z. the lower ones pay higher interest but are higher risk as well the average yeild is 5%on x,6% on y, and 10% on z investor wants to invest 2 as much in x bonds than z. To get a return of 2000 how much should he invest in each? === Subject: Re: can anyone help > investor wants to invest 30,000 in bonds that are rated x,y,z. the > lower ones pay higher interest but are higher risk as well the average > yeild is 5%on x,6% on y, and 10% on z investor wants to invest 2 as > much in x bonds than z. To get a return of 2000 how much should he > invest in each? Set up a system of equations. x+y+z=30000 .05x+.06y+.1z=2000 z=2x 3x+y=30000 .25x+.06y=2000 -.18x-.06y=-1800 .25x+.06y=2000 .07x=200 x=2857.14 z=2(2857.14)=5714.29 y=30000-3(2857.14)=21428.57 David Moran === Subject: Re: can anyone help > investor wants to invest 30,000 in bonds that are rated x,y,z. the > lower ones pay higher interest but are higher risk as well the average > yeild is 5%on x,6% on y, and 10% on z investor wants to invest 2 as > much in x bonds than z. To get a return of 2000 how much should he > invest in each? > Set up a system of equations. > x+y+z=30000 > .05x+.06y+.1z=2000 > z=2x <...> Should be x=2z. y+3z=30000 .06y+.2z=2000 ==> x=20000, y=0, z=10000 -- Darrell === Subject: Pi and ACT by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3MDlEt21467; First, the speed of light (@186,000 mps) squared gives you pi. Not quite, not yet. A large decimal point problem. The only way to move that decimal point is by exerting great force of Big Bang magnitude. But how to circumvent my ACT score? When ACT and PELL both moved to Iowa City, was that not in fact Big Bang II? Someone check my math. === Subject: Re: Pi and ACT > First, the speed of light (@186,000 mps) squared gives you pi. Not > quite, not yet. A large decimal point problem. The only way to move > that decimal point is by exerting great force of Big Bang magnitude. > But how to circumvent my ACT score? When ACT and PELL both moved to > Iowa City, was that not in fact Big Bang II? > Someone check my math. Somebody check his head! === Subject: Re: Pi and ACT > First, the speed of light (@186,000 mps) squared gives you pi. Not > quite, not yet. A large decimal point problem. No, it doesn't. It doesn't even come close. 186000^2 is 34,596,000,000. Pi would be 3.14159 > The only way to move > that decimal point is by exerting great force of Big Bang magnitude. ...or an overactive imagination... > But how to circumvent my ACT score? When ACT and PELL both moved to > Iowa City, was that not in fact Big Bang II? > Someone check my math. Did you take any math classes? -- Darryl L. Pierce Visit the Infobahn Offramp - What do you care what other people think, Mr. Feynman? === Subject: Distance: Related Rates by support1.mathforum.org (8.11.6/8.11.6/The Math Forum, $Revision: 1.9 primary) id i3ME6Uf24268; Let x and y be differentiable functions of t and let s= squarert.(x^2+y^2) be the distance between the points (x,0) and (0,y) in the xy-plane. (a) How is ds/dt related to dx/dt if y is a constant? (b)How is ds/dt related to dx/dt if neither x nor y is a constant? (c) How is dx/dt related to dy/dy if s is a constant? Can someone please assisst me with this problem by today April 22, === Subject: Re: Distance: Related Rates > Let x and y be differentiable functions of t and let s= > squarert.(x^2+y^2) be the distance between the points (x,0) and (0,y) > in the xy-plane. > (a) How is ds/dt related to dx/dt if y is a constant? > (b)How is ds/dt related to dx/dt if neither x nor y is a constant? > (c) How is dx/dt related to dy/dy if s is a constant? > Can someone please assisst me with this problem by today April 22, Use the change rule. === Subject: Re: Distance: Related Rates > Let x and y be differentiable functions of t and let s= > squarert.(x^2+y^2) be the distance between the points (x,0) and (0,y) > in the xy-plane. > (a) How is ds/dt related to dx/dt if y is a constant? > (b)How is ds/dt related to dx/dt if neither x nor y is a constant? > (c) How is dx/dt related to dy/dy if s is a constant? > Can someone please assisst me with this problem by today April 22, > Use the change rule. And if that fails try the chain rule ;-) Hint: it is easier to work with s^2. If you're not too confident, write s(t), x(t) and y(t) to remind yourself that t is the variable. (BTW, take another look at (c)). -- Paul Sperry Columbia, SC (USA) === Subject: Re: Second Derivative > .... > The question reads: Given: 4x^2 - 5y^2 = -1 . Simplify Fully. > The answer is: 4/25 y^3 <----How do i get this?.... First, I think you meant the question to include Find the second derivative y'' of y with respect to x. Second, the answer is actually 4/(25 y^3) i.e. with the y^3 in the denominator. You need to think of the equation 4x^2 - 5y^2 = -1 as defining a function intrinsically, and differentiate the whole equation: 8x - 10 y y' = 0. The term 10 y y' comes from the chain rule: (d(5 y^2)/dy)(dy/dx). Now you can find y' in terms of x and y, differentiate again to find y'', and simplify. This problem is artificially rigged so that the answer includes (4 x^2 - 5 y^2) which you can simplify from the original equation. Can you make progress now? Ken Pledger. === Subject: Re: Math Help: Second Derivative > .... > And could you show me the work for this question > I'm sure he could, as could I. But this is a problem that _you_ are > supposed to do. JVN has already provided you with a nice outline of > the necessary steps; you should be able to do the rest yourself if > you are to deserve the extra credit.... Right. I saw the question under a different title, and probably gave too much help before finding this thread. It may be too late to matter anyway. Ken Pledger. === Subject: [MMA/Combinatorica] enumerations and graphs I posted this in MMA forum, but like i said in my last post, i wanted to post in here too. I hope someone can shed soem light. sorry for cross posting. ------ hello group, this is kinda like a follow up on my last post. i have a way to enumerate the problem i posted earlier. I wanted to find all tours that are possible with the edges given. Now I have to plot them. consider the following. it will generate 64 lists of edges within a list. la= { {1, 4}, a, {1, 5}, a, {4, 2}, a, {4, 5}, a, {5, 2}, a, {5, 4}, a }; li ={ {1, 4}, i, {1, 5}, i, {4, 2}, i, {4, 5}, i, {5, 2}, i, {5, 4}, i }; lai=Outer[List,Sequence@@Transpose[{la,li}]]//Flatten[#,Length[la]-1]&//ToSt ring; StringReplace[lai,{a,[Rule] EdgeColor -> Blue},{,i,->EdgeColor-> Red},{, a-> EdgeColor [Rule] Blue, i->EdgeColor-> Red}]//ToExpression; Partition[%, 6] %//Length now what I woudl like to do is use combinatorica to show all 64 enumerations as 64 separate graphs using the same vertices as follows. { {{0.0, 1.5}, VertexLabel -> a1}, {{0.0, 0.0}, VertexLabel -> b2}, {{0.5, 0.75}, VertexLabel -> c3}, {{-0.5, 0.75}, VertexLabel -> d4}, {{-1.5, 0.75}, VertexLabel -> e5} } how do I do that? at least how do map those vertices to those edge sets generated by the enueration above so that i canuse in ShowGraph command? sean === Subject: Re: Replacement to MMA and Maple > On the other, each time I try to use > MATLAB for doing pretty graphics, i.e. > high quality for publication, it is > a huge struggle. I am spending a LOT > of time trying to do clean graphics, > the way I want them to be. > Moreover, what I see is NEVER what I get. I don't know what the standards are in your community. But exporting Matlab graphics to eps and using them within LaTeX works quite OK for me. > MATLAB's basic graphic features are nice, > advanced ones suck. It's the same with Maple, > and probably with MuPaD too. If you need advanced graphics in Maple, Mathematica and (probably) Mupad, you sit down and use the Maple/Mathematica/Mupad language to program these graphics on your own. These systems are programmable. As is their graphics subsystem. It's up to you what they do for you. > I repeat: I do not understand this effort > for advanced graphic capabilities. > For other features, I think it is important > that programs can exchange datas, via some > normalized protocol. I don't think it is > necessary for a program to do everything. If you are visualizing a formula (as opposed to some numbers in an array) your data may be a very complex program. To visualize it properly within finite time you may have to adapt your sampling rate (evaluate more points where the function is complicated and less where it's simple and smooth, depending on the size and type of the graphics you generate, etc.). That's why it's good if the tool which knows how to evaluate expressions in the Mupad, Maple, Mathematica language also knows how to plot more or less decent graphics. Besides, some users prefer to purchase one program instead of two. And they prefer to learn one language instead of two. And they may prefer to use their CAS of choice to create completely new types of graphics. === Subject: Re: Replacement to MMA and Maple ... > If you need advanced graphics in Maple, Mathematica and (probably) > Mupad, you sit down and use the Maple/Mathematica/Mupad language to > program these graphics on your own. These systems are programmable. > As is their graphics subsystem. It's up to you what they do for you. I think I know Mma quite well, and I know perfectly well how I might get the system to produce decent publication-quality graphics. In my experience, however, it just isn't worth the effort if I want, or need, to control the graphics and its labels with a lot of detail. What I usually do is just export the numerical data and use it with metapost, which has the added advantage of perfect integration with TeX / LaTeX. As for the value of visualization: in my work, this is of only minor importance - in fact, I usually run Mma in an emacs buffer on the console, and the ASCII graphics there is more than sufficient for me. Also, there is some danger that shiny graphics lead some people to forget about the need to understand their functions - I have seen more than one case where people produced a lot of beautiful pictures but had a hard time interpreting them. (Which is also why I am sceptical about too much emphasis on visualization in teaching, but that is a different topic.) ... ... > For other features, I think it is important > that programs can exchange datas, via some > normalized protocol. I don't think it is > necessary for a program to do everything. Doesn't more or less everything allow you to write a plain text file with the data? That should be sufficient for just about anything, and if it isn't, some simple sed or Perl or whatever scripting goes a long way. Albert. === Subject: Re: Replacement to MMA and Maple Originator: bellenot@haar >What I usually do is just export the numerical