The Museum of HP Calculators

HP Forum Archive 21

[ Return to Index | Top of Index ]

[WP34s] Possible problem with matrix inversion
Message #1 Posted by Dieter on 5 Jan 2013, 10:01 a.m.

I just tried the matrix inversion command with a simple 4x4 matrix. Calling M-1 directly from the keyboard makes the calculator shut down (!). Sometimes the VERS screen is displayed instead.

However, if the same M-1 command is used from within a program, everything works fine:

  001 LBL A
  002 M-1
  003 END
Used this way, the command is executed as expected. The same applies to M-1 calls from the library, e.g. the matrix editor MED.

All this applies to Version 3.1 3325 as well as the new 3.2 3350 (updated a few minutes ago). So, is it the firmware or is there a problem with my hardware?

Dieter

      
Re: [WP34s] Possible problem with matrix inversion
Message #2 Posted by fhub on 5 Jan 2013, 10:22 a.m.,
in response to message #1 by Dieter

Quote:
I just tried the matrix inversion command with a simple 4x4 matrix. Calling M-1 directly from the keyboard makes the calculator shut down (!). Sometimes the VERS screen is displayed instead.
Have you tried it on the emulator, too?
I just did it (also 4x4) and had no problems.

Franz

            
Re: [WP34s] Possible problem with matrix inversion
Message #3 Posted by Dieter on 5 Jan 2013, 12:00 p.m.,
in response to message #2 by fhub

The emulator works fine. It's "the real thing" that behaves that way.

Dieter

      
Re: [WP34s] Possible problem with matrix inversion
Message #4 Posted by Walter B on 5 Jan 2013, 12:57 p.m.,
in response to message #1 by Dieter

Hmmh, I just tried with build 3335 on my "real thing" and everything worked fine.

My matrix (chose some arbitrary values without thinking much):

 1   2   3   4
 0   1   4   9
 7   6   5   4
 1   0  -1   1
How's your matrix looking?

d:-)

            
Re: [WP34s] Possible problem with matrix inversion
Message #5 Posted by Dieter on 5 Jan 2013, 2:11 p.m.,
in response to message #4 by Walter B

Quote:
How's your matrix looking?
It's looking great, of course. 8-)

Since the early Eightes I'm using a simple test matrix, taken from a book on numerical algorithms I bought back then:

      5   7   6   5                  68 -41 -17  10
      7  10   8   7                 -41  25  10  -6
A  =  6   8  10   9      Inverse =  -17  10   5  -3
      5   7   9  10                  10  -6  -3   2
This, as well as a 4th...6th order Hilbert matrix, usually is the first thing I try with new matrix software.

But I do not think the matrix itself is the reason for this strange behaviour. I fear something went wrong with the crystal/IR upgrade and it's actually a hardware problem. The same BTW happens with the LINEQS command. I just wonder why everything is fine as soon as the same functions are called from within a program. Strange.

Dieter

                  
Re: [WP34s] Possible problem with matrix inversion
Message #6 Posted by Walter B on 5 Jan 2013, 4:27 p.m.,
in response to message #5 by Dieter

I fear you're right. Your matrix inverts on my calculator just fine. I can't tell you, however, how the quartz / IR shall effect your calculations - maybe Marcus knows more about that particular case?

d;-)

                  
Re: [WP34s] Possible problem with matrix inversion
Message #7 Posted by Marcus von Cube, Germany on 5 Jan 2013, 4:28 p.m.,
in response to message #5 by Dieter

Dieter, I suspect your hardware is broken. What does BATT say? Can you switch to SLOW mode and try again?

                        
Re: [WP34s] Possible problem with matrix inversion
Message #8 Posted by Dieter on 5 Jan 2013, 5:33 p.m.,
in response to message #7 by Marcus von Cube, Germany

BATT returns 2,9 V, and after SLOW the same thing happens: M-1 turns the device off, and in some cases it displays the VERS message.

Dieter

Edited: 5 Jan 2013, 5:33 p.m.

                  
Re: [WP34s] Possible problem with matrix inversion
Message #9 Posted by Paul Dale on 5 Jan 2013, 8:02 p.m.,
in response to message #5 by Dieter

M-1 and LINEQU function pretty much the same underneath. They both perform a Doolittle LU decomposition with partial pivoting and then solve the system of equations (many times for M-1).

I've no idea what is causing your problem :-(

- Pauli

Edited: 5 Jan 2013, 8:03 p.m.

                  
Re: [WP34s] Possible problem with matrix inversion
Message #10 Posted by John B. Smitherman on 7 Jan 2013, 9:51 a.m.,
in response to message #5 by Dieter

Dieter, have you tried re-flashing the 34s software?

John

                        
Re: [WP34s] Possible problem with matrix inversion
Message #11 Posted by Dieter on 7 Jan 2013, 3:07 p.m.,
in response to message #10 by John B. Smitherman

Yes, of course a firmware update was the first thing I tried. But the problem showed up with version 3.1 3325 as well as 3.2 3350. It seems to be a hardware problem.

Dieter


[ Return to Index | Top of Index ]

Go back to the main exhibit hall