A389

following way, apparently some unwanted ID's appeared in the code. I'm 
still wondering why did that happened (I'm still a newbie). The checksum 
was different than yours, although it compiled correctly and it was working 

like the EVAL command.   
%%HP: T(3)F(.);
!RPL
!NO CODE
::
  CK0NOLASTWD
  '
  ID cmds
  NOTcase
  ::
    No saved commands
    FlashWarning
  ;
  INNERCOMP
  DUP#0=case
  ::
    DROP
    No saved commands
    FlashWarning
  ;
  FALSE
  {
  }
  BIND
  NULL$
  BINT0
  '
  ::
    BINT5
    FALSE
    '
    ::
      NoExitAction
      {
        {
          EVAL
          ::
            TakeOver
            TRUE'
            STO
            FPTR 2 76
          ;
        }
        NullMenuKey
        NullMenuKey
        NullMenuKey
        {
          CANCL
          FPTR 2 77
        }
        {
          OK
          FPTR 2 76
        }
      }
    ;
    TRUE
  ;
  FPTR 2 72
  NOT?SEMI
  ABND
  case
  ID GaENTER
  InitEdLine
  EditString
  CODE 00015 8FB97608F558628DC7530
  STO_CURS_POS2
;
@
 
 
I guess (haven't tried it) that the unescaped quotes () are to blame.
Maybe you can escape them () or use C$ $ in front instead of the
surrounding  .
I'm glad to hear the program's useful for you =)
 
 
 I'll try that in the future. It is strange though, it worked fine 
for the GaEnter program. One more thing, I noticed that the program does 
not keep the history events for things entered through the EQW and MTRW 
unlike the normal CMD history, not even when you DUPlicate or PICK and 
EVAL on the stack. I wonder why is that.     
 
 :
Only the top-level system outer loop is calling AlphaENTER;
it's also not called from EDIT[B], by the way.
IIRC the 48G does store new matrix elements, however,
which shows that it's calling AlphaENTER from there also, apparently.
 
 
fine 
does 
and 
Because GaENTER gets only called if you exit the command line with
ENTER, not if you exit an application (like the editor, EQW and MTRW)
which pushes something onto the stack.  To force the object on stack
level 1 into the history, use a program like
 :: EDITDECOMP$ ID GaENTER ;
 
 
If you find an hyperterminal which can use the usb port you may be able to
send file with xmodem protocol.
 
  
Well, some old programs (that still use the same tokens)
see to work when transferred BACK to the 48G in binary!!!
 
 
No UserRPL programs whatsoever can be directly transferred in binary
are at different ROM addresses, and so are most commands.
no valid addresses at all when transferred back from 49G to 48G
as a *binary* object -- that's why FIXOB etc. are dangerous!
What *is* binary compatible, IIRC:
Maybe some SysRPL programs, containing a very limited set of
entry points, some objects (Grobs, Numeric Matrices, Complex#s,
strings, most but *not*all* real values, user binary integers,
and lists of same).
But always first test on an inconspicuous area (e.g. Emu48 :)
 
 
 
I'm not sure of understand your problem, but when you were in an INFORM
screen, pressing the button CALC (press NXT before), you'll be in the 
stack.
Do your calculations, and press OK.
 
 
That was exactly what I was looking for. Again, Stat48Pro has been
great, better than great, but backing though all the GUI windows was a
pain.
Currently I.82m using this for PORT 2 and use the checksum, 
but it is
way to slow!
Well, if it is fast that would do it. All changes of PORT 2 would be
monitored, if only the stack is used (storing/purging via the stack,
if the object already exists, the calc complains so it has to be
purged before it can be stored under the same name again). I could
check the filer, which can replace objects in a port directly,
seperatly.
Unfortunatly I do not have enough knowledge to do it myself in ML.
I probably forgot to mention that I use local named variables in my
Begin/Until-Loop (also for exiting the loop) and that this might cause
the calc crashing when used in the library config object.
Do INV EVAL INV.
you should first copy the rom to the directory where the convert
program is, then you should go to MS-DOS prompt, go to that directory,
hope this helps.
I am a big fan of HP-Tools (only because I sit in front of a computer
all day and the tools work well with EMU48)
A couple of question, when you transfer the files, I am assuming that
you are using the HP conectivity kit, ensure that you have binary mode
set.
Now, the likely cause, is the need for a header at the top of your
source code to tell the calc. what version you are using, something
like this:
ASSEMBLE
        NIBASC  HPHP49-C
RPL
::
A simple program may look like this
ASSEMBLE
    NIBASC /HPHP49-C/ 
RPL
::                      ( defines the start of a source object )
  CK2&Dispatch          ( check if there are two real numbers on the
stack )
  2REAL 
  ::                    ( if so...)
  ;             
;                       ( ends the main source object )
When I use nbumeric solver for this equation X^4-8X^3+8X^2+32X-44=0 I
only get one answer.
When I plot the graph I can see that it has 4 roots. How can I solve
for all the solutions.
use CAS. I want to see how it solves it.
How do I do it?
You can set flag -109 (for numeric solutions) and -103 (for complex
mode). Then enter the polynimial, enter its variable and SOLVE. If
with I want to see how it solves it you mean that you want to see
the internal procedures, then I must disappoint you, because this
can't be done in this case. (Or did somebody found a way?)
This might not be one of the methods used by the solver, but one way
is to find the eigenvalues of the companion matrix. As cybernesto
found, this seems to work well on the TI-89/92+.
It will work also on the HP49G, but very very slow. And as I never
tried this method, I can't tell if step by step or verbose will show
anything useful then.
It won't.
Yes, that'll work, but is a bit slow on the '49. One advantage in exchange
for the usually massive computing effort needed, is that this method
(eigenvalues) is one of the most robust finding close (and/or multiple)
roots.
I seem to recall how the '49 solves univariate (and to some extent also
multivariate) polynomials;
Roots of polynomials of order 1 and 2 are computed directly. If symbolic
roots are wanted, the GCD and first derivative is computed, which in turn
can yield the roots. If the polynomial has only rational coefficients, all
rational solutions are isolated by testing the finite number of num/denom
pairs made up by the constant coefficients and leading coefficient divisors
(polynomial remainder theorem).
If exact solving fails, completely or partially, numerical methods are used
to find the rest of the roots. These methods are mainly different
factorization algorithms - examples are Bairstow and Laguerre methods.
I believe the Berlekamp algorithm is used too, on the '49.
When solving polynomials by hand, I find that factorization is usually the
most obvious way to do it. When the complete factorization is not obvious
(and hence almost impossible on paper), the eigenvalues method in some form
or another will always get the job done. It mayl take you all day, but it'll 

work...
PS. If anyone is wondering what a companion matrix is, look at
http://mathworld.wolfram.com/CompanionMatrix.html.
The comapnion matrix of the above polynomial is then [[0 0 0 44][1 0
0 -32][0 1 0 -8][0 0 1 8]], and EGVL on that matrix yields [-1.96811878507
2.50401716993 5.96811878507 1.49598283007] (the roots of
'X^4-8*X^3+8*X^2+32*X-44').
The fastest way to solve 'X^4-8*X^3+8*X^2+32*X-44' is by using 
Polynomial49.
the companion matrix method uses 10-12 seconds and SOLVE uses 3.84 seconds.
My HW2 TI89 spends 3.96 seconds on the solution in comparison.
And setting the flag for the VERBOSE state doesn't help more...
To use the CAS, try SOLVE with 'X' (or SOLVEVX where 'X' is the default
variable -aka the VX-). But that won't tell you how to proceed exactly as
you certainly know what to do : a FACTORing or a X COLLECTing first...
Make sure the flag 01 is unchecked for general solutions.
Both the solve application and the stack commands
for finding roots (SOLVE and PROOT)return the former
an array and the latter a list with all four real roots:
1.496, 2.504, -1.968, 5.968. I don't understand what
you did or didn't do to come up with the same results.
That'll of course only return one slution (That's how it's supposed to
Try X^4-8X^3+8X^2+32X-44
SOLVEVX in approximate mode
Numeric Solver - Solve Poly.
Enter:
[1 -8 8 32 -44]
Press solve and get:
Roots:
[1.49... 2.50... -1.96... 5.96...]
This version rounds to nearest integer square root, and returns 
TRUE / FALSE.
If you do not care about TRUE / FALSE flag, ie. then you might 
test to see if another iteration is necesarry, and in many cases
avoid last iteration. ( which would reduce run time with about 
20-40 % for large integers.) 
This version is only efficient for integers with size below 250, but will
work for any size.
Further it doubles the accuracy for each iteration.
I believe it does not cycle between results, as may be problem 
if  N = ( A -1 )*A , where a is integer.
* returns square root if Zinteger, with flag for exact result
* i.e. same as ZSQRT , but for any size of Zinteger 
*  uses Newton-Raphson iteration with doubling of accuracy
*  for each iteration
*  x' = x + (y-x*x)/x /2
* --------------------------------------------
DEFINE QAdd     FPTR2 ^QAdd 
DEFINE QSub     FPTR2 ^QSub 
DEFINE QMul     FPTR2 ^QMul 
DEFINE ZDIVext FPTR2  ^ZDIVext 
 
  CK1NoBlame 
  FPTR2 ^CK1Z
  FALSE SWAPDUP 
* calculate 16 digits of square root, convert to Zinteger
*  ID Z2%%2 %%SQRT %%FLOOR ID %%2z 
*  calulate at least 8 correct digits  FPTR2 ^ZSQRT
  DUP OSIZE BINT10 #-    ( d digits in square root )
*  #8                                      ( n dynamic size, init.  2 * 8)
   #4                                      ( n dynamic size, init.  2 * 4)
  DUP                                   (a digits to be removed )
 {{ a n d }}                    
  BEGIN 
    ROTDROP
    DROPDUP !n              ( update dynamic size, but max d)
    d SWAP#- !a               ( digits to remove from x  )
    OVER a #2* ID noff    ( downsize y )
    OVER a ID noff           ( downsize  x ) 
    DUPDUP QMul
    ROTSWAP QSub       
*  check if exact result 
    DUP ZINT 0 EQUAL 5UNROLL 
    SWAP 
    ZDIVext DROP 
    ZINT 2 ZDIVext QAdd  ( correct rounding )
    #0 a #- ID noff               ( back to orig. size )
    DUPUNROT 
    QAdd SWAP 
    ZINT 0 EQUAL  
  UNTIL 
  SWAPDROPSWAP
  ABND    
;
@
:: 
* ID noff 
* truncate ZINT 
* ZINT #off 
* guard against underflow 
* Gjermund Skailand 2002-03-06  
CODE 
  GOSBVL POP#
  GOSBVL SAVPTR
  C=DAT1.A D1=C D1+5 C=DAT1.A 
  C-6.A 
  C-A.A LA 50000
 A=C.A 
GOVLNG PUSH#ALOOP 
ENDCODE 
 ID setsize   
; 
@
::
* ID setsize 
* ZINT #  
* guard against underflow 
* Gjermund Skailand 2002-03-04  
* # is digits only, does not include sign
NULLHXS SWAP#1+ EXPAND     ( include sign )
CODE 
  GOSBVL SAVPTR
  CD1EX D1=C D0=C
D0+5 C=DAT0.A D0=C A=DAT0.A  % type
C=DAT1.A D1=C DAT1=A.A            % write type
D1+5 C=DAT1.A D=C.A
D0+5 A=DAT0.A
% point to tails
B=A.A AD0EX A+B.A AD0EX
B=C.A CD1EX C+B.A CD1EX  
?C<A.A SKIPYES { C=A.A }
C-5.A
GOSBVL = MOVEUP
GOSBVL = GETPTRLOOP
ENDCODE 
SWAPDROP   
; 
@
The ZSQRT2 was bugged, and wrong version was posted.
This version should work.
It  does not cycle.
However, if the number  is not a perfect square, it will not necessarily
return 
the next lower integer. Since the original ZSQRT does,  then perhaps it
should, However I made the ZSQRT2 for use with floating point square roots
(long reals).
It compiles to checksum # 5819h    and 289.5 bytes
::
* ZSQRT2    2002-04-09 Gjermund Skailand
* returns square root if Zinteger, with flag for exact result
* i.e. same as ZSQRT , but for any size of Zinteger 
*  uses Newton-Raphson iteration with doubling of accuracy
*  for each iteration
*  x' = x + (y-x*x)/x /2
* --------------------------------------------
DEFINE QAdd     FPTR2 ^QAdd 
DEFINE QSub     FPTR2 ^QSub 
DEFINE QMul     FPTR2 ^QMul 
DEFINE ZDIVext FPTR2  ^ZDIVext 
 