data and use it with >metapost, which has the added advantage of perfect integration with >TeX / LaTeX. This sounds interesting, is there a url with some examples? -- Steven Bellenot http://www.math.fsu.edu/~bellenot Department of Mathematics real.email@line.below Florida State University bellenot at math.fsu.edu Tallahassee, FL 32306-4510 USA +1 (850) 644-7189 (FAX: 4053) === Subject: Re: Replacement to MMA and Maple MuPaD in particular, and I didn't want to > start a war. I didn't think you wanted to. > Yes, it is very nice to plot easily > within a mathematical software. It is > very convenient and helps a lot for > debugging. I think that CAS should be > able to do graphics. But not pretty > graphics. graphics are a valuable tool for exploration, which is a large part of (my) research. And the step from good and useful graphics to pretty graphics is actually a very small one -- even features of pretty graphics yet missing in MuPAD such as texture mapping could come in handy every now and then for actual research. And don't forget teaching -- images are a very important aspect of teaching mathematics. > MATLAB's basic graphic features are nice, > advanced ones suck. It's the same with Maple, > and probably with MuPaD too. It depends on the precise results you want to achieve. I have no experience with MATLAB graphics, but I can assure you that MuPAD's graphics are a very long way ahead of Maple's. Have a look at the small galleries at www.mupad.de/gallery.html (actual commands included) and http://www.mupad.com/?&ID=66737&menu=2 to get an impression. > I repeat: I do not understand this effort > for advanced graphic capabilities. Most of the intelligence for advanced mathematical plots is there to make exploration easy. Getting shiny models in 3D is nowadays not much more difficult than drawing anything at all in a way you can do interactive manipulations such as rotating and zooming -- again things you really want to do when looking at some yet unknown structure. > For other features, I think it is important > that programs can exchange datas, via some > normalized protocol. I don't think it is > necessary for a program to do everything. Remember: Standards are great -- everyone should have one. Sure, I'd love to be able to export MuPAD graphics to, say, POVray or something like that, but I haven't seen a tool yet outside a mathematical program that I'd wish to use for interactive manipulation of such objects. -- +--+ +--+| |+-|+ Christopher Creutzig (ccr@mupad.de) +--+ Tel.: 05251-60-5525 === Subject: Re: Replacement to MMA and Maple > I repeat: I do not understand this effort > for advanced graphic capabilities. May be becuase many see scientific visulization as a very important tool to help understand the data they analyze? A CAS system without good support to plotting and powerfull graphics (2D, 3D, animation, zooming, view angles, exporting to different image formats, etc...) would not be too successfull these days. > Using the right tool for the right job > is important. That's what professionals > in any field do, except in informatics, > it seems. So, you'd rather export your data to disk file, then try to import it to another application specialized just for plotting, which hopefully will be able to read the format from the first application, rather than have everything integrated into one application?? good luck selling such a CAS system. Check matlab 7 (to be released soon) new grphics features, they have added more support to better manage these things. Which is a good thing (tm). Matlab has always been strong on graphics, and this is one of its strong selling points. this is a first look screen shots: http://homepage.mac.com/samchops/B733177502/C1647066619/E649021504/ http://www.mathworks.com/products/new_products/R14_prerelease.html === Subject: Re: Laplace transformation software? > It seems that Matlab herits the symbolic mathematics modules from > Maple. I wonder whether there exists any big difference (for example, > the intrinsic function library for Laplace transformation) between > Matlab and Maple especially in symbolic mathematics. My information may be outdated, but AFAIK Matlab uses (parts of) Maple for symbolic calculations, so the only difference should be that you might not be able to access everything Maple offers from within Matlab and that the Maple version used by Matlab might be an older one than you can buy separately. -- +--+ +--+| |+-|+ Christopher Creutzig (ccr@mupad.de) +--+ Tel.: 05251-60-5525 === Subject: Re: Laplace transformation software? >> It seems that Matlab herits the symbolic mathematics modules from >> Maple. I wonder whether there exists any big difference (for example, >> the intrinsic function library for Laplace transformation) between >> Matlab and Maple especially in symbolic mathematics. > My information may be outdated, but AFAIK Matlab uses (parts of) > Maple for symbolic calculations, so the only difference should be > that you might not be able to access everything Maple offers from > within Matlab and that the Maple version used by Matlab might be an > older one than you can buy separately. The Symbolic Math Toolbox does use the Maple kernel. As of version 3.0 of the toolbox, Maple 8's kernel is being used for the symbolic calculations. -- Steve Lord slord@mathworks.com === Subject: Re: Computer Algebra System <92rf80hfki7lqrqlcd4rdqq33d3rs1nld2@4ax.com>Polynomials use a denser representation, where a reference to some > The denser vector of polynomials is perhaps minor. It depends on what you want to achieve. For understanding a general system, true. For understanding polnomial arithmetic algorithms, I would recommend such a structure. And for faster calculations, it is almost required. > An exponent can put into all nodes. It would be an integer. It seems > like a good idea since most of the algorithms can't cope with an > exponent and shifting it into a node createss simple special cases. It would be difficult to support this transparently to the user and I would not expect significant benefits -- to the contrary. > * How is information about the identity of the CPU stored?. >> Why should it be stored at all? > There would be at least a stack for each CPU. Oh, sorry -- MuPAD is not multithreaded, so the question does not arise. >> Certainly in an OO system such as MuPAD each object has a pointer to >>its domain type. (Kernel types do not have a pointer but rather a >>small number, but that's an implementation detail.) Oh, and you did >>not include numbers, functions etc. in the list, which are of course >>pointed to as well. > In Ada 95 a record can be extended without changing its type and by > adding adjacent fields (with no pointers). There must be information stored somewhere about the type. In an interpreted language (with lazy evaluation), this means having some sort of pointer (actual pointer, index into list, ...) to the domain stored inside the data structure. > * What is the interpretation of out-of-bounds lengths of the array?. >> I'm sorry? Which array are you talking about? > Have you got the source code. A URL would be nice. I understand this question even less than the one before that. > I request the reasoning for not getting source file released. You question the very idea of commercial, closed-source software -- fair enough, but this is certainly not the right forum for that. > The topic was made by me to be on the 3GL internal design of Mupad. > I.e. the exact design and not summaries or clarifications of source > code that is not in the public domain. You kept writing to > sci.math.symolic and advocating use of Mupad. I still don't understand your question. The design is that each domain type may have its own idea of what is or is not zero and that it can implement this idea by overloading the function iszero. If you want more detail, see ?iszero for what the kernel domains regard as zero and read the source code for the library domains not covered by the documentation, if there are any. Your question is completely unrelated to closed source vs. open source and I certainly will continue to advocate the use of MuPAD, since I regard it as a valuable tool -- I do not think narrow-minded criteria like but it's not open-source should limit my choice of tools. > An example of a problem with storing the 3 and the x in 3x, in different > nodes/records: > simplify (3*x - 5*y <= -y), > versus: simplify (x*3 - 5*y <= -y) As you can easily verify (and is documented in ?*), products of basic types are reordered to a canonical form. Quoting from the documentation: ,---- | In particular, a numerical factor is internally stored as the last | operand. On the screen, a numerical factor is displayed in front of the | remaining terms: `---- > The data has been online in the past ?. >> Again, I do not understand what you mean. Please clarify. > A yes or no would be a preferred form for the response. I want to > get to the information speedily. Perhaps Mupad is not selling well > enough. I do not understand the question, so yes or no would be inappropriate. Which data are you talking about? -- +--+ +--+| |+-|+ Christopher Creutzig (ccr@mupad.de) +--+ Tel.: 05251-60-5525 === Subject: Particular solution to the Navier's equation Hallo! Is there anybody who could help me? For I'm on a desperate search for months, without success so far and also with dwindling hope that it ever will turn secceed. I'm seeking for a particular solution (in closed form) to the Navier's equation in 3D space with non-constant body forces being described by some function with decay behavior, i.e. some function which influence decreases as the distance from its focus increases. Example of such a function f is e.g. f = exp(-r), f = exp(-(r*r)), f = 1/(1+r), f = 1/(1+r*r) or alike (here r is the Euclidian distance, or radius from the source point to the observation point where the three displacement components ux, uy, uz are sought). On this place I should point out the difference between the Navier's equation (which describes the static equilibrium in a unbounded elastic body) and the Navier-Stokes equation (which describes the dynamic equilibrium or fluid flow). I'm interested only in the Navier's equation which is less complicated being time-independent. Is there any solution for some f function with decay behavior already known? Or does somebody have any idea for which function f could it be possible to find a closed-form solution? As a second alterantive to the suggested f functions, would come into question also the so-called compactly supported radial basis functions (CS RBFs). They have form of a cut-off function and thus the integral over such a function is bounded, what's the necessary property for the functions I need. I've already tried to solve the