HP PRIME solves a equation wrongly

[Williams369]

We write in the CAS

"csolve(((x+400)*0.8/(π*0.04^4/32)) = (0.4*((x+400)*0.8/(π*0.04^4/32)+x*0.5/(0.141*(20*√2)^4*10^-12))),x)" RESULT −1661.90655858

"zeros(((x+400)*0.8/(π*0.04^4/32)) = (0.4*((x+400)*0.8/(π*0.04^4/32)+x*0.5/(0.141*(20*√2)^4*10^-12))),x)" RESULT −1661.90655858


"fsolve(((x+400)*0.8/(π*0.04^4/32)) = (0.4*((x+400)*0.8/(π*0.04^4/32)+x*0.5/(0.141*(20*√2)^4*10^-12))),x)" RESULT 2492.85983787

Plugging "−1661.90655858" into the equation, the result is nonzero. It is a linear equation, how can it solve it wrong.

The correct answer is the one provided by fsolve. Why is this so? Why is HP Prime wrong when solving a simple equation of the first degree?

[Arno K]

Hi,
you may use csolve(equation,x,x>0), that calculates 2492.85983787.
Hope that helps
Arno

[Nigel (UK)]

(05-07-2023 01:23 AM)Williams369 Wrote:  We write in the CAS

"csolve(((x+400)*0.8/(π*0.04^4/32)) = (0.4*((x+400)*0.8/(π*0.04^4/32)+x*0.5/(0.141*(20*√2)^4*10^-12))),x)" RESULT −1661.90655858

"zeros(((x+400)*0.8/(π*0.04^4/32)) = (0.4*((x+400)*0.8/(π*0.04^4/32)+x*0.5/(0.141*(20*√2)^4*10^-12))),x)" RESULT −1661.90655858


"fsolve(((x+400)*0.8/(π*0.04^4/32)) = (0.4*((x+400)*0.8/(π*0.04^4/32)+x*0.5/(0.141*(20*√2)^4*10^-12))),x)" RESULT 2492.85983787

The correct answer is the one provided by fsolve. Why is this so? Why is HP Prime wrong when solving a simple equation of the first degree?

Using the web version of Xcas (solve(...,x)) gives the correct answer, whereas the Prime does not.

Nigel (UK)

[Wes Loewer]

(05-08-2023 02:00 PM)Arno K Wrote:  csolve(equation,x,x>0)

Is the use of that 3rd parameter documented anywhere?

[Arno K]

Nope, but I had something in mind from solve(), 3. Param can be taken to describe an interval and I simply tried this one.
Arno

[Williams369]

How can I know that the answer is greater than zero? It is also a first degree equation, there can only be one answer, and when replacing the answer it gave, it does not give zero, but another value, that is, it is wrong. Why is he wrong in a first degree equivalency.

[Williams369]

I have the physical HP prime, I can't use a cell phone in the exam. And I see that I will fail the exam if I use this calculator.

[phobos]

Always use exact values when in CAS. (Press the decimal to fraction key).

https://i.imgur.com/2jO5OIc.png

[Albert Chan]

(05-08-2023 02:00 PM)Arno K Wrote:  you may use csolve(equation,x,x>0), that calculates 2492.85983787.

Last argument, x>0, is not the reason for getting correct answer.
Besides, for complex number, x>0 is meaningless.

There is a bug, equation does not properly converted to numerical.
We can explicitly turn equation numerical.

Cas> m := approx(
(x+400)*0.8/(π*0.04^4/32) =
0.4*((x+400)*0.8/(π*0.04^4/32)+x*0.5/(0.141*(20*√2)^4*10^-12)))

(3183098.86184*(x+400.)) = (0.4*(3183098.86184*(x+400.)+5540780.14184*x))

Cas> solve(m)      → {2492.85983787}
Cas> csolve(m)     → {2492.85983787}
Cas> zeros(m)      → [2492.85983787]

[jonmoore]

(05-10-2023 01:29 AM)Albert Chan Wrote:  

(05-08-2023 02:00 PM)Arno K Wrote:  you may use csolve(equation,x,x>0), that calculates 2492.85983787.