  CK1NoBlame 
  CK&DISPATCH0
#FF ::
  DUP ZINT 0 Z< case FPTR2 ^Z<0ERR
  DUP ZINT 0 Z= caseTRUE  
FALSE SWAPDUP 
* calculate 16 digits of square root, convert to Zinteger
*  ID Z2%%2 %%SQRT %%.5 %%+ ID %%2z 
*  calulate at least 8 correct digits  
   FPTR2 ^ZSQRT DROP
  DUP OSIZE BINT10 #-   
* ( d digits in square root )
   #4                                      ( n dynamic size, init.  2 * 4)
  DUP                                   (a digits to be removed )
 {{ a n d }}     
  BEGIN 
    ROTDROP
    DROPDUP !n    
    d SWAP#- !a           
    OVER a #2* ID noff    
    OVER a ID noff     
    DUPDUP QMul
    ROTSWAP QSub   
    DUP 5UNROLL
    SWAP 
    ZDIVext DROP 
    ZINT 2 ZDIVext 
    OVER ZINT 0  Z= IT :: ZINT 0  2DUP Z< NOT_IT SWAP DROP ; 
    QAdd                       
    #0 a #- ID noff    
    DUPUNROT 
    QAdd SWAP 
    ZINT 0 Z=
  UNTIL 
  SWAPDROPSWAP
  ABND    
  ZINT 0 Z=
 ;
;
@
..
I have amended my SQRT2 program, my claim is that it:
1, returns true for perfect squares
2, if the number is not a perfect square, e.g. Z = A*A -1 then it returns
A-1
3, it does not cycle. 
The below may not be a complete proof, however
I found that cycling may occur with Newthon -Raphson when a number  Z = 
A
* (A+1). 
Because the iteration is a' = a + (Z - a*a)/a/2, 
if a = A then you get
a' = A + (Z - A*A)/A/2 = A + (A*(A+1) - A*A)/A/2
a'= A + 1/2
if a=A+1 you get
a'= A+1 + (Z - (A+1)(A+1))/A/2 = A+1 + (A*(A+1)-(A+1)*(A+1))/(A+1)/2
a'= A+1 + ( A*A + A - A*A - 2A - 1 )/(A+1)/2= A+1 + (-A-1)/(A+1)/2
a' = A+1 - 1 / 2 = A - 1/2
Thus the correction is +/- 1/2, since we are using integers, the result
depends on the following:
if  1/2 is rounded upwards, you get +/-1 and never 0 correction.
if 1/2 is rounded downwards, you don't cycle, but you need to adjust the
result
if you want to always get the lower result. 
Further you need to check if  Z= A*(A+1), since the correction would be 1/2
and 
that would be rounded down to zero, and would result in returning A+1 or
cycling,  
depending on how you check for iteration to finish.
::
* ZSQRT2    2002-04-10 Gjermund Skailand
* returns square root of Zinteger, with flag for exact result
* i.e. same as ZSQRT , but for any size of Zinteger
* returns nearest integer, or rounded down, see comment
*  ZSQRT rounds down, 
* uses either ZSQRT or %%SQRT for initial value, note #4, #8
*  slow for small integers, 
*  uses Newton-Raphson iteration with doubling of accuracy
*  for each iteration
*  x' = x + (y-x*x)/x /2
* --------------------------------------------
DEFINE QAdd     FPTR2 ^QAdd 
DEFINE QSub     FPTR2 ^QSub 
DEFINE QMul     FPTR2 ^QMul 
DEFINE ZDIVext FPTR2  ^ZDIVext 
 
  CK1NoBlame 
  CK&DISPATCH0
#FF ::
  DUP ZINT 0 Z< case FPTR2 ^Z<0ERR
  DUP ZINT 0 Z= caseTRUE  
FALSE SWAPDUP 
* calculate 15-16 digits of square root, convert to Zinteger
*  ID Z2%%2 %%SQRT %%FLOOR ID %%2z 
*  calulate at least 8 correct digits  
   FPTR2 ^ZSQRT DROP
  DUP OSIZE BINT10 #-   
* ( d digits in square root )
* #8 or #4
   #4                                      ( n dynamic size, init.  2 * 4)
  DUP                                   (a digits to be removed )
 {{ a n d }}     
  BEGIN 
    ROTDROP
    DROPDUP !n    
    d SWAP#- !a           
    OVER a #2* ID noff    
    OVER a ID noff     
    DUPDUP QMul
    ROTSWAP QSub   
    DUP 5UNROLL
    SWAP 
    ZDIVext DROP 
    ZINT 2 ZDIVext 
    OVER ZINT 0  Z= IT :: ZINT 0  2DUP Z< NOT_IT SWAP DROP ; 
    QAdd                       
    #0 a #- ID noff    
    DUPUNROT 
    QAdd SWAP 
* exit if 0 correction and full precision
    ZINT 0 Z= a #0= AND
  UNTIL 
  SWAPDROP
  ABND    
* the result is now rounded to nearest integer
* final test to see if we should round down
  OVER ZINT 0 Z< IT :: ZINT 1 QSub ;                    
* test if correct result, return TRUE/FALSE 
  SWAP ZINT 0 Z=
 ;
;
@
 Should compile to checksum #2683h , 329.5 bytes. 
I have also amended the test for iteration to finish.
Presently the test ZINT 0 Z= will return false if you test against ZINT 00
( leading zeros),
but this may change in the future, and then we would get premature ending
since 
intermediate calculations are performed with reduced precision.  
The program  could probably be cleaned up some, all the tests are using a
lot of bytes. 
http://www.mum.edu       http://www.tm.org
Another question: How do I recall a individual element from a stored
matrix or list in System RPL?
.
,
I'm having a bit of a problem with units in the following user RPL
program on a HP49G:
        << 43.01 MENU
        ; do some calculations here and leave it on the stack...
OK.  The 43.01 MENU command displays the UNITS | LENG menu so that 
I can enter a numeric value with a length unit; ie.1.25_ft.
Pressing the [ENTER] key places 1.25ft onto the stack; it's missing
the required underscore in the unit object.
The 'ft' is stored in n1 leaving 1.25 on the stack, with the same
process repeated for the next two INPUT statements.  This is obviously
not what I want to have happen, but I'm a bit stumped on how to fix
this.
I recall seeing some units programming information in one of the two
48SX manuals that came with it, but I sold the 48SX when I got my 49G
(I really goofed there).  Any Ideas on how to get around this problem?
The problem is that INPUT invokes only a text editor,
with the PRG (programming mode) indicator on,
and for menu definitions which are unit objects,
the menu keys in PRG mode merely echo the name of the unit
into the text, rather than performing their ordinary
immediate mode unshifted definition, which is: '1_anyunit' *
Curiously, each *shifted* definition still echoes what would have
been executed in immediate mode: '1_anyunit' CONVERT or /
How about Enter Number 1? PROMPT instead?
The PROMPT command halts and allows completely normal
keyboard operations to be used to create data.
For each PROMPT command, you'll need to continue the program
via CONT, rather than via ENTER; this can be expedited
by displaying your own custom menu instead, e.g.:
{ 1_ft 1_in 1_m 1_cm { } { OK CONT } } TMENU
Enter N1, press OK PROMPT 'n1' STO
Enter N2, press OK PROMPT 'n2' STO
Enter N3, press OK PROMPT 'n3' STO
I was going to suggest trying INFORM here, but although you can
get a nice-looking form with three labeled values to enter,
and each value can be required to be a unit object,
the exact same problem occurs when trying to use unit menu keys
when entering data from a form!
If you want to stick with INPUT, you'll just need to insert
the _ characters, using either the keyboard underscore key
or your own custom menu key; if you want to allow combinations
of units to be built, all the operators (e.g. ^ or /)
will require manual entry.
But wait!  A dumb trick could also be used:
You enter 23ft; the value 23_ft gets stored into 'n1' !!!
But anything this absurd is almost as bad as using a TI ;-)
Your explanation (as always crispy clear), John, about menu commands
that are put as text on the command line during INPUT, brings an idea.
If we don't need many different units but only a ffew of them, say ft
and in, then we can make a menu:
{ _ft _in }
and use TMENU prior to INPUT:
        << { _ft _in } TMENU
        ; do some calculations here and leave it on the stack...
In this case when INPUT prompts us for data, pressing the menu key
[_ft] puts the unit with its underscore after some input number.
If we don't like the undercore on the label of the menu key, then:
{ {ft _ft} { in _in }} TMENU, so that the menu labels are 
ft
and in but pressing the keys still puts the units with underscores
on the command line.
Well, actually, they *are* called dummy variables, more so in contexts
like tensor analysis (dummy indices in that case).
-
a long time ago (when mathematicians were philosophers), it was
believed that a number had to be either a boy or a girl. even numbers
were male and female numbers were odd. the circle was the most perfect
body and therefore heavenly bodies were perfect. also, the male sex
was more perfect than the female. therefore rounded eggs, being nearer
circles, must be males. :)
Riemann? ;-) Or do you mean Gauss?
When I was studying general relativity, I was talking to my friends
about smooth manifolds. I guess I was the only one dumb enough not to
make the correspondence, because my friends were laughing at me,
saying that I liked smooth manifolds, and I had no idea what they were
talking about :(((
Wishing I were 'normal',
Oh, since we know that embedding them in appropriate spaces can smooth
them, it's only a matter of finding that appropriate space ;-)
We're getting dirty here, aren't we? Mwahahahaha!
.
No, let us hear it. Hope that the next calc will also conatin these
things in its CAS ;-)
Hmmm, I though it was love and potence = marriage, which leads to the
a algebraic structure with + that works like 3 + 8 = 5.! + in this
case is a symbol for the operation subtract number to the left from
number to the right.
Did he add numbers for money, or do we assume { -oo , 0, +oo }? ;-)
And what happens to 2 when 8 has already taken place ? Does it
disappear? ;-)
.
is there any way to get the results of opperations on matricies to show up 
normal decimals to show as a fraction, but i can't figure out how to do it 

to
a matrix.
Just out of curiousity how do I use the XQ command. I can't find it 
anywhere? Not even in the manual. I have a 48G. not the + or the GX, 
just G.
Of course, you've to convert to lists therfore... 
The HP48G[X] doesn't have MAP but you can use APLY instead;
APLY comes in the examples created by TEACH on the 48G[X], or see:
The above alternate version of APLY is even useful on the 49G,
if you want to avoid fighting the CAS over flag settings,
because MAP, which once left flags alone, now [1.19-6] demands
changing them to suit the CAS (and purging a variable, too!)
Variable taxation and Flag-changing without necessitation is tyranny!
But HP's original APLY program from the 48G[X]
*does* handle symbolic matrices, if that's what results,
and will work as indicated; its output is then, of course,
in the form of a list (for a vector) or a list of lists (for a matrix),
which is actually exactly the same as it is in the 49G,
except that the 49G gives it a new object type prolog,
and also continues to display, edit, or process it as an array.
I wonder whether they had the 49G in their mind's eye
and has bugs, anyway (ouch, sorry for this awful comparison :)
Try it if you have a 48G[X]:
Q.E.D.
Yes, I know the APLY program, and have fiddled around with the TEACH lib
many years ago. I would just point out that the program wouldn't return
symbolic matrices.
Hum... According to me { { '6/5' '5/2' } } is a symbolic matrice. I mean a
valid list (its elements are only reals, complexes, algebraic, variables 
and
Zintegers) well formated (i.e. a valid matrice dimensions) is a matrix. So 