Navier's equation for the aforementioned functions f directly using Mathematica and also for many other similar functions using the known fundamental solution as a volume integral over the product of the fundamental solution and the f function. Everything without success. I would be thankful for any answer, whether positive or negative, though positive would be preferable. Andrej === Subject: got stuck with an integration Hi ! I am working on a problem in acoustics, and I need to express a pressure function from the time domain to the frequency domain, i.e. through a Fourier transform. / the integration is: |exp(iwt)/sqrt(t*(t+a))/(t+b) dt / from 0 to infinity, with a > b > 0. I have tried all possible ways, with no success so far. The question is if it is possible at all to make get an explicit form to this integral. Any hint is warmly welcome ! D. Ouis === Subject: Re: got stuck with an integration > Hi ! > I am working on a problem in acoustics, and I need to > express a pressure function from the time domain to > the frequency domain, i.e. through a Fourier transform. > / > the integration is: |exp(iwt)/sqrt(t*(t+a))/(t+b) dt > / > from 0 to infinity, with a > b > 0. > I have tried all possible ways, with no success so far. > The question is if it is possible at all to make get > an explicit form to this integral. > Any hint is warmly welcome ! > D. Ouis Here is a special case... b=1, a=2 > lprint(%); Int(exp(I*w*t)/((t+1)*t^(1/2)*(t+2)^(1/2)), t = 0 .. infinity) = (1/2)*exp(-I*w)*(Pi+I*w*( 2*BesselK(0, -I*w) -Pi*StruveH(1, w)*BesselK(0, -I*w) -I*Pi*StruveH(0, w)*BesselK(1, -I*w) )) in the notation of Maple. -- G. A. Edgar http://www.math.ohio-state.edu/~edgar/ === Subject: Re: testing programs in lapack does not run with ifc 7.1 > LAPACK/TESTING directory. In this directory make single complex > complex16 works but when I do make double the computer hangs at > ./xlintstd < dtest.in > dtest.out 2>&1 I am using Intel(R) Fortran Compiler for 32-bit applications, Version > I built the blas and lapack libraires using -O2 -Vaxlib options when > optimization is needed. I used -O2 -Vaxlib when optimization is not needed. You need to recompile a couple of subroutines without optimization for > this too work. Try: > ifc -O ?lamch.f xerbla.f > ar -r liblapack.a ?lamch.o xerbla.o > ranlib liblapack.a > otherwise the testing executable goes into an infinite look. Also I > would add a: > -mp -u > to your make file (and perhaps a -tpp[4,5,6] and/or -x[i,M,K] for better > performance). > Jason The correct way to achieve that is to explicitly specify NOOPT = -O0 (That's -O zero) in the Make.inc file in LAPACK main directory. Whenever a hook exists, use it. And yes, I've been hit by this too; apparently you have to force zero optimization onto the Intel compiler, not telling it anything is not enough. Hope this helps Salvatore Filippone === Subject: Re: testing programs in lapack does not run with ifc 7.1 > The correct way to achieve that is to explicitly specify > NOOPT = -O0 > (That's -O zero) in the Make.inc file in LAPACK main directory. > Whenever a hook exists, use it. > And yes, I've been hit by this too; apparently you have to force zero > optimization onto the Intel compiler, not telling it anything is not > enough. > Hope this helps > Salvatore Filippone Oops, typo.. should be -O0, however I did have NOOPT= -O0 -FI in my makefile and I still needed to do the little trick in the last post to get the test suite to run. I don't think all of the routines that need it are compiled without optimization. Jason === Subject: controlling relative errors when solution = 0 In many general numerical codes for solving ordinary or partial differential equations, one is given an option of controlling relative errors of the solution obtained. However, in many cases of practical interest, the solution may approach zero values, so that the relative error control surely encounters a problem with divide by zero (or equivalently the problem with never satisfying the convergence condition). My question is: are there any known tricks or methods to avoid this problem while designing a general code? I address this question to those who have experience with writing such programs. I'd be happy to find relevant references or program descriptions explaining the workarounds. L.Bieniasz *-------------------------------------------------------------------* | Dr. Leslaw Bieniasz, | | Institute of Physical Chemistry of the Polish Academy of Sciences,| | Department of Electrochemical Oxidation of Gaseous Fuels, | | ul. Zagrody 13, 30-318 Cracow, Poland. | | tel./fax: +48 (12) 266-03-41 | | E-mail: nbbienia@cyf-kr.edu.pl | *-------------------------------------------------------------------* | Interested in Computational Electrochemistry? | | Visit my web site: http://www.cyf-kr.edu.pl/~nbbienia | *-------------------------------------------------------------------* === Subject: Re: Solving Linear Systems in loops Sorry for posting myself. I noticed a profoundly mistake in my Question: I'm looking for numeric computation methods of linear TIME-INVARIANT systems, such that can be written in the form x'(t) = A*x(t) + B*u(t) y(t) = C^T*x(t) + d*u(t) (where A is for example a state space matrix of 3rd or even 4th order) Once again, I would be grateful for any kind of entrance point to that topic. Thanx! _uli -- > Hi all! > I'm a newbie to