Last argument, x>0, is not the reason for getting correct answer.
Besides, for complex number, x>0 is meaningless.

There is a bug, equation does not properly converted to numerical.
We can explicitly turn equation numerical.

Cas> m := approx(
(x+400)*0.8/(π*0.04^4/32) =
0.4*((x+400)*0.8/(π*0.04^4/32)+x*0.5/(0.141*(20*√2)^4*10^-12)))

(3183098.86184*(x+400.)) = (0.4*(3183098.86184*(x+400.)+5540780.14184*x))

Cas> solve(m)      → {2492.85983787}
Cas> csolve(m)     → {2492.85983787}
Cas> zeros(m)      → [2492.85983787]

Well answered as per usual Albert.

Ref the bug. I did a quick check on KhiCas for the the Ti Nspire, the Linux flavour of XCAS and PocketCAS on iOS and none of these showed evidence of the bug, so I suspect that the XCAS flavour in the Prime has missed a recent Giac service update. Below are the repository pages of both the general releases of Giac/XCAS and KhiCAS which detail recent updates:

https://www-fourier.univ-grenoble-alpes....c/?C=M;O=D
https://www-fourier.univ-grenoble-alpes....i/?C=M;O=D

Bernard's main installation page for the various flavors of XCAS/Giac contains no reference to the HP Prime at all.

https://www-fourier.univ-grenoble-alpes....install_en

Considering HP calculators were the first to feature the symbolic CAS engine (primarily) developed by Bernard Parisse and that the Prime remains the canonical calculator UX implementation, it's sad that the Prime isn't mentioned at all. On more practical terms, it also makes it difficult for Prime owners to check for specifics of the current CAS version included with the Prime. This information is only available via the About page available in the Prime's onboard help system. And this shows the CAS to be v1.5, which lags by a large margin behind the current CAS version - v1.9.0-49

It's highly possible that the onboard Prime help About page is badly outdated, but we can only make a judgement based on the available information.

[parisse]

Actually the Prime *is* mentionned on my webpage, it is the first item in the Calculators menu.

I hope that I will be able to update the CAS version on the Prime soon. But as you can guess, it's not like for KhiCAS on the Casio, TI Nspire or (old) Numworks where I have full control on updates.

[jonmoore]

(05-10-2023 10:32 AM)parisse Wrote:  Actually the Prime *is* mentionned on my webpage, it is the first item in the Calculators menu.

I hope that I will be able to update the CAS version on the Prime soon. But as you can guess, it's not like for KhiCAS on the Casio, TI Nspire or (old) Numworks where I have full control on updates.

Hi Bernard, I didn't intend to be critical of your efforts. I have nothing but admiration for your efforts to supply a free alternative to the main CAS vendors (Mathematica, Maple etc).

I am aware of the XCAS pages where the Prime and 48/49/50 heritage is mentioned. However, the best page to keep up with XCAS releases across all available platforms is https://www-fourier.univ-grenoble-alpes....install_en and the actual repository pages I linked to in my last post. I have all XCAS/Giac related pages bookmarked, inclusive of both English and original French language documentation pages.

The broad point I was making was that HP calculators are no longer the canonical implementation of XCAS on mobile platforms. The fault for this situation has nothing to do with your efforts. But the situation is nonetheless sad considering the roots of Erable on the 48.

I hadn't expected the (very welcome) update Moravia made to the HP Prime firmware. But now that they appear to show a commitment to the HP Prime in the immediate future, I'm hoping that they can co-ordinate with you to ensure that the CAS reflects recent developments to the XCAS project. I'm sure that this isn't always practically possible, but I'd hope that the Prime CAS is at the very least a match for KhiCAS on the TI Inspire.

[chromos]

(05-10-2023 12:34 PM)jonmoore Wrote:  ... I'm hoping that they [Moravia] can co-ordinate with you to ensure that the CAS reflects recent developments to the XCAS project. ...

+1