{
{ 1 3 } { 2 4 } } is a 2x2matrix but { {1 3 } 2 4} is not a matrix... Many
symbolic algorithms on 48 series use that trick and 49g too (exept that a
structure is the same, and AXL just change the type naming to suit some
routines)
A better original APLY program (including symbolic results) for 48G[X]:
Revisiting the APLY program - all new, fast, no bugs [1996/07/21]
found that XQ does it in a single command!
But, it should take you back to exact mode... and Variable taxation and
Flag-changing without necessitation is tyranny! (sic). Hum... Maybe Mr
Parisse supposed that if you want to convert (for just seeing ?) to exact
result then you want to perform exact computation (and then need the exact
mode) ..?
to lists...) You may also use the command XQ, but you'll be turned to
exact mode.
Does the last sentence mean that you already have a list?
Which ones do you think you want to use?
Have you tried decompiling them with Nosy or Jazz?
This document names many Kermit entry points, without description:
http://www.hpcalc.org/hp48/programming/entries/ent_srt.zip
I find the following useful, especially with Emu48/49:
These entry points translate between external ascii representations
rmdr is any trailing characters of the original string
which represent an incomplete backslash sequence, e.g.
(the next packet would then be concatenated with this remainder,
to continue translating and compiling, packet by packet).
Each of the above respects the current translation mode
(last value in IOPAR), which can be set by xTRANSIO (DOTRANSIO),
ranging from zero (no translations) to three (all translations).
K[IN]VISLF inserts/removes a carriage-return before each linefeed
(for external PC text), except when the translation mode is zero.
UserRPL programs which employ the above (useful in Emu48/49):
==
I've been trying to write a program to solve a quadratic equation. here's 

what i have so far:
when i enter 3 variables into the stack, and eval the program, i get an 
answer to the first expression, but the second just comes up as the 
expression i.e. '(-b - (b^2 - 4*a*c)^-2)/2*a'. is there any way to use the 

local variables a b and c for both expressions?
local variable only exist for the next equation or program, thats why it
worked for the first and not the second equation.  Enclose both in program
delimiters to get it to work.
..
what i have so far:
answer to the first expression, but the second just comes up as the
expression i.e. '(-b - (b^2 - 4*a*c)^-2)/2*a'. is there any way to use the
local variables a b and c for both expressions?
Your proposal, Steven is almost OK, but it needs two additional EVAL
after the two algebraic objects that are enclosed between << and
  << '(-b + (b^2 - 4*a*c)^-2)/2*a' EVAL
      '(-b - (b^2 - 4*a*c)^-2)/2*a' EVAL
The reason for this is: When an algebraic object follows directly the
local variable, then it is evaluated. But when the algebraic is in a
program that follows directly the local variables, then it is simply
put on the stack, just like any algebraic object in any program does.
Algebraic objects that follow directly after local variables
assignments, are treated in a special way, they are evaluated.
Of course you couldn't omit the two EVAL, if you would use RPN syntax
in the program after local var assignments:
  << b NEG b SQ 4 ....and so on
,
What are those backslashes for? They seemed to interfere with entering
the program in RPN mode. I had to get rid of them and add brackets
around the (1/2) and the (2*a) to get it to work.
A backslash indicates that what follows is a representation of a 
non-ASCII symbol as used by the HP calculators.
sigma and SIGMA represent the lower and upper case sigma symbols, 
and so on.
This reminds me of a USER Rpl challenge long ago. This program was submitted 

by
some clever programmer (I can't remember who, but I put it in my HP48
directory!) It is sort of instructional in the use of the stack and RPN as
opposed to algebraic solutions. And only 50 bytes! Place a, b, and c on the
stack, then execute:
Lastarg must be enabled for this to work. Also the HP49 has PICK3 available
saving 2.5 additional bytes!
Simple, for the 48/49G:
This can give completely wrong results;
three different coordinate display modes (RECT, CYLIN, SPHERE),
because in the latter two modes it takes some of its three arguments
to be *angles*, in which case the automatically converted
*rectangular* components of the resulting vector
will not be the original arguments!
Note that the last two components are *angles* !!!
If you now press cursor-down (to edit in Matrix Writer),
you will see that the actual vector result is:
[ 4.5 2.59807621135 3 ]
which are the corresponding *rectangular* coordinates!
A fact to remember:
*converts*coordinates* in CYLIN or SPHERE modes,
Yes but it will not work for symbolic coeffs a,b,c. 
.
This will merely get the two expressions echoed to the stack. Add an EVAL
after each expression, or calculate the solutions via the stack (the two
expressions evaluated in RPN).                                               

                                                                             
 
                                 
In the 19-5 beta roms there is a numeric solver called MSLV. Rumours say 
that this is a numeric solver for systems non-linear equations. However I 
can't seem to find any documentation on it. The questions are:
1) Is it really a numeric solver for systems of nonlinear equations?
2) Does anybody know how advanced it is - does it just simple Newton-
iteration or does it employ more sophisticated methods?
3) How is it used? It's not interactive so my guess is that it needs 
equations and names on variables on the stack, but in which format and 
which order?
 
If you need an equation solver that can handle multi occurrences
of a variable in an every equation (unlike MSLV) and so can handle
a larger range of (non-linear) equations try the excellent 'GBASIS' 
(Groebner
bases)command. I have used it successfully with results comparable to the
classic 'solvesys' program although the latter gives you direct numeric
solutions while the former supplies you with a set of polynomial equations 

that
in one variable each that form a basis for the solutions.
!Demeter!
Yes, but it is only for polynomial things. (For which it works very
good.) If you have other things, (trigo ;-)) then it is not directly
suitable.
Anyway, has somebody counted how many solvers and specialized solving
commands  the tiny HP49 has? :-)
Hmm, a numeric solver should find numeric solutions for *any* kind of
equations, since it uses numeric methods. So it shouldn't matter if
you have linear or non-linear or anything else. Or do I miss something
(again)?
Well, it is, because of the reasons above.
No idea! It does its work OK, though.
Enter the vector with your equations, example [X^2-SIN(Y) 'Y-1=3*X']
Enter the vector with the variables to solve for, [X Y]
Enter guess solutions for X and Y in a vector, [1. 1.]
Then use MSLV to find the solutions. MSLV returns (after a while) the
vector of the equations, the vector of the variables and the vector of
solutions. You can test how good the solutions are, by using these
arguments for the program:
<<
  SWAP AXL SWAP AXL =
     
 If you like interactive solving then perhaps you should also try
the Multi Equation Solver, yet another numeric solver that lives
somewhere in the HP49G ROM.
This is mostly a semantic questions. But, imho, one cannot ignore the type 
of problem when choosing a solver. Eg it's hard to solve differential 
equations with a solver that only find roots of polynomials. In the case of 

the HP49, you can't input a system of non-linear equations in the Solve 
equation... interactive solver (the menu reached by r-shift 7). :)
Then there is the problem of some problems that can't be solved numerically  

 
(at least not with the methods known today) unless the initial guesses are 
the exact solutions (let's ignore that exact is not a very good word in 
the numeric world), not even if an infinite number of steps in the given 
algoritm are taken. Of course, I don't have a good example, but I bet 
someone can come up with a differential equation that illustrates. :)
lot! :)
It doesn't seem to be in any of the obvious places... Is 19-6 beta ROM 
needed, perhaps - I use 19-5 right now.
Totally forgot the existence of differential equations and some other
beasts. I only thought about normal equations, but then again, you
are right, there will be enough cases where one numeric algorithm
finds solutions and the other doesn't.
Fortunately enough the HP49G has a big number of different numeric
solver software built-in, including solvers for differential
equations. :-)
I use 19-6 and it is menu 116, which you get by 116 MENU. Do you get
that also with 19-5?
*trying* Yes, I get a menu and when I press NXT I can choose MES which 
gives the MES LIB MENU with items MSOLVR, MINIT, MITM etc. However none 
of these commands have help and none of them brings upp a form to fill in.
Guess it's time to upgrade to 19-6?
http://bugs.hpcalc.org/show_bug.cgi?id=6 [scroll way down]
Note that MES is a purely numeric solver, which works only
with certain sets of equations; the criteria for its use
is not linearity, but rather that there can be found
a sequence for solving the equations in the set for which
only one variable in each equation will be unsolved as yet,
upon which the original single-equation numeric solver is then
applied; MES simply figures out the sequence and automates
the repeated invocation of the one-equation numeric solver,
using status bits for each variable to remember which ones
are known or solved for vs. unknown as yet.
This happens to suit all the equation sets in the original
HP48 equation library (of various common physical systems),
which is just what MES was developed for and included with,
first as a plug-in card for 48SX, later built into 48G[X],
then omitted from the earliest 49G, then finally restored again :)
It is the Multiple equations SoLVer that came with the EQuations 
LIBrary
on the 48 series...
For the use of it, try
http://www.hpcalc.org/hp48/docs/programming/aug012.zip or
http://www.hpcalc.org/hp48/docs/programming/hp48gaur.zip or
http://www.hp.com/ ?
Is the library aviable for the 49g?
You can use: matrix {13 12} GET where matrix is your matrix.
Or 'matrix(13,12)' EVAL.
I am using the G series ml entry point ChkLowBat, 325AA. Jim Donnely
described it in his book. It works well at detecting a low battery state of
the main batteries. When the batteries have just been replaced, it still
indicates a low battery state for a few hours, though. A warm start does 
not
clear this false indication although the alarm annunciator does not 
indicate
a low battery state. How can I use this entry and successfully determine
that the batteries have just changed.
If the LowBat condition was detected during I/O activity, it may be stored 

into
PrtStatus (ChkLowBat checks this 1st for an earlier LB condition).
OpenUart will reset it. Or add this piece of code before calling ChkLowBat:
    
    D0=(5)  =PrtStatus    (80924)
    C=DAT0  A
    CBIT=0  =sLowBatErr    (7)
    DAT0=C  A
                            
Sample of a better solution:
    CODE                        ( Check for low battery condition )
        GOSBVL  =SAVPTR
        GOSBVL  =DisableIntr
        C=0     A
        GOSBVL  =ChkLowBat
        GOSBVL  =AllowIntr
        ?ABIT#0 1                       System battery weary ?
        GOYES   BattNOK                 yes
BattNOK GOVLNG  =GPPushT/FLp
    ENDCODE
-- 
If we use unsupported entry points and get rid of the C=0 ( the
contents of C aren't needed in ChkLowBat ) then it shortens down to :
CODE
        GOSBVL  =SAVPTR
        GOSBVL  #3213E
        ?ABIT#0 1
        GOYES   BattNOK
BattNOK GOVLNG  =GPPushT/FLp
ENDCODE
----------------------------------------------------------------------------

ChkLowBat mucks around with the printer status bits 
( PrtStatus : # 80924h ) which causes problems for successive battery
readings because it sets the low battery status bit ( 7 ) in PrtStatus
if a low battery condition is detected and from then on reports any
reading as such *unless* bit 7 is cleared in PrtStatus . So an easy
fix would be to do something like :
PrtStatus       EQU     # 80924
        D0=(5)  PrtStatus
        C=DAT0  B
        CBIT=0  7
        DAT0=C  B       
        ST=0    15
        GOSBVL  =ChkLowBat
        GOSBVL  =AllowIntr
The above can actually be shortened by using some unsupported entry
points :
        GOSBVL  #32A2C          * Get PrtStatus into C[B] and ST[7:0]
        CBIT=0  7
        DAT0=C  B
        GOSBVL  #3213E          * Disables interrupts, calls ChkLowbat
                                * and then re-enables them via
----------------------------------------------------------------------------

you have to disable interrupts first to get a reasonable result from 
ChkLowBat.
Use ST=0 =sALLOWINTR before the call.
Raymond
Portland State Bookstore has them for $29.   Any better deals out
there? 
Also, is there a version designed for DCE instead of DTE?  (Example:
to connect to a modem instead of a PC, receive and xmit data pins
reversed.)
Or is there a good connector which can be used to connect to the
hp48sx side, and I could then make my own cable?
Why not try http://www.cynox.de for the cable.  If I recall correctly,
it is $14.00 with a shareware CD ROM.
http://www.cynox.de/en/intasch.htm will take you right there.  The
price is $14.01, and while you're there, why not snag some other
deals.  I bought the HP 48 G+ with 256K for $121.23, which is a hard
to find bargain.  They obviously open it up and replace a memory chip
with a bigger one
if anybody knows how the scores work. (not the scores file, i mean how
points are given)
If anybody knows anything, please post here or email me at the address
miz.
If I remember well, each 'type' of brick has a different score associated 