computational numeric algebra. I want to simulate linear > (and later on nonlinear) systems with C++, MTL and CLAPACK. I know that for > solving linear systems there are some good algorithm I could use quite > comfortable together with MTL or ITL to solve my systems. > Actually my systems (like filters) are mostly of the 3rd order and are > within loops. I would be grateful for some information, hints and literatur > citations about the ways to solve those problems within a bigger system most > efficiently (but with some comfort, since I'm just an engineer ;) > I am of the opinion that for fixed Systems of 3rd order within loops (i.e. a > loopfilter) it is still the best way to solve them by hand just computing > the result with my state space coefficients and eigenvalues instead of using > matrix computations. (Together with Matlab it's definitly the way) > I'm primary interested in learning more about Numeric Computation, so I'am > thankful for any information! > _uli === Subject: Contraction Mappings I am trying to solve the following question Question: Let X = {x elementOf R: x>1} and let the mapping T:X->X be defined by T(x)=x/2 + 1/x, Show that T is a contraction mapping. How do I go about showing T is a contraction mapping? What are the steps involved here? I have searched the net for examples of showing a mapping is a contraction mapping but to no avail. Does anyone have references to such examples. any help appreciated, pat === Subject: Re: Contraction Mappings > I am trying to solve the following question > Question: Let X = {x elementOf R: x>1} and let the mapping T:X->X be > defined by T(x)=x/2 + 1/x, Show that T is a contraction mapping. > How do I go about showing T is a contraction mapping? What are the > steps involved here? > I have searched the net for examples of showing a mapping is a > contraction mapping but to no avail. Calculate a Lipschitz constant (upper bound of the norm of the derivative) and show it is smaller than 1. Arnold Neumaier === Subject: Re: System of ODEs > My aim is the solution of a huge, sparse system of ordinary > first > order differential equations of the general form > > dy/dt = A y = (B + c D) y > > Here, A and B are Hermitian matrices and D is a diagonal > matrix. > The > solution is to be obtained for a series of equally spaced > values of > the real parameter c. How can I make use of the special > structure > of > the problem? Assuming a solution via the matrix exponent is > attempted, > is it possible to use the eigenvectors obtained for one > value > of c > to > aid the calculation of y for subsequent values of c? By the > way, c > spans several orders of magnitude and, hence, c D may not > always be > regarded a small perturbation. However, adjacent values of > c > differ > only slightly. > > Any ideas? > > Johan I assume D is a positive matrix (not really necessary, but > helpful), > and > that B and D are time-independent. Then Step 1: transform the dependent variable, x = sqrt(D) y; Step 2: now the equation is dx/dt = Cx + cIx where I = > unit > matrix > and C = [sqrt(D)]^{-1} B [sqrt(D)]^{-1}; > and since [C,cI]=0, x(t)=exp(ct)exp(Ct)x(0). If [B,D]=0 then > y(t)=exp(cDt)exp(Bt). So c is not the problem but rather having > to > calculate exp(Ct) or exp(Bt) when A is huge and sparse > (selfadjointness > is > of no advantage except that the eigenvalues of A are real) is a > problem > and > is usually done via Kyrov subspace techniques. Indeed, B and D to not commute. I forgot to mention that before. > I > consider using expokit in order to calculate exp(A t)*y_0 before. > Still I hoped I could somehow make use of previous results when > stepping c. > calculate > the vector exp(Ct)D^1/2y(0) (computationally expensive) and have > D^-1/2exp(ct) (computationally inexpensive) operate on it over the > range of > c. Doesn't come much better than that, :-). Yes, you are absolutely right. Unfortunately Julian's transformation > does > not apply. > Johan Johan, Julian's rotating frame transformation is indeed applicable. D is > diagonal. > You need D^1/2 = (exp(D_i,j) d_i,j) where D = (D_i,j) and d_i,j is the > Kronocker delta. You also need D^-1/2 = 1/D^1/2 = (exp(-D_i,j) d_i,j). > You > *don't* need D^-1 (there was a typo in Julian's first post but the > physics > was bang on). Good luck Johan, Gerry T. No, I messed up here (hey, the Toronto Maple Leafs won): exp should have > been sqrt. > But all is not lost, having gone from the y-representation to the > x-picture > and subsequently to the z-picture there is nothing to compel you to go > back > to the y-picture (other than that D^1/2 is noninvertible but physics > doesn't care so long as you know y(0), a given). This is all you need: > x(t) = exp(ct)exp(Ct)D^1/2y(0) > and you're done. > Oops, did it again: C needs D^-1/2. > How about using the Trotter formula > exp(B + cD)t = Lim exp(Bt/n)exp(cDt/n) as n --> oo? n doubles as a scaling > factor in computing the exponential matrices. Indeed, that is an excellent idea! Initially I'm going to try Trotter-Suzuki's formula, which is correct to fourth order. Ideally, I imagine to calculate exp(B+cm D)t at an intermediate value of c, cm, and to obtain an approximation to the matrix exponent at adjacent values of c using the above approximation. I'm going to do some numerical testing and return in order to report on problems or success. Johan === Subject: Re: choice