HP Forums

Full Version: HP32SII Solver Bug
You're currently viewing a stripped down version of our content. View the full version with proper formatting.

Hello, try this, I think it is a bug, but not(?!?!?) important - maybe annoying:

===========================================================

CASE 1:

Type the following equations into the calc:
A+B=C (eq1)
D+E=C (eq2)

Press [ON] to exit EQN mode
Press [ON]+[C] to soft-resetting calculator (try without it - I used, for safety)
Press [CLVARS] to clear variables
Go to EQN list, and search (eq1)
Press [SOLVE] C
On A?0 prompt type 2 [R/S]
On B?0 prompt press [ON] to cancel SOLVE
Go to EQN list, and search (eq2)
Press [ENTER]
On D?0 prompt type 3 [R/S]
On E?0 prompt type 7 [R/S]

The EQN anunciator will flashing and on the screen seems
C?0 or C?1, and you will get C=10.

===========================================================

CASE 2/a:

Type the following equations into the calc:
A+B=C (eq1)
D+E=F (eq2) <-- CHANGE IS HERE!!!

Press [ON] to exit EQN mode
Press [ON]+[C] to soft-resetting calculator (try without it - I used, for safety)
Press [CLVARS] to clear variables
Go to EQN list, and search (eq1)
Press [SOLVE] C
On A?0 prompt type 2 [R/S]
On B?0 prompt press [ON] to cancel SOLVE
Go to EQN list, and search (eq2)
Press [ENTER]
On D?0 prompt type 3 [R/S]
On E?0 prompt type 7 [R/S]
On F?0 prompt type 10 [R/S]

You will get C=1 or C=0.
(If you after CLVARS deleted the stack, you will get always C=1.)

===========================================================

CASE 2/b:

Type the following equations into the calc:
A+B=C (eq1)
D+E=F (eq2)

Press [ON] to exit EQN mode
Press [ON]+[C] to soft-resetting calculator (try without it - I used, for safety)
Press [CLVARS] to clear variables
Go to EQN list, and search (eq1)
Press [SOLVE] C
On A?0 prompt type 2 [R/S]
On B?0 prompt press [ON] to cancel SOLVE
Go to EQN list, and search (eq2)
Press [ENTER]
On D?0 prompt type 3 [R/S]
On E?0 prompt type 7 [R/S]
On F?0 prompt type 11 [R/S] <-- CHANGE IS HERE!!!

The EQN anunciator will flashing, and you will get NO ROOT FND error.

===========================================================

For a good reproduction try this with empty stack
(clear the stack after CLVARS).

Any comments??? HP33S users can it to reproduce???

Csaba

Edited: 15 July 2004, 7:06 a.m.

...can anybody tell me, what is the problem with my postings? Or just only me is interested about this things? ;) Or I wrote wrong place here about this? (I hope it is not seems too "agressive"... ;) )

Csaba

Hi Tizedes,


I am interested---just have not had time to try yet!


Regards,

Bill

Bill posted:

"Hi Tizedes,"

Bill, his first name is "Csaba", not "Tizedes". "Tizedes" is his last name.

He's already angry at not getting any comments to his (very interesting) post, so don't make it any worse by addressing him by the wrong name ! :-)

By the way, Csaba, please excuse my not commenting your post but I do not have an HP32SII at hand to try, sorry.

Best regards from V.

.

Hi Csaba,


Interesting you shold post this observation, as I actually did notice this issue just a few days previous.

Basically, the "feature" or bug is that the solver does not cancel when you get out of the middle of a solve, when in an "equation". The "1" you find in the C register is the first guess that the solver has tried.

If you run through a complete solve, everything works as expected.

you know, this strange behavior may actually be useful----have do do some experimenting....

regards,

Bill

Wow... that is most perplexing. I recieved the exact same results on my 32sii, but got slightly different results on my 33s.

On the 33s I never recieved a "C=" for any of the cases. For case 2/b, I recieved an 11 and -1 on my stack. For 2/a, I recieved a 10 0 on my stack, and for case 1, it prompted me for c, unlike on the 32sii.

Apparently there are some differences between the 32sii and the 33s.

Hmm.... most perplexing.
Good catch,

Ben

"...the solver does not cancel...The "1" you find in the C register is the first guess that the solver has tried."

Exactly this is my opinion, too! Thanks your answer!

Csaba

Hello Ben,

this bug is cureable, if you press an [RTN] after [ON/CANCEL] in solving procedure. After this everything will works like as manual wrote.

So, SOLVE is take one level of subroutines...(?)

Csaba

Hi Ben And Csaba,

For my 33s, I find that the results are entirely "Bug-free" and that the solver does in fact CANCEL when hitting C.

Results for 2/a:

t  3
z 7
y 10
x 0

This is entirely as should be. <ENTER> evaluates by putting a MINUS sign in place of the equals. Therefore, 3+7-10 results in a 0.

the 3,7,10 in the stack are from the prompting. Just as with the 32sii, a prompt for a variable will also put it on the stack.

results for 2/b:

 [pre]
t 3
z 7
y 11
x -1

also follow the rules: 3+7-11 gives a result of -1.

For 1, again, we have results which are consistent with the documented properties:
after cancelling the solve of eq1, and ENTERevaluating eq2, we put in 3, R/s, 7, R/s, and then we are prompted for "C?"( whatever value was in C will be on the x-line) and again, it will be evaluated.

So, in my opinion, form the reuslts of the tests, on my machine, the 33s has no bug.

The machine I have tested is from waht I call the "second" batch--an Amazon.com CNA41505109 unit purchased after April.

the Walmart units released initially before April, have a different ALG X<>Y implementation, and may hacve other differences, too!.

Which is yours, Ben?

Edited: 27 July 2004, 12:10 p.m.

I have Walmart version. It wasn't preordered, but after there were reports of Walmart shipping, I camped the site and as soon as they became available (2 day later or so) I ordered one. It is serial: CN40401698

-Ben Salinas