to
it. Each time you break a brick, it's 'value' gets added to your score.
Well, the problem with that is that I think I've had a higher score
with a game where I got to a lower level (and so have broken less
bricks).
Also, I've found that in most (if not all) levels there are some
ghost invisible bricks which you can shoot (not breakable with ball)
instead of breaking normal bricks. That is, if you shoot three of
these ghost bricks, you finish the level when there are still three
bricks left. In fact there are more strange things, like level 2 where
you can't shoot past the second row until you have broken some certain
brick on the left.
If I remember well, there is a bonus given depending on how fast you clear
a level. Any other thing (like gost bricks) could only be a bug...
not that I know of...
An old joke between Collin and me. I have a Ginger chicken on my monitor
(from chicken run) and Collin used to 'pinch' the chicken, and hide it
somwhere, or do something with it (like hide it in the plants, mimicking an
attack with a plastic snake) and send me emails entilteled: 'Have You Seen
the Chicken' :-) I might have some picture somewhere....
I've heard that APPS, TOOL and CAT did nothing on the first emulators.
Seems logical. TOOL is useful, CAT is so-so, but APPS... Who would
call the Equation Writer thru it, for example? While some commands are
hidden, other appear in three different menus (or more) without any
good reason. 48 had some repeated commands, but not that way
I still think that lack of time was the reason not to improve those
rough edges. The size of ENTER is not a problem, but following 49's
path, ENTER would be available thru APPS menu too.
: I've heard that APPS, TOOL and CAT did nothing on the first emulators.
: Seems logical. TOOL is useful, CAT is so-so, but APPS... Who would
: call the Equation Writer thru it, for example? While some commands are
: hidden, other appear in three different menus (or more) without any
: good reason. 48 had some repeated commands, but not that way
: I still think that lack of time was the reason not to improve those
: rough edges. The size of ENTER is not a problem, but following 49's
: path, ENTER would be available thru APPS menu too.
You can easily change the assingments, e.g. with Keyman. This is what I
did:
- Swap APPS/FILES
- Swap HIST/CMD
- Swap SYMB/EVAL
- Mode form on CUSTOM
- Programming custom menu on MODE
- Math custom menu on TOOL
- Normal tool menu on long-TOOL
- Graph screen on RS-LEFT
- OVER on LEFT
- 3 PICK on long-LEFT
- ROT on LS&LEFT
- UNROT on long-LS&LEFT
I have heard the term Meta Kernal used several times and I am not sure
what this is. I went to HPCALC and seen that it can be downloaded from
their site and installed on the 48GX. I have an older HP 48GX, ROM
version 'P'.
Would this help my calculator out any. Would it add any extra features
not already available on the GX.
An decent Text Editor, Equation Writer, Pretty Print Stack(like the 
TI-92+),
RPL and ASM programming...
If you can get an 128K card, try it. Sure that you will simply LOVE it!
Download the documentation and read. It's the best for a 48GX (with 128 Kb
At the beginning it can seem difficult, but it is not. Simply different.
I think it is a must.
Can I use it on EMU48?
I'll e-mail you an Emu48 with MK installed
Would it be possble to e-mail me one also.
Al
Ok
Well. First of all, it's not 'Meta Kernal', but 'Meta Kernel'.
It will add a good functionality: speed enhance to the general behaviour
of your HP48GX. You'll have a new stack program, some SysRPL and ASM
compilers/decompilers, the same filer as the one in the HP49, and some
other nice stuff.
card (1 PMEM gives me 35 bytes left).
-- 
-----
nous semblons .90tre entour.8e d'un quarteront de d.8enigreurs, qui 
non
seulement n'apportent pas de r.8eponses, mais en plus s.8fment le doute 
dans
l'esprit des demandeur... Restez modeste .8da fait du bien.
-+- Joe in: Guide du Neuneu d'Usenet - Neuneugaulle, le recours -+-
I wonder if anyone can answer a question about the CAS system for me. 
I'm working on a 40G but I believe that the 49G system is similar enough 
that you may be able to help.
I'm differentiating y=(x+1)^4/(2x-3)^3.  Whether I use the DERVX 
function or the d/dx function (the steps of working are slightly 
different but the result the same) I get an answer of 
(2x^4-12x^3-48x^2-52x-18)/(16x^4-96x^3+216x^2-216x+81).
If I now paste the FACTOR command into the top and the bottom I can get 
it to give me an answer of 2(x-9)(x+1)^3/(2x-3)^4
My question is, why did it not give me the second answer in the first 
place when it is clearly the better answer?  I would have thought that 
it would give me the second answer and require that I use EXPAND if I 
wanted the first one.  ie. If the CAS system is going to symplify at 
all, then why is it not simplifying correctly to the better answer?
Is this some flag (on the 40G a 'config settting') or is it some 
philosophical issue that I'm not aware of that makes the first answer 
'better'?
I take your point about the fully expanded form being easier to 
differentiate a second time but surely what is of greater interest is 
which form is more useful - the fully expanded form or the form of 
2(x-9)(x+1)^3/(2x-3)^4.  I would have thought that the single most 
common application for the derivative is to find where it is equal to 
zero and the second form is far more useful for this (if you're going to 
work algebraically).  In fact, the fully expanded form is almost useless 
for that purpose.  Consequently I don't see why the CAS defaults to 
giving the fully expanded form.
 
Fair enough.  I suppose my perspective is that of a high school teacher 
and so I concentrate on the uses I make of the derivative and forget the 
others that I haven't used much since uni days.
Ok.  May I humbly suggest that it is an issue worth considering if there 
is ever a major revision?  :-)
 
Why isn't (x+1)^4/(2x-3)^3 a representative of a generic fraction, in
other words what would be such a representative? Does it have to do
with the fact that the deniminator is zero for x=3/2? And how is a
generic fraction defined in this case?
Nick.
 .
more often involved in problems? In other words, if we encounter a
fraction, is it more probable that it will be generic?
 
 
The factored answer may be better from your perspective, for the problem
that you're solving. But the unfactored rational expression can be more
useful, for example if you want to determine its limit towards infinity
(1/8). The 40/49 does not attempt to simplify beyond the immediate answer,
it's up to you to tell it to go further. The first answer is neither better
nor worse than the first one, it all depends on what you're doing. It's 
true
that, as you're studying a derivative, the factored expression may be more
useful to you, but the calc has no way of knowing that. The whole 
philosophy
of the CAS is to make you think about what you're doing, not letting the
calc do everything for you. It may seem wierd at times, but in the end 
gives
you more freedom to do what you want to do. That is my point of view, at
least.
Jeremy Gibbons
gibbons@enst.fr
3CB64E2B.305@iinet.net.au...
G'day Folks
Upgrading from a 39g to a 49g and just wondering if the 49g has a way of
storing equations like the 39's 'SOLVE' applet?
I find it really convenient to place commonly used equations there, 
although
10 places is small. The best part is that I can then quickly solve for
another variable in the equation.
cheers
/Victor
 
Did you try [<-i]&[F1(A] (Blue left shift key and F1 key : Y=)
The Tremendous Tiran of TETRI, Peter Geelhoed proved that he
got indeed a score of over 10,000 points in the TETRI game.
Moreover, he produced a self-performance of his marvellous 
game which is on my site under the name TDEMO.
He must have been working very hard: first of finding a suitable
random seed allowing to survive 10 levels, next to find the winning
strategy, and finally to record his strategy in such a way that
it's reproducable by itself. Therefore, he deserves the reward,
a nearly new HP49, which I will sent to him as soon as possible.
Here the link to Pivo's self-performing TETRI game:
ftp://ftp.math.fu-berlin.de/pub/usr/raut/HP49/games/TDEMO.zip
Note that in contrast to all other Tetris versions, TETRI allows
if an obvious mistake was made. 
 
PS. Some collegues (mathematicians) say the TETRI game alone
is worth the Dollars for a new HP49 - and I agree.
 
 
instead directory.
 
 
Yes - but I can do this only towards end of May. Maybe there
is somebody who could do it earlier, its not particularly
difficult with JAZZ(light).
- Wolfgang
Sorry for asking more: and a SAVE option. Is it not interesting? Like
pressing NXT twice, but it would let us use the calc for other tasks before
reasume the game.
This would be very hard. It can be done but I fear that the reproducability
of the game is gone. Perhaps saving all variables and then running the 
right
amount of %ran's would do it. I tried it in my demo but failed miserably.
--
This message was written with 100% recycled electrons
Pivo
What lacks on the 49 is a general suspendor which freezes
all background applications including the screen and perhaps
also PICTURE. As I was told by Raymond, the HP38G has such a
suspendor or Task-Switcher. The HALT command is insufficient;
it does not save the screen and perhaps some other parameters.  
 
 
TTRM's in ML than SysRPL:-)
Then I just played a lot of games. (45 mins a piece)
Writing an interpreter for the keyrecording was easy, took me about 30
minutes.
The most work was removing the extra keypresses from the higher levels,
the interpreter starts at level 1
and therefore it takes about 15 minutes before you reach level 8.
Anyways, I get a new 49.    
 
 
  
Does anyone know if an application that can be used to view and search 
through text files exists for the HP48GX?
I've found a few text viewers, which works fine - but I would like something 

where I am able to search through the textdocuments in e.g. a directory in 
the memory for a specific word. 
 
 
Rummage through these, and see whether any fit your purpose:
http://www.hpcalc.org/hp48/apps/database
http://www.hpcalc.org/hp48/apps/pims
http://www.hpcalc.org/search.php?query=search+AND+string&hp48=1&title=1&desc

ription=1
 
 
I have once more hacked together a hopefully useful program during my
holidays. They're over on Monday, so you won't get another one :-(
Anyway, these Extable Tools are there to help you personalize your
extable. The library consists of two commands: One converts an extable
into a string which lists all its entries, and the other does the
opposite.
One of the more useful applications of this is that you can add the
entries in rament.S to the table. You'll immediately notice the effect
in Nosy =)
Source directory, library and documentation are availiable as usual,
either through mail or soon on hpcalc.org.
 
-- 
 
 
Maybe, you could integrate your 2 tools in the original extable, this would
allow us not to have 2 libs on the system!
 
 
 
A way to manage entries as we do with Jazz were needed... You give us more.
 
Oups. was (a kinda EC and others were in need). Sorry for the mystake, I
didn't read back before to send.
 
Three months ago (on Jan 25th) I asked about the trouble I had with
the EQV application in this app: fin49a.zip from 2002/01/16 by John H
Meyers (ported to the 49 as FIN49A.BIN, an HP49G version of FIN421B
[HP48G] authored by Don [MacDonald R.] Phillips)
John answered quickly and tried to help me as usual. 
I didnt't write you back then, because I quit trying. 
Anyway, I'd just updated my ROM (for the very first time) to VERSION
119.6 -from my previous original Release 110-, and everything is
working all right, including the FIN49A app, that doesn't make any
mistake now.
 
Whoever wants a fine financial app (containing 11 independent
financial applications), you can get it at:
http://www.hpcalc.org/hp48/apps/misc/fin421b.zip [48G only]
http://www.hpcalc.org/hp49/apps/misc/fin49a.zip  [49G only]
 
Does anybody remember an application about ERROR and Uncertainties
from Jay Kile, for the HP 48? and knows where to find one similar for
the 49G ?
Carlos 
Including the Txt file from then. 
ERROR (Uncertainties by Jay Kyle)
This is a program that adds, subtracts, multiplies, raises to a power,
etc uncertainties or in other words numbers such as 2 .96 0.5.  The
program has full instructions.
Item: 237 by _jk at hpcvbbs.UUCP
Author: [Jay Kyle]
  Subj: Uncertainties
  Keyw: 'ERROR' (x.96dx)  [that's x-plus-or-minus-dx, you non-IBMers!
 