of program/language Matlab is very good, but very expensive. > There is 'octave' of course the open source, free as in free speach. IMHO, I would avoid Object Oriented languages (e.g., C++) unless you can > identify clear benefits from using them. The clear benefits tend to come as the program gets bigger. > Since most programs start small and manageable, but grow big, > complicated and unmanageable, when do you decide to go OO, > when it is too late? > In the end it is mostly a matter of person choice. It is > possible to use non-OO languages in a OO sort of way, but > why make it hard for yourself, unless the client wants > to avoid OO languages. > Jack supports OOP. info. Fortran 95 has derived types (analogous to C structures, but allowing private data and functions), and OOP in Fortran 95 has been discussed in a book Object-Oriented Programming via Fortran 90/95 by John Akin. Fortran, like and C and C++ but unlike Java, does not force you to use structures/classes. Often, you can do the job using scalars and arrays (possibly multidimensional) of the basic types. In numerical work, the structures/classes you do want to use will often be pretty simple (as opposed to those found in a graphical user interface), and you will often use multidimensional arrays of them. Fortran 95 makes it easy to access the components (including array sections) of arrays of derived types, as shown in the following code. I do not think C++ or Python has this functionality. module ij_mod implicit none type, public :: ij integer :: i=0,j=0 end type ij end module ij_mod program xij ! demonstrate arrays of derived types in Fortran 95 use ij_mod, only: ij implicit none integer :: i integer, parameter :: n = 3 type(ij) :: aa(n) character (len=*), parameter :: fmt_i=(100i4) real :: xi do i=1,n aa(i) = ij(i,i**2) end do ! OUTPUT: print fmt_i,aa ! 1 1 2 4 3 9 print fmt_i,aa%i ! 1 2 3 print fmt_i,aa(2:)%j ! 4 9 print fmt_i,aa((/1,3/))%j ! 1 9 end program xij === Subject: Re: choice of program/language John, Obviously if people misuse any tool or method the result is disaster. I find many programmers do not use OO because they are ignorant of OO and think they can normal block structuring is enough. When the application is mostly implementing formulae (as in stats and so on) then you are probably correct, normal block structuring is adequate. How many applications stop at just implementing formulae? At the very least data must be stored and made available at the appropriate time. I could name a big open source C library project that is very difficult to use because it is written in very old fashion block structuring. If parameters are not set up just so, followed by initialisation functions in the correct order, then nothing works. The library has been extended to add wrappers around the base functions, but this hardly helps at all, just another level of complication. As I said at the beginning it comes down to personal choice. If you have vast experince in block structured programs then moving to OO will be a big investment, and probably not worth it for you. But for new entrants to programming I think they should seriously consider OO as a necessary skill. Jack === Subject: Re: choice of program/language >I would only question whether the application requires the use of OO >techniques. Just because a program is big doesn't mean OO techniques are >required. If one cannot identify a need for encapsulation, inheritance and >polymorphism in an application, why bother using OO? I just think its a >little too parochial to believe that every program in the world will benefit >from OO. Lapack is a perfect example of F77 code (non-OO) that is >representative of good structured programming practices. It is not unusual >for OO programs (especially in C++) to become bloated and buggy because of >the improper application of OO techniques. OO requires a large investment >of time up-front before a single line of code is written. Most people skimp >on that investment and proceed directly to developing a plethora of classes >and creating obscure inheritance heirarchies that quickly lead to a >quagmire. Just ask Bill Gates. Yes. I'm sure there are plenty of hard core OO programmers out there who could probably make their codes cleaner and more maintainable simply by dropping their OO in favor of GOTO's all over the place. On a more serious note, I use GOTO on rare occasions. There are certainly times where a GOTO can make code a LOT cleaner and better than jumping through hoops to avoid the GOTO. Dan :-) === Subject: linear function can someone please help me with this problem Let F:V-->R be a linear function. Let W=kerF and {x1,...,xs} a basic of W. Let y be element of V and assume that F(y) is not equal to 0. We wish to show that {x1,...xs,y} span V. Let v be an arbitrary element in V. If v is in W then v is the sum aixi, ai is eleent of R since {x1...xs} is a basic of W. If v is not in W, then we try to find beta so that v-betay element of W. If we succeed at this, then we can write v-betay = sum aixi, or v = sum aixi + betay . This is what we wish to establish. So we have to show that such beta exists. A vector is in W if F evaluated on that vector is zero. So we need to show there exist a betat so that F(v-betay)=0. === Subject: Taylor expansions at different expansion points! I've a question regarding the taylor expansion around two different