Common error analysis is done in generally two different ways, an
absolute
method, and a random or systematic method. The absolute method
compounds
absolute errors ie. (2.960.5) +(3.960.3)= (5.960.8) whereas the 
random
method
uses a less pessimistic approach.  In the case of this program, I
utilized the
general formula:
     dz = SQRT[(df/da)^2*da^2+(df/db)^2*dB^2+...)]
                 ^
                 |
            (partial derivatives)       for   z=f(a,b,...)
which has a 90% confidence in the answer.
  The program has both methods in it.
  The program also utilizes a custom menu for choices of operations.
I've
included a work directory as part of the program simply to go into to
do any
calculations, a single program called WIPE exists that wipes the work
slate
clean  so to speak. This is just a quick way to clear work directory
after
you're finished with a particular formula. I borrowed this concept
from an old
hp 28 program.
beta
testing it for me.
(User.programs)
Item: 239 by _jk at hpcvbbs.UUCP
Author: [Jay Kyle]
  Subj: Uncertainties math program
  Keyw: x.96dx, error, uncertainties
  This is an updated version of my uncertainties program, ( v 1.2 )
that now works with unit objects as well as real numbers. The main
description is in message #237. I have renamed it ERROR, short for
error analysis.
error (5904 bytes)
[Uncertainties math program for lab data]
Enjoy
Jay Kyle
UBC
Chemical Engineering student
Did anyone tried this out ? 
First, I received a WARNING Invalid Card Data. 
After the PINIT routine, I still had problems and basically, it seem
written for another machine.
 
 
works well 
  
It seems that there are exercises out there that every college student
can solve by hand but which I can not solve (easily and directly) with
my HP49. I want to solve the above equation symbolically. With SOLVE
for example. But even a SQ command on both sides does not get me any
further (still saying not reducible to a rational expression). I
have to isolate manually (!) the remaining root, then square again
etc. Maybe I missed a command which does it directly.  By the way, the
result is x=2. PC Programs like MathCad do it directly (sorry for
comparing two worlds, I just wanted to make sure that it IS possible).
 
 
Forgive me if I'm mistaken, but from what I understand, what you're 
suggesting is under the assumption that A^2+B^2 = (A+B)^2 which is WRONG!
The correct answer as previously stated is x=2. 
If you square both sides you get the equivalent and harder expression to 
solve: 
3*x+7+2*SQRT(2*x^2+9*x+10)=x+23
Factorizing and simplifying the original expression you get (not easy):
(1/-4)*(50-2*SQRT(841)) == 2  Correct Answer! 
The other apparent solution comes from: 
(1/-4)*(50+2*SQRT(841)) == -27
 
 
Try to solve it with the numeric solver.
 
  I don't know if anyone has suggested this, but try the following:
Let T = SQRT(2*X+5), U = SQRT(X+2), and V = SQRT(X+23).  Then we can
rewrite the original equation as T+U-V=0.  Also we can get rid of the
square roots by
squaring both sides to get T^2 = 2*X+5, U^2 = X+2, and V^2 = X+23. 
array of variables), and now run SOLVE.  Does this work?
  
 
Your manual manipulation technique in the EQW is the best way to go here.
It's pretty neat, but a shame SOLVE can't do it. SOLVE give up very 
quickly.
With your method, you get two solutions: X=2 and X=-27. The X=-27 is of
course false, and appear when you square the equation (You of course know
that, but it's a good example of introducing falso roots when manipulating
expressions).
 
When x = -27, sqrt(2x+5) = +/- 7i, sqrt(x+2) = +/- 5i,
and sqrt(x+23) = +/- 2i.
So, by choosing the signs properly (7-5=2), the solution is validated!
Of course one can do that. But as the original equation gets more
complicated, including more square (or even higher) roots, the whole
process of choosing the right signs and right roots gets also more
complicated and can cause errors, simply because my eyes and brain get
confused.
But on the other hand, what else can we do?
http://web.jet.es/leobueno/
How to call the chars editor from an User program? (MetaKernel 2.30 in a
48GX)
As I said, I'm using a 48GX (with MK and Erable): 22,3 KEYEVAL is what I
need... BUT I get the original HP chars editor, and I want the MK chars
editor (is like 49's: much faster, etc)
It's been a while since I've used a HP48GX with the MetaKernel in it.
from the stack. It should do the trick.
I had a quick look at the source code, and I don't think there is another
way of doing it.
In fact, calling Shift CHARS does exactly this: put the system font on the
stack, call EDITB then remove the remaining font from the stack
Yes, I posted yesterday that I'm using this way, but when exit (in edit
mode), the screen is not refreshed... I can see only the actual line of my
program and the rest of the screen is of chars editor.
When MK is installed in a 48, the new chars editor is got in the *same way*
that the original one: pressing RS CHARS ( 22.3 code key). The same occurs
with the matrix editor, the filer... The new MK applications simply
substitute the old ones. No special commands or User keyboard are needed.
But 22.3 KEYEVAL or -22.3 KEYEVAL (Erable command), gets the original
instead the new one... and that's the problem.
*There is a way* for doing it but with a problem that I don't know how to
solve: I explain the story.
What I want is create a new CHARS key in User mode: pressing CHARS I'll get
a CHOOSE box with several options, like change font size, UFL editor, etc,
and CHARS editor.
And it works.
I assign the new definition with IF PRG option of Keyman, for using it
when programming.
But when I use it in this case, the screen is not refreshed when exiting 
the
editor.Impossible to read. And I must press ENTER, wait for the sintax
revision of the program and edit again. It's boring..
 
Is there a way to graph a polar equation from a specific interval? For
example the graph r=3cos(theta) from the interval 0 to pi?
  
On the 49g
I believe what you need is to go to Plot Window, (F2 key) and set the
On the 48
I believe is done by going to Plot, (8 key) and Opts (D key) and set
the indep varible and the hi and lo
  
Sounds strange, but any body know how to do it?
I think:  Plug an HP-TO-PC serial cable into an TI-TO-PC serial cable(use 

an
serial adapter).
I want to send pictures(created with the Geometry application in TI-92
Plus[hardware 1]) to HP-48.
  
i think TI uses a propietary serial protocol.
you could use a microcontroller with build-in UARTs and convert protocols.
that would require a small project box witth a batt 
or if you could connect a TI to a pocketPC (i doubt this) and download into 

the 49.
 
Then copy your HP48 ROM to your computer!
Install Emu48 and stop with that thing in comp.sys.hp48!
 
It's true. TI89/92+ seems too much user-friendly, i found it laggy compared
to a HP49(assume that TI92+ has an 12MHz processor!).
FYI, i am 14 and I am learning now ASM(!!). I've did some SYS-RPLing.
It's like getting used to Linux when everybody uses Windows.
    
I typed in:
2:  '4*x^2+3*x-4+(2-x)=4*x-2'
1:  'X'
Typed solve and...
SOLVE Error:
Problem Unitialized.
  
I'm guessing you have the SIMP library installed (L1785). That simplex 
library has a conflicting SOLVE command. You're actually calling that 
solver and not the built-in. A while back John Meyers posted the following 
solution:
The S~N command is in the built-in development library
(APPS menu, usually item 12)
You need to have library 256 attached. 
  
As you have entered it, you are trying to solve for a variable, 
capital X, which does not occur in the equation, which only contains 
lower case x.
Make sure that all occurrences of the variable agree as to case, 
since variable names are case sensitive.
  
   
It works fine here - you may have a problem with SysRPL or ASM programming
gone bad?
 
 
Anybody know how to type math expressions with [] and {}, like:
{3+5*[9-(17/4])}
Can't type [] or {} into expr...
 
 
Use ( ) - they mean the same.
 
 
 
Anybody know how to make that thing works?
Tried DDE48, no luck.
 
 
No chance for direct loading. You must use a cable.
  
 
I defined this units in a new Unit Folder called STORAGE.
bit
Now tried to convert 11_byte to nibble. Shows INVALID UNIT.
 
=  pe #2 RES. Select a poorer quality.
   
Buy a speed-up modul?
Switch to TI?
   
Don't forget the funny TI-BASIC(funny even to Basic programmers!)
Prefer 1000x HP's INFORM command than TI Dialog...EndDlog.
Compare:
(a step=a command).
TI-89:
: MyProgram()
: Prgm
: Dialog
: Text Nothing new!
: EndDlog
: EndPrgm
7 steps.
HP-49:
.8e
5 steps, counting the { }.
(used choose to make titlebar).
Which circuit solver and Controls programs are you talking about?
Danny
Controls:  Bode-Routhe
Circuit Solver:  CSim  (numeric only)
Nice programs.
JLA
I was trying to write something like a control program and wanted to make 

it
100% SysRPL. But... How to R/W from serial port?
I have some old computers(an 486 and an 286) and old serial cards that i
made sort of an data collector...
It outputs something like(made an program in BASIC to I/O it):
date/time: 01/01/2002 21:33:44
observation number: 100
system speed: 66 mhz
math copro. working: 1
mem: 8mb
Last log-in: 01/01/2002 21:45:29
Last calculation made:
[4*5]-(2/{3-5})+9*6
And the program should write:
turn off to exit system
continue to get more info
lag to slow it
s to speed up.
I don't want ML code(i know nothing about ML!)
 
1. Enter the null directory(I don't remember the SYSEVAL).
2. Start Filer.
3. You will see a lot of junk on the folder list.
4. Nothing serious; press ENTER on one, home directory.
 
 
Since it was never intended to
use the hidden directory as a current directory,
you venture into a grey area to call it a bug
No doubt it's possible for JYA to protect against this
      http://www.mum.edu     http://www.tm.org
 
[#9d SYSEVAL] ( clear mem )
 
 
 
Are you saying about VECTORS?
Simply, while in MTRW, press VEC softkey to go to vector mode.
Then enter 1 line x n cols vector.
 
   
1st time that used an 48: i sometimes slammed on keyboard(wanting to type
[CapitalYAcute][derivative], slammed SHIFT, got some sort of ...).
 
 
No. on HP49 i made an 2048x1536 grob:137 KB.
 
 
Yes, as my later posts say, I agree on that. I was talking about Erable and
the '49 CAS. Can you provide the example that resulted in your 205 kb
expression?
 
 
200000 MAKESTR 'name' STO  -- no problem at all!
(it just can't be copied into port 1 or port 2).
  
  
Well... in a port maybe. But in main memory you can have quite a bit
more. I just tried the following:
CODE
SAVE
% Just so that the kernel won't try to show a string
D0-10 LC(5) DOEXT3 DAT0=C.A
GOVLNG GPPushR0Lp
ENDCODE
The resulting object was 222032 Bytes big.
 
 
True, you can build objects manually that are only a little smaller than 
256
kb, maybe 230-240 kb. Neither Erable nor the '49 CAS is able to do that,
when returning from internal routines, since they usually keep a copy of 
the
data worked on in one or more local variables. There are exceptions to 
this,
but I still haven't seen one in practice.
 
 
 
Yes - compared to the programming abilities and the number
of build-in functions of that little marvel - the book was big.
The current 49G Manual is just a Quick Start Guide.
To get the User Guide and Examples on using CAS
you'll have to buy the Urroz books...
 
 
 
 
http://www.greatunpublished.com/Authors/Gilberto Urroz.htm
I received them in one week. VISA.
 
 
I found these books very helpful. 
I bought them via Internet at:
http://www.greatunpublished.com
  - Science and Engineering Mathematics with the HP 49 G -
21 US$ each. 
I did the download of the e-book version (pdf) and printed it on a
laserprinter.
This may be a bit more expensive than the printed version (25 US$
each) but there is no trouble with taxes or long shipping time. (I
don't have any experiences in that business)
There were some difficulties with the print-job. 
1. I didn't succeed in printing with zoom (120 %). It only worked wizh
100 %
2. I was not able to print the whole book in one. I had to make parts
of about 50  pages.
But these problems may be connected to my printer driver (Kyocera
3700+, Adobe Acrobat Reader 5.0)
The conversation with greatunpublished.com was easy and very friendly.
 
 
One Link: http://www.engineering.usu.edu/cee/faculty/gurro/myBooks.htm
Follows the hyper threads...
 
 
somebody knows where the unsupported pointer #14E6D (also known as
How can one find an unsupported pointer in a 49G ?
 
 
Dunno, but you could easily substitute the underlying code with your own;-)
DOTRUE,DOFALSE)
 
 
Do you mind posting the code, since I don.82t know how to dig 
it out of the 48GX?
 
 
Is there any way to graph polar equation from an interval? For example, to
graph the equation r=3*cos(theta) from the interval 0 to pi, could you do 

it
on the 49G?
 
Yes, in the PLOT WINDOW - POLAR screen enter 0. for INDEP Low and 3.14
 
 
I wonder what is the simplest way to add 
a screen shot to program in hpcalc.org?
 
 
Most 49G CAS commands pop up dialog boxes demanding mode changes
(e.g. RADian mode) which remain permanently changed thereafter;
they also force you to purge the current variable named in 'VX',
whether or not that variable has anything to do with the command,
and sometimes *don't* ask you to purge another variable which *does*
need to be purged to get a correct result.
The functions below eliminate all prompts, and restore
all modes, flags, and any purged current variable
exactly as they were before executing any CAS command;
they also set the *right* variable for commands which *don't* use VX.
%%HP: T(3)F(.); @ start here for Kermit download
HOME
<< RCLF -55. CF RCLVX DUP IFERR RCL THEN END
      EVAL THEN 1. ELSE 0. END ERRN SWAP
        xx. vx. IFERR STO THEN DROP2 END fl. STOF
@  end of downloadable file (execute once, then discard)
Usage examples:
'X' STOVX
'COS(X)' { INTVX } XC           @ commands which use VX
'Y^2-Y-6' 'Y' { SOLVE } OVER XV @ command argument is temporary VX
100 17 { IQUOT } '~' XV         @ VX is irrelevant to the command
{ HALT } XC                     @ do manual commands, CONT to restore
Creating a sample CUSTOM menu of selected CAS commands
(using this menu preserves all flags and variables)
  { } XQ STOVX } MENU
NOTES:
Requires ROM version 1.19-6 or later.
Backing up memory is recommended before use.
Would you like to be able to choose
between several different pre-stored custom menus?
<< Choose CST menu { 'Mymenu1' 'Mymenu2' ... }
As you probably know, a custom menu saved in HOME
will be the default custom menu for all lower directories
which don't have their own custom menus, and so on.
 
 
 
I just received my HP48Gx with the speed module and have found some
odd things about it and I wonder if someone has an explanation.
The first simple test I ran was timing the internal clock by watching
the clock display on the screen.  Took 30 seconds to see one minute go
by, indicating that the internal clock was indeed running at double
speed.
But Then I ran same actual calculation tests and found that with the
double speed module on, calculations took an average of 75% of the
time that they did with it deactivated.  I say average, bacause,
seemed that, different type of operations were perfomed faster than
others, with those which accessed the display repeatedly being faster
than those that were calculation intensive.
Performing ON-D and then A with the speed mudule off, gave a speed of
3918033, with the module activated I got a speed of 2431280 (62%
difference).  I'm not sure how to interpret this result.  It would
seem to me that if the clock display runs at 2x speed then the second
number should be exactly 1/2 of the first.
Can someone throw some logic into these findings?
By the way, after using my 49g for a couple of years and playing with
this 48GX, I realized how true it is!!:  The 49G keyboard is a piece
 
 
Pressing ON-D and then A produces a nunber which is suposed to be the
clock speed of the calculator.
How does the calulator come up with this number?
As far as I know, in order to measure the speed of a clock, you need a
second clock to use as a refference.
 
 
It would certainly seem so to me. Bill Wicke's HP 48 Insights
says that the time clock is based on a (nominally) 32768 Hz
quartz crystal oscillator. I expect that the CPU clock is also
there is a second, *independent*, time counter that's used for
the cursor blink. I *guess* that the ON-D A test measures the
CPU frequency in terms of the cursor blink counter; probably not
especially accurate, but accurate enough to detect any gross
error in the functioning of the frequency multiplier circuits.
If anyone can add to this or correct my assumptions I'd
appreciate it.
 
 
        Basically, the calculator executes a sequence of instructions
of known cycle count for 1/16 of a second (using nibbles 1-2 of TIMER2
as a reference) keeping a count of the number of iterations. The final
iteration count is then multiplied by the cycles per loop. This is the
clock speed divided by 16.
        Actually, the whole chip is based off of *one* 32768 Hz
crystal controlled clock. This signal is multiplied by an PLL to get a
high frequency clock which is then suitably scaled down by various
on-board components for their individual use. The timer, display and
LED sections are based directly off of the 32768 crystal clock. 
        All the cursor blink counter consists of is a 16-bit
software counter ( =T1COUNT ) controlled by TIMER1  which is itself a
16-bit hardware counter controlled by bit 8 of TIMER2 (therefore
TIMER2 has to be running in order for TIMER1 to function).
  I meant to say 4-bit here. Guess my subconscious was thinking
of 2^4 - bit instead. There goes my brain again. ;)
----------------------------------------------------------------------------

 
 
In case anyone is trapped in self-test mode, you can get out of
it by pressing ON and C together, which will cause a
warm-start, which, among other things, clears the stack and
aborts any suspended programs.
 
  
http://www.hpcalc.org/search.php?query=clock+AND+speed&title=1&description=1

Joe Horn copies clock speed test from 48 ROM:
http://www.hpcalc.org/viewzip.php?id=2882&file=spdg.txt
Preston Brown (of HP) explains that GX is a DX2
http://www.hpcalc.org/hp48/docs/misc/speedup.txt
Dave Arnett explains speed adjustment jumpers on GX board:
http://www.hpcalc.org/hp48/docs/misc/jumpers.txt
Dan Kirkland's deep explanation (and his own code):
 
 
is there a speed routine for the 49G?
 
PS: rcobo: how about hacking the 49G
    to speed it up with an external adjustable clock?
 
 
 
i remember a trick i saw in TV. a tourist (in a tourist attraction)
put a coin in a mailbox-like machine, then the machine started talking
about the history of the place. when the tourist left, a small
trap-door opened an a midget came out for a smoke break.
PS: i saw an atomic clock application that *was* feasible to build for
the HP49G. when i called the company, they told me that they didn't
sell it no more  
  
Another way to preserve the time is to drive the CPU and
Timer/LED/Display sections by separate clocks, which is indeed
possible. ( by setting certain bits in the Yorke MODE register - which
can only be done by interfacing with the external Saturn bus AFAIK. )
This would have the advantageous side effect of reducing the power
consumption since only the CPU ( and UART ) are being overclocked.
   
http://www.trancenet.org/law/denarot.shtml .
 less):  www.mumbull.com .
 
http://www.geocities.com/bbrigante/bull2.html .
 
www.suggestibility.org .
   life. a paradox?
       How do you submit any arbitary equation, to the solver, in a 
program?
manually.
   
Hmm, does the HP48 have a command SOLVE, that works like the HP49G
command? I only know of the programmable commands QUAD, ISOL and ROOT
on the HP48.
Usage:
2:   equation     (Polynomial of degree 2 or less)
1:   variable
QUAD
2:   equation     (Variable to solve for has to appear only once)
1:   variable
ISOL
3:   equation     (Anything goes)
2:   variable
1:   initial guess (real or list of reals)
ROOT
ROOT is only for numeric solutions.
 Yes... with Erable. I always say the same: MK, Erable, Alg48 are must
 
Well, if your SOLVE works like mine, then you should enter the
equation, enter the variable to solve for, and then press SOLVE. You
don't need to stor in EQ first.
  
On an Hp48 I can think of three possible cases. First, finding the
real root(s) x(if any) of a general equation f(x)one at a time; second
the roots (including complex) of a polynomial with constant
coefficients and third the roots of a quadratic equation. An example
for each case is shown below:
 I'm learning to calculate laplace/inverse laplace transforms and as usual 

my
textbooks don't give answers to the exercises, so this is when the HP kicks
in. But there are strange things happening...
SIN(X)*Heaviside(-X) LAP    =    1/(SQ(X)+1)
should be 0 IMO: Heaviside(-X) = 0 for every X in the integration interval
[0-,+inf] and this pulls the whole integrand to zero (Mupad agrees with me
on the result).
X*e^(-X)*Heaviside(-X) LAP  =    1/SQ(X+1)
should be 0 (my opinion and Mupad's)
Heaviside(X+2) LAP          =    INV(X)*e^(2*X)
should be INV(X) (according to Mupad)
Delta(X+2) LAP              =    e^(2*X)
should be 0 (according to Mupad)
converts
it into a editable string. Finally, when finished, ASM compiles the
string.
I need this two commands for the 49
   They are in the built-in library 256. Just type
256 ATTACH
BTW there's an ASM2 somewhere (library 257?) which automatically calls
ER if errors have been found.
 
OT49 by Wolfgang automatically attaches it, I suggest you install it on
your 49. It does many useful things in the lands of black magic ;-)
  
if you have the extable, (needed for sysrpl compilation), it also attaches
it.
if I remember well, there is also a flag that if set automaticaly attaches
the lib
 :
Never heard or read on a System flag automatically
attaching lib 256 if set.
 
Well, you should have: flag -86, it's in the Header.h file. If set, lib 256
will attach itself automatically
  
Why you guys don't post the complete list of all hidden flags of Hp49 
system? 
  I only want compile a 49 prg in Emu for my 48GX
  I have installed on my HP-49 library for HP48 and now
my HP49 freeze on start up. I can not run my hp. ON-F1-F6 don't help.
What is command ON-BACKSPACE or something like that to skip start up
and then you can delete corrupted library?
 
 
You just turn it ON and then hold down the backspace key until the boot
process has finished.
  
   
Well, is that simple i have the james D. Purdy's eqlib49 library of
equations, and when i get the formula i only need to press SVLR on
that library and presto! i got the variables name in soft menus, but
when i try SOLVER directly from the stack i get nothing, the only way
to resolve a formula is to get torought the menu under the N.Solver
(Shift + number 7?) and then it took me  the solver screen, thougt i
still have to specifi which equation solve...   so do i have to learn
programming or what?  i am in RPN with ROM 19-6...
 
 
it's not as tidy or possibly as quick as it could be.
%%HP: T(3)A(D)F(.);
@ CSLVR is an extension of J.Myers CSLVR using utilities from
@ Wolfgangs OT49 to extend it's capablities and flexibility.
@ - Equation list to be solved should be on stack in format suitable for 
the
@   HP48 Numeric Solver.
@ - It finds all variables including variables within referenced equations.
@ - If a variable list isincluded it preserves the order of the original
list.
@ - If Programs are included in the variable list they are collected into a
Menu.
@     The programs can be used to set options or any other purpose ie 
preset
@     variables. If a Setup program exists it is run before starting.
@ - The Menu also incudes a general equation and variable viewer and option
@     to purge the variables.
@ Note: '$VMempPrg' is used to temporarly store the menu structure.  It can
be
@ deleted after finishing with CSLVR.  I tried to avoid using this 
temporary
@ variable but everything I tried was signifigantly slower.
@ Stephen Nash  ssnash.nospam@bigpoond.net.au
<<
@ Equation to solve is on stack
  DUP STEQ
@ Generate list of variables VL
@ Equation to solve is on stack
  END
  <<
@ Split list into list of variables and program selections
    { } VL 1.
    <<
      IF DUP TYPE 5. ==
      ELSE +
      END
    VP REVLIST DUP 'VP' STO
@ Generate Menu structure
@ VP is on stack
      { PURGE Variables
    << { } RCEQ 2. GET 1.
      <<
        IF DUP TYPE 5. =/
        THEN +
        ELSE DROP
        END
'$VMempPrg' STO
@ Generate EQ list format for Solver.
    RCEQ { } + 1. GET
    { { Menu
    << Menu $VMempPrg 1.
      IF CHOOSE
      THEN EVAL
      END
    DUP STEQ
@ Run any setup program
      IF 2. ==
      THEN SWAP
        IF Setup ==
        THEN EVAL
        ELSE DROP
        END
      END
    END
@ Start Solver
    1. GET 30. TMENU
 
 
 
I am trying to learn how to program my hp49, but I am not sure were to
get started.  I learned at hpcalc.org that there are four different
ways to write programs: hp basic, user rpl, system rpl, and assembly. 
I found online explanatins of system rpl and assembly and they both
looked too complacated.
On the internet it said that user rpl was faster than basic so I
thought that I wanted to learn user rpl.  I read over the programing
method in the HP 49 manual (assuming it was the basic language?) and
then got on the interenet to learn user rpl.  I could find no
explanations of user rpl for the HP49 so I looked at the one for HP48.
 The HP 48 user rpl explanation looked similar to  what I read in the
HP49 manual.
What is the difference between HP basic and user rpl?
Which language is in the HP 49 manual, user rpl or hp basic?
 ..
User RPL is the language that was built-in to the calcs since the
HP28. Using this to program, you work exactly like you would work
manually with the stack. For example to add two numbers you type the
first, press [ENTER], then type the second, and then press [+]. The
arguments for commands must be on the stack before the commands
execute.
On the other hand HPBasic is (almost) programming by writing the
command first and then its arguments in parentheses, like
GET({1,2,3},2). Exceptions are for example the infix mathematic
functions. You don't write +(1,2) but 1+2.
 