expansion points. I can say for example : f(x0+h)=sum_n=0^Inf (d^nf/dx^n)|x=x0 h^n/n!. If I expand this function around another point x1 such that x0+h=x1+h' I get: f(x1+h')=sum_n=0^Inf (d^nf/dx^n)|x=x1 h'^n/n! Both these functions are evaluted at tha same position so they must be equal. Therefore there is a relation ship between the different expansion coefficients. It is this relationship that I'm after. I aspecially want to understand the relation ship for a finite series instead of an infinite. If someone can give me a clue in which field of math I've to look I'm happy. Gr, Maurice === Subject: Re: Taylor expansions at different expansion points! >I've a question regarding the taylor expansion around two different >expansion points. >I can say for example : >f(x0+h)=sum_n=0^Inf (d^nf/dx^n)|x=x0 h^n/n!. >If I expand this function around another point x1 such that x0+h=x1+h' and hence x0=x1 >I get: >f(x1+h')=sum_n=0^Inf (d^nf/dx^n)|x=x1 h'^n/n! >Both these functions are evaluted at tha same position so they must be >equal. Therefore there is a relation ship between the different >expansion coefficients. It is this relationship that I'm after. >I aspecially want to understand the relation ship for a finite series >instead of an infinite. >If someone can give me a clue in which field of math I've to look I'm happy. >Gr, >Maurice maybe we change this to x0+h=x1-h=x and evaluate both series at x? and change h to (-h) in the second series if both series converge then you have of course the possibility to express all derivatives at x0 and x1 also by a series at x and this gives you the relation you are searching? hth peter === Subject: Re: MPI SPOOLES question > If the data has to be partitioned > before SPOOLES can use it, is there a particular free(open source) > partitioner that you recomend? Parmetis and Chaco are popular choices. V. -- email: lastname at cs utk edu homepage: cs utk edu tilde lastname === Subject: Re: MathML runtime I would suggest to write an XSLT Stylesheet to transform MathML to C# code (or other) that uses your specific math function library. As MathML input control you could use MathML .NET Control. Bernhard www.soft4science.com > Do you have a solution ? > AC > from >C++ to a more open standard - MathML to be more specific. Is anyone >aware of a piece of software that can read in MathML and create code > (C, >C++, C# etc.. ) from it? >This would mean we could remove any gaps between the algorithm we say >we're using, and the actual implementation. >4Space === Subject: Covergence... If i have a sequence generated by a mapping by which the (k+1)-th term of the sequence is generated by T( k-th term ) e.g. x1 = T(x0)...etc How do i show that it converges to a fixed point on the real line in the interval [a,b] if |T'(x)| < 1 forall x elementOf [a,b] Where do i start? What steps to take? Any help appreciated! pat === Subject: Re: Covergence... >If i have a sequence generated by a mapping by which the (k+1)-th term >of the sequence is generated by T( k-th term ) e.g. x1 = T(x0)...etc >How do i show that it converges to a fixed point on the real line in >the interval [a,b] if |T'(x)| < 1 forall x elementOf [a,b] >Where do i start? What steps to take? >Any help appreciated! >pat never try the impossible. what about T(x)=log(1+exp(x)) for arbitrary real x? lookup some introductory text e.g. on numerical analysis or functional analysis and search for the contraction mapping theorem. you will find another condition , named selfmapping. this one may be harder to show than |T'(x)|<1, which is alo not sufficient, you need |T'(x)|<=L<1, L independent of x hth peter === Subject: Re: Square Number Problem <6U9jc.28$li2.126@news.oracle.com> I am distantly acquainted with a participant in this year's >> Mathematics Olympiad. When I heard about this I looked up some of the >> qualification problems here http://www.bmoc.maths.org/ to see what the >> standard was. This was the first problem, from 1993: >> Find, showing your method, a six-digit integer n with the following >> properties: >> (i) n is a perfect square, >> (ii) the number formed by the last three digits of n is exactly one >> greater than the number formed by the first three digits of n. (Thus n >> might look like 123124, although this is not a square.) >> I didn't have any flashes of inspiration so I started churning out >> algebra eg letting sqrt(n) = 100a+10b+c and then squaring, and got >> nowhere. Generally when I resort to pages of algebra I figure I've >> lost and the answer lies elsewhere. Can anyone here solve this problem >> and provide *hints* about how to do it. >> cheers >> dd >I can't solve it, but I can do some algebra. >n is square, so n=y*y >n is also of the form x*1000+x+1 >so: x*1001-y*y+1=0 1001 * x = y*y - 1 1001 * x = (y+1)*(y-1) 7 * 11 * 13 * x = (y+1)*(y-1) Since y*y is a 6-digit number, 316 < y < 1000. Neither y-1 nor y+1 can be 1001, so the 3 prime factors must be distributed between the 2 terms. That cuts the number of y's to check down to a few dozen: multiples of 77 +- 1 multiples of 91 +- 1 multiples of 143 +- 1 I found some. (Spoilers below) 183184 is one solution (y=11*13*3-1). 328329 is another (y=11*13*4+1). There are probably a more, since 2 out of every [7,11,13] will work. Exactly 6 of them, at a guess. --Keith Lewis klewis {at} mitre.org The above may not (yet) represent the opinions of my employer.