 
The difference between basic and rpl?
One is a language.
The other is  a egaugnal
 
 other, a language)
   
 
The HP-BASIC in the 49G is more of an algebraiced
version of the UserRPL language.
 .
  
I have a directory containing a equation.
s/SQR(n) with the name SV
In RPN I type SV to recall it to the stack
now I enter
5 s STO
9 n SOT
and RS ENTER it to eval it.
Now I want to do the same calculation again with different values.
so I recall it again.
Type
9 s STO and here I get the error: BAD ARGUMENT TYPE.
If I delete the variable s from the directory then I can do the STO
command.
Is there no command to replace the value? Or should I write a program
to delete all values from the directory?
What would you do?
 
  
Unless you have deliberately set flag -55 to disable
last argument saving the arguments of the command which
errored are always left on the stack; in this case,
you should have found whatever had last been stored into 's'
on the stack, rather than the name 's', and that is why
the STO command objected -- you can not store 9 into 5,
for example, because 5 is not a variable name.
You need to put 'quotes' around a variable name,
to indicate that you don't want the calc to try to
substitute the contents of the variable instead
(or to run the program, if that's what's stored in the variable).
Actually, it's much easier to use the keyboard shift keys
when you want to store into a variable; when the VAR[iables]
menu is displayed, the left and right shift keys act like
STO and RCL if pressed just before pressing the menu key;
this is one of the deep, dark secrets that HP
didn't want you to know before buying the calc (or even after),
but here we unmask all :)
Just be sure that you release the shift key before you press the
menu key, because holding down the left-shift key as you press
a top-row key on the 49G would otherwise invoke those other
graphing functions that are printed above the keys.
   
 
try: 9 's' STO, this should work (if you type just s, it will try to 
evaluate 
the variable).
 
 
Just wondering how to connect my 49g to the PC.
I am using the hpcomm 30r4 software but can't seem to connect to the calc.
The serial cable should be alright because it works with the 39g.
The calc is an ID946XXXX.
Basically it justs sits with 'Awaiting server cmd' and eventually times 
out.
  
 
I don't know why but I have to be in the Home directory and the stack
cleared to reliably connect to the PC.
 
  
    
 
1st. - Where could I find the 'assembly instruction set' for the Saturn
processor of the HP49G;
Are there any official books or documentation regarding it?
2nd. - What are the impressions of people using the different programming
methods, and the straight differences between them?
(assembly, hp basic, user rpl and system rpl)
3rd. - Besides the documentation reported by greatunpublished.com books 
by Gilberto Urroz, does anyone knows any other publications on the HP49G?
 
 
You can get entries.srt from hpcalc.org. It is a 48 document but describes
the 49 well too.
Carsten Dominik's Entry database should help you as well
http://zon.astro.uva.nl/~dominik/hpcalc/entries
 Also get my HP49G Assembly Language Examples
documents
Assembly rules speed-wise in specific problems (EG games, string processing
etc)
SRPL is used mostly for more general programs which need speed. it is 
usually
smaller than ASM or URPL.
URPL useful for programming small programs that are not speed critical, you
can't crash
the calc with it.
BASIC - dunno never used it as it is for Algebraic mode
You can find all the documentation you need (and more) on hpcalc.org
 G. Urroz's books are very thorough and guide you through the use of the
calculator in detail, including the advanced functions.  There are lots of
worked examples based around engineering problems.  The two volumes provide
about 800 pages of documentation.  I have them both, and have found them
useful.
One tip...  Pay and download the first volume.  If you like it, get the
second and order yourself hard copies of each immediately.  Great
Unpublished should let you order the hard copy at only the incremental 
price
(about $5 per volume).  If you wait too long (end of the month) they won't
do this for you.  I lost out and decided to print my own copies.
 
 
    Mr. Gilberto Urroz has written some amazingly useful books that do just
what you want.  To find them, go to
http://www.greatunpublished.com/bookstore/bookstore.html
Science and Engineering Mathematics with the HP 49 G - Volume 1
and
Science and Engineering Mathematics with the HP 49 G - Volume 2
These are what the HP manuals should have been!
 
  
your reccomendations and ordered both volumes now.  they may be the only 
books that ever get written for the calculator.  it sounds like everyone 
loved them.  I appreciate the input.
   
  
I also had one other question.  I own the 49g but it has that wierd 
serial cable that seems to only allow me to connect to a PC or an old 
fashioned mac with a serial port and some funky adapter.  I have a Mac 
G4 desktop without a serial port.  Is there a way yet to connect the 49g 
to a Mac via USB so I can download the latest stuff and install it onto 
the calculator?  Can I use USB on a PC yet?
If not, are they ever planning on getting this to the users, since 
serial is sort of going away?
   
   
     Probably your best bet would be to purchase a serial-to-USB adapter.
Belkin has some very nice ones, such as this one:
http://catalog.belkin.com/IWCatProductPage.process?Merchant_Id=1&Product_Id=

 
    I'll eat my hat if HP releases a USB calc-to-pc cable.
 
  
  
http://catalog.belkin.com/IWCatProductPage.process?Merchant_Id=1&Product_Id=

 
  
the HP49 was build using a relatively old CPU, and has the RS232 build in
the die, which makes is easy to support. For some reasons that I do not 
want
to go in, we could not redesign or change the CPU. Therefore, the only way
for the 49 tu support USB would have been to have an external USB module.
and these things are not cheap, nor are they simple to drive. the HP49
software would have had a hard time handeling USB requests especially as 
the
OS is not design for multi tasking and the CPU is really slow.
  
 
For those of us that aren't familiar with it, could you write a brief
description of what the advances in the Xpander were?
 
  
  
Your PC has an IrDA interface, the HP48 has an IR xmitter. Unless you
develop the IrDA stack for your HP48, you won't be able to make both
communicate by IR.
 
  
I have become dependent on my HP48 for my Accounting and Statistics
classes and want to backup my equations. When I click the archive
tool, the instructions say to press the OK icon followed by the EVAL
on the HP48. As soon as I press the OK icon, things go back and forth
If I press EVAL, I get a time out message on the HP48. Any ideas?
  
 
It sounds like your using the PC connectivy App to Backup your calculater.
The procedure you described works perfectly for my HP49. The << ARCHIVE
on the Calc when you press EVAL.  The fact that you see this instruction
appear on the calc indicates that the comms are working.  On my HP49 
nothing
much seems to happen after that but on the PC you can see the bytes being
exchanged.  It takes some time for it to complete.  You end up with a file
on your PC like 01-Apr-2002.hpbackup.  A handy thing about this file is
you can drag and drop it to EMU48 and use RESTORE to synchronise EMU48's
contents to your calcs.
   I'd like help you but I don't understand When I click the archive 
tool...
 
 
When I use the HP software on my PC, there is a command called
archive. I presume it will make a backup copy of my equations. I have
had a few lockups. The window that opens following that command shows
a file name and instructions on how to archive.
 
 
Here are some useful links for your HP49G:
http://www.hpcalc.org/hp49/science/physics/inerc12.zip
http://www.hpcalc.org/hp49/science/civil/inerci49.zip
http://www.hpcalc.org/hp49/science/civil/seccion.zip
If you can't find the HP49 program you need, try with these ones for the
HP48: [Emu, get a 48, port them to 49 or hire someone to do the conversion]
http://www.hpcalc.org/hp48/science/physics/inercia.zip
http://www.hpcalc.org/hp48/science/physics/inertia.zip
http://www.hpcalc.org/hp48/science/physics/perfiles.zip
http://www.hpcalc.org/hp48/science/physics/sectv10.zip
http://www.hpcalc.org/hp48/science/physics/sec_prop.zip
http://www.hpcalc.org/hp48/science/civil/seccion.zip
http://www.hpcalc.org/hp48/science/civil/seccion1.zip
http://www.hpcalc.org/hp48/science/civil/seccion2.zip
http://www.hpcalc.org/hp48/science/civil/sepd50.zip
http://www.hpcalc.org/hp48/math/misc/sectn11.zip
Perhaps you are lucky enough to own an HP-28S:
http://www.hpcalc.org/hp28/apps/propsec.zip
Butcher, published in Datafile V20N3, which you can order as a back issue.
( http://www.hpcc.org/datafilev20.html#V20N3 )
 
-
  
I like to work in engineering notation, but my 48GX seems to have other 
ideas...I put it in that mode, but it somehow switches back to standard 
notation.  Has anyone else had this problem and is there a fix for it, other 

  
 
It is probably a program you have that resets the STD mode. Proper programs
save flags before they change anything and reset them upon exit.
If you can find out which program changes it you could add the flagsaving
yourself
with
<< RCLF 'flags' STO
(insert program name here)
 
  
Ok, I entered this thing into my calc with the name YONPY where you said to 

put 
the program name and saved it as 'RESFL'.  I put it in the YONPY directory, 

and 
tried it to see what it would do.  It didn't work.  As a matter of fact, it 

started YONPY.  Should I have put the YONPY in tick marks or something?  
I've 
looked in the User's Guide, but it's not terribly helpful with this.
  
 
It is SUPPOSED to stary YONPY :)
But after RESFL runs, your flags will be restored and ENG will still be
enabled, unlike running YONPY! Its like magic!
 
 
Yes, but it didn't actually WORK that way.  When I quit YONPY, it still has 

changed the numbers back to STD.  Argh!
 
 
Ok then we try the less elegant, but rigorous way!
I think that YONPY exits with an error, this would
explain why the flags are not restored.
Here is a new RESFL
<< IFERR YONPY 2 ENG
   THEN 2 ENG END
This resets to engeneering mode (with 2 decimals)
whether YONPY errors or not.
  
  
The only program I have on it is YONPY.  And it switches back to STD even if 

I 
haven't been using YONPY.  Today, I was in class and I entered a list of 
numbers, and they all displayed in ENG, except the last one.  When I went to 

modes to check, sure enough, it was set to STD, and when I set it back 
to 
ENG, everything displayed like I like it.
Where do I put the << RCLF 'flags' STO (insert program name here)'flags' DUP 

all?  
I have absolutely NO programming experience with this calculator.
 
 
Ack! No programming experience! Then what does YONPY do!!
If you don't use it, remove it and see if it helps.  Also note if you press 

+/-
while in MODE screen it changes to STD if you were in ENG before
culprit. 
 I use it frequently, and don't want to take it off the one I use most right 

now, so I guess I'll have to live with it changing number formats until I 

can 
figure out how to put that thing Peter Geelhoed gave me into YONPY.  Unless, 

  
 
I just had the same thing happen to me last night.
Try entering the number as a real or putting the calc in Approx mode.
That should do the trick.
 
   
But, Peter, the HP48GX doesn't have Exact mode. It is always in
Approx mode. Or did I forget so many things?  .
 
I guess I was thinking of the 49G.
The 48 doesn't have an integer data type anyways so I must have been
thinking of a different animal all together.
 
Ok, I just looked up how to do that in the User's Guide, and no dice.  I 
don't 
have the AUR, and the index of the User's Guide doesn't have approximation, 

Approx, or anything like it.  How do I do that?
Also, I was reading about complex numbers in the UG, and it got me thinking 

that if I had an operation that produced a complex number (which 
occasionally 
happens) it might be forcing a number format change.  So I did an operation 

which had a complex answer (arcsin(5)), and checked the number format.  It's 

still in engineering mode.
What did you do that may have made your calc switch on you?  And how do I go 

about putting it into Approx mode?  Please bear with me.  I've had this 
calculator for almost 3 years, but have only recently begun to explore what 

it 
can really do.  I don't know that I'll end up programming, but I do want to 

be 
able to learn how to use programs that are already out there, and the 
capabilities that are already present in the calc.
   
 
  
The previous guy was incorrectly talking about the HP49G, not the HP48GX.
Normally, the engineering mode should change at all. What may happen is 
that
you have a library installed that reset some flags (usually this happens
during a reboot).
But as far as I know, the HP48GX, by itself will never reset the display
settings, except if you crash it
 
 
The only program I have on it is YONPY.  And it switches back to STD even if 

I 
haven't been using YONPY.  Today, I was in class and I entered a list of 
numbers, and they all displayed in ENG, except the last one.  When I went to 

modes to check, sure enough, it was set to STD, and when I set it back 
to 
ENG, everything displayed like I like it, and all I had done was enter 
numbers.  
I hadn't been using YONPY for days.  So far as I know, the only libraries I 

have installed would have come with the calculator.  I haven't added 
anything 
to it but YONPY.  Could that be the culprit?  And I've never crashed it.  It 

does this in the middle of exams, which is annoying, even though it's not 
critical.  Just takes time I could be using to do the exam...
  
  
to explore what it
to be
If you want explore this calc (the best in the world ;-) I strongly
recommend you use MetaKernel 2.3 and Erable 3.2. You will find them in
good keyboard.
 
   
If you look at the advanced user guide thread you will find a link to
http://web.archive.org/web/20000815225830/www.hp.com/calculators/techsupport

/graphing/49g_userguide/
you can download a german pdf manual from there, I checked it, it works!
  
   ;
  
Try www.hp.com or order the manuals for free at german calculator hotline,
 
 
i take a look on hp.com and i only find an english pdf version
can you search the number for me ??
 
 
http://web.archive.org/web/20000815225830/www.hp.com/calculators/techsupport

/graphing/49g_userguide/
  
For what it's worth, I'm using my HP 48 G/GX manuals to supplement my HP 
49G
manuals.  So far, so good!
In so doing, I've done a few 48-49 performance comparisons.  One big
surprise was how much longer the 49 takes to recall a previously graphed
function.  Using PICTURE on my 48 GX, the graph displayed quite quickly.
However, the graph and, in particular, the menu keys took very long to
display using the GRAPH command on the 49G.
http://web.archive.org/web/20000815225830/www.hp.com/calculators/techsupport

/graphing/49g_userguide/
Take a look on www.hp.com. There are all manuals available.
Nope, HP seems to have dropped the manuals from their site entirely.
Probably the first step to them abandoning that stillborn child of
theirs...
Does anybody know if they are still manufacturing 49Gs? My guess is,
with Office Depot and others having removed it from their programs
(and sometimes selling it for 20 bucks or so at their garbage tables),
that they are just selling remaining stock, and that will be it. At
least that would be the usual procedure for a failed product.
Have you tried www.samsoncables.com?  My dad got me the ME Apps card, used, 

How can I change the number of decimals the calculator uses to a
number bigger than 11.
11 is not close enough to the number of decimals I need.
developed a library called LongFloat which can calculate to virtually
infinite precision. See hpcalc.org.
There's also a built-in 15 digit data type (long real), but you can
only use it from SysRPL or ML.
Hmm, the 49G doesn't provide more than 12-digit precision for the
normal user. And when you say not even close enough, I think that
for your need even the long floating libs available will not suffice.
Is a disequation something like a!=b, where != stands for not
equal?
In this case, couldn't we replace != with =, SOLVE, and replace = with
!= in the solutions?
Example: x^2-x+1!=0.
SOLVE { x=blabla ...}
SOLVE can give you solutions, as long as there are no other symbolic
names than the variable variable, which you solve for.
    A few minutes ago I started to play with a little code that I have
started
today (at morning). 
    I has saved my source code in port 2 and thought that everything
was ok, (so... if I make some shit with my code, the backup is
done...).
    I started to test the code and I got a TTRM, I pressed YES and the
system has started.
    When I look in port 2 to find my code, surprise! No source code
there!!!
    Now I've lost my code (or there's any way to use the CQIF to
search for string inside FlashROM (What's the port?)) and I'm very
confused about the Flash
Rom's security. 
    I have another question... There's a limit of files that the port
2 can support? Now I have 63 Files ...
 
I remember something... I tryed to make another backup, but a Low Batt
screen was shown. Well, is there any possibility to recover my code?
 
 
Well, the Flash is suposed to be really secured. It takes a lot to erase
things from the flash, and it certainly can not happen by luck (or lake
off), or even because of a faulty program, so the problem is not there.
now; if, as you said, you have a low bat message, it is well possible that
your data has never been stored in flash! and in this case, you are kind of
out of luck.
In any case, you can always consider storing things in port 1, it does not
need the same amount of power, and is also quite secured.
There is no limit to the number of files in flash besides the memory and
your ability to manage the data.
 
  
Be warned! It is not nearly as secure as the flash is. I only recently
lost a part of my Port 1 data    
due to some weird bug. And this was not the first time.
On the other hand I have not successfully gotten rid of my flash data
yet ;-) I once had a corrupted bank, but that seemed to be due to the
low batteries or - maybe - a bug in the ROM, because back then I didn't
know SysRPL or ML.
 
It happens that these last days (after updating my ROM to 1.196), I'd
had problems attaching classic libs like the Organizer, along
others.
I would like to translate many libs and directories programmed for
the 48 series, like e.g.:
1) the physis series from Le Mafioso ID Lib # 98X
hp49/science/physics/physique.zip:BTS CIRA & STL 
2) ERROR (Uncertainties by Jay Kyle)
...  a program that adds, subtracts, multiplies, raises to a power,
etc uncertainties or in other words numbers such as 2 .96 0.5.  ...
Item: 237 by _jk at hpcvbbs.UUCP
Item: 239 by _jk at hpcvbbs.UUCP
 
Keyw: x.96dx, error, uncertainties [that's x-plus-or-minus-dx, you
 
This is an updated version of my uncertainties program, ( v 1.2 )
that now works with unit objects as well as real numbers. The main
description is in message #237. I have renamed it ERROR, short for
error analysis.
 
No standard but several approaches.
* Kalinowski's hpconv and hpdis are worth a try:
  http://www.hpcalc.org/hp48/pc/programming/hpconv14.zip
* The manual method: Jazz.
keystroke
  detection routines (and menus, flags ...), adapt to 49 enhancements: 
fonts,
  integers ... (this can be VERY hard), look for unsupported entries (a 
true
  art form) or avoid them completely. Finally, ASS (or MASD), backup to PC,
  CRLIB and debug.
* Contact the author.
  Who knows? The version for the HP49 of my LKT library was in response to 

a
  petition from ONE user.
* Ask for help here.
  Even for your most specific questions. Perhaps, you can find someone
  to convert your programs for free: Many people are willing to show how
  much they know and that could be a good way.
  Gather all of the libraries and programs you wish to be converted
  and look for somebody to do the job. Of course, that guy should be
  proficient with both machines, honest, with a reputation to maintain ...
  ( http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/adforum.cgi?read=7067 

   
  
http://www.microsoftmobilitypress.com/site/extras/pics/normal/oem/hpcommanch

  
  
  
I just got an HP 49G and I was wondering why they ship with a serial cable
converter for the end that plugs into the unit itself. What is the other 
end
for and why would I be using the serial converter without the calculator in
the loop?
Finally, is there a converter for the other end so I can skip by buying the
PC connectivity kit, or is that a you have to situation?
 
  
So you can connect the HP49G to another HP49G or (with the converter) to a
HP48.
the
I don't don't think so but you can build your own PC con. kit. Have a look
at www.hpcalc.org.
  
  
My old trusty, but very lightly used 48SX developed a problem out of the 
blue
the other day. The ON button stopped working properly. I have to push it 
much
harder and on the left side of the button. Is there an easy way to open the
unit and clean the contacts (assuming it's of that type of design). I've
noticed this on other keys as welI so I assumed this problem was due to 
some
type of oxidation since I only use this a few times a year.
  
Of course you can open your calc and try to clean the contacts. But be
warned this is no easy operation! You can find some instructions how to 
open
the unit on www.hpcalc.org. (Read them before you try to open the calc;-))
 
How much time should this message long ? (after crash)
I downloaded eqlibhp49 (along with stat49pr.lib) and installed them
in Port 1.- After attached, I wanted to try th first on out and the
49G crashed. The .87Recovering Memory message had been going for 
more
than half an hour now and I don't know when should I stop it or if I
should do it at all.
I pressume there is a relation with the stuff loaded before, but there
was less than 300 k.
by the way, do anyone had any trouble with this specific lib
(eqlibhp49.lib) ?
I haven't got any trouble when I updated my rom to 1.196, but since
then I'd been experienced many situations when trying to download new
applications. The PINIT routine didn't help, as Tim Wessman put it out
in his 20 Essential Things to know about the HP-49G Guide for the
hopelessly confused (like myself)
Am I doing something wrong here ? 
 
 
About the Recovering memory message, my experience is that it doesn't
recover anything when it lasts so long. I once or twice had the
patience to wait until it comes to an end (or I forgot it because I
had something else to do ;-) ) and after the end of the procedure
nothing was recovered. (Only dome directories named dir.01 , dir.02
that didn't contain anything.) I backup all contents on port 2 from
time to time, and when I have a crash (seldom in the last time) I
answer No to the question for recovery and then I restore the
backup.
  
  
  
Have a look at my site (http://members.iinet.net.au/~ccroft) if you want 
info on what's available for the 40G.  My site covers the 38G & 39G but 
the 40G is the same as the 39G except that the 39G has infra-red and no 
CAS while the 40G has CAS and no infra-red.
The manual for the 40G is excellent.  The only thing that it doesn't 
cover is the CAS and you can download the CAS manual from my site.
If you want to try before you buy then you can download an emulator kit 
which will let you emulate the 38G, 39G and 40G from my site.
The CAS system on the 40G is VERY powerful and is basically the 49G 
system translated.  However the 40G is essentially a machine designed 
for high school students or uni students not requiring really heavy 
mathematical use - ie not engineering or physics or maths.
  
how am I suposed to work this out, as you probably have guessed I'am a
newbee to this.
I have a hp49g, and I would like to update to ROM version 1.18 (I bought my
calc about 2.5 - 3 years ago, and it hasn't been updatet at all).
I have downloaded the update file (HP49g118.zip) to my laptop, and conected
the cable between the laptop and my hp49g. AND HERE IS THE PROBLEM: what do
 
 
First of all you schould use the newest (beta) ROM 1-19-6 instead of 1.18.
Although it's a beta ROM it is more stable than version 1.18. The
instructions for updating the HP49G with a new ROM should be in the 
ZIP-file
too. You can find the 1.19-6 ROM on http://www.hpcalc.org. Read the 
attached
instructions (you have to unzip the ZIP-file of course;-)) for updating 
your
calc.
  
 I just uploaded the update to the Complex Numbers Marathon. The new
document contains the first part with some corrections and the new
second part. The second part is about 30 pages. Some of the contents
of the second part:
About complex numbers in the form of algebraic objects, X+Y*i
How to use flag -27 to display such complex numbers as (X,Y), why some
of them are still displayed as X+i*Y and programs that convert such
complex numbers so that they can be displayed as (X,Y).
Behavior of commands like EXPAND etc. applied on complex numbers in
algebraic object form. Usage of TEXPAND TSIMP in such cases.
Entering complex numbers as algebraic objects in polar form with or
without units of angle, and a small idea for yet another shell.
Factoring complex numbers in the form of algebraic objects. The HP49G
doesn't factor for example 10-5*i. (It should return (2-i)^2*(2+i),
but it errors out with Bad Argument Value. While we are at it: try
to factor for example 39+26*i or 68-17*i with the built-in command
FACTOR and be embarassed that it gives an error. FACTORS doesn't error
but gives you *erroneous* results! ) How to use PA2B2 in a program to
factor also such numbers. Gaussian integers and Gaussian primes.
Programs for completely factoring quantities like (16-23*i)*(25+
i*SQRT(3)), which the built-in command leaves unchanged, while it can
be factorized to i *(2-i)*(11-6*i)*(-25*i+SQRT(3))
Some additional explanations for commands like DOSUBS, MAP etc.
An alternative formula for calculating the angles of complex
quantities in the interval from -pi to pi. (Alban Schann's idea.)
As always we try to keep band width usage of Eric's site low. So if
you want the document, then you can of course also mail me and I'll
send it to you as soon as possible.