Hi everyone,

Thank you for this awesome product, we really like to work with it. Unfortunately, a colleague of mine is worried about the verification/certification of calculations in SMath, since it's open-source software.

Could anyone tell me more about how it is ensured that calculations in SMath are correct or where I can find documentation about this? The EULA does not include anything about this, I could not find the terms and conditions anywhere. Having worked on software development I could imagine that there is an elaborate testing suite for the programme and that it uses imported modules from verified repositories.

Thanks to anyone that can help!

Smath is a CAS, it sources functions from same
as any other CAS, i.e: from universal libraries.
No flaws from Smath solvers either.
Sanity some typical [ln(x), sin(x)...] 15 D vs Abramowitz.

Wrote

Hi everyone,

Thank you for this awesome product, we really like to work with it. Unfortunately, a colleague of mine is worried about the verification/certification of calculations in SMath, since it's open-source software.

Could anyone tell me more about how it is ensured that calculations in SMath are correct or where I can find documentation about this? The EULA does not include anything about this, I could not find the terms and conditions anywhere. Having worked on software development I could imagine that there is an elaborate testing suite for the programme and that it uses imported modules from verified repositories.

Thanks to anyone that can help!

First of all, SMath is not open source. It is free to use. Completely different concept.
Second, open source softwares are more easier to find and correct bugs/mistakes.
This is why merely all servers are linux/BSD (unix) based.
Open source software/systems are easier to repair and control.
I wish SMath was an open source program, but it is not and not will soon. BUT it is free.
I understand Andrey's concerns, since he sells engine of SMath to commercial companies.
That should give you an idea for the proof of its correct calculations.

As your original question, my guess is no software can give you that kind of verification.
Smath is a program, prone to bugs/mistakes as all out there.
This is an excellent project while it is a one man show (actually 3-5) except plugins.
There is no enormous software development team behind SMath yet it calculates as it should.
I don't think there are huge mathematical errors in it. I didn't encountered any of it.
But since it is a program there is a possibility of bugs/errors/mistakes as all CAS software could have them.

Maybe there were some errors in the past but those are already corrected.
Maybe there are some errors (very little possibility) and if someone find it it will be fixed immediately.
This is the good part of SMath, you can directly communicate with its creators and experienced users.
Errors shall be fixed within the next release and if it is critical a new release shall be distributed as soon as possible.
On the contrary, Mathcad/Mathematica/Maple/etc won't get that kind of quick fixes.
Anyone should not get cold feet to use SMath as a productive tool, just for that reasons.

Regards

Wrote

...
Could anyone tell me more about how it is ensured that calculations in SMath are correct or where I can find documentation about this? ...

I don't know any (that is: not any one) that offer that. Can you please show an example of that?

Some counter examples:

Maplesoft, at https://de.maplesoft.com/documentation_center/Maplesoft_EULA.pdf

"... YOU ASSUME THE ENTIRE
RISK AS TO THE USE AND PERFORMANCE OF THE SOFTWARE OR DOCUMENTATION IN TERMS OF
CORRECTNESS, ACCURACY, RELIABILITY, CURRENTNESS, OR OTHERWISE. WITHOUT LIMITING
THE GENERALITY OF THE FOREGOING, IN NO EVENT SHALL MAPLESOFT, ITS AGENTS OR ANYONE
ELSE WHO HAS BEEN INVOLVED IN THE CREATION, PRODUCTION OR DELIVERY OF THE
SOFTWARE AND/OR DOCUMENTATION BE LIABLE TO YOU OR ANY OTHER PERSON FOR ANY
DIRECT, INDIRECT, SPECIAL, OR CONSEQUENTIAL OR INCIDENTAL DAMAGES ... "

Matlab, at https://www.mathworks.com/license/mll/license.txt

"... The Materials are provided with all faults, and the entire risk of satisfactory quality, performance, accuracy, and effort is with Licensee ..."

Best regards.
Alvaro.

Wrote

I could imagine that there is an elaborate testing suite for the program and that it uses imported modules from verified repositories.

Plugins as is have to be checked vs other CAS systems.
1. The Smath Gamma(x) SS 6179 is incomplete and ± 4 decimals.
That does not meet Γ(x)from CAS repositories.
2. The native SS 6179 integrator is Simpson.
As is, it ranges from exact in rare instances to absolute crap.
In the mean time, two more integrator have been developed
and published in the Smath Community [Forum].
Further, Simpson does not recognize Frechet-II
It fails to evaluate the power function X^Y as a scalar function.
Have a good day, keep enjoying Smath ... Jean

Certification.sm (59 КиБ) скачан 26 раз(а).

Wrote

Hi everyone,

Thank you for this awesome product, we really like to work with it. Unfortunately, a colleague of mine is worried about the verification/certification of calculations in SMath, since it's open-source software.

Could anyone tell me more about how it is ensured that calculations in SMath are correct or where I can find documentation about this? The EULA does not include anything about this, I could not find the terms and conditions anywhere. Having worked on software development I could imagine that there is an elaborate testing suite for the programme and that it uses imported modules from verified repositories.

Thanks to anyone that can help!

There have been various industry and professional articles about this in the US.

1. in the US, the "engineer" is responsible and liable for verifying ALL calculations. This includes software generated. All software will have a disclaimer to that effect and our licenses will NOT allow us to "blame" the software. I can track down at least 2 articles on this subject if you like. REMEMBER, garbage in-->out!
   
2. You/your company has to run a QA/QC program to verify software ... any software, including items like word processors, spreadsheets, etc. This is one reason software has updates. How do you/your company currently confirm a spreadsheet?

Having previously worked in various critical industries (as well as others on this forum), my previous employers had certain sample problems that were hand calculated and verified via secondary software, etc. AFIK/recall, a software can/may be used to verify other software. Many times, a difference was found and the established calculation (old spreadsheet or hand calc) was found to have been "not as accurate".

The example that is most noteworthy for me was a tank calculation. I found at least 5 spreadsheet versions on a company server and they were all within round-off error of each other, but added margin at the end. We had a failure (PRV/Rupture Discs popping) and I was tasked to re-verify calculations. I had an MathCAD/Smath calculation that showed a "larger" than installed tank was required, WITHOUT applying any margin. As my calculation was "different" from 5 others, I had to dig into the spreadsheets (MathCAD/SMath really spoils you). All of the spreadsheets had "SIMILAR" round off conversions (e.g. pi=3.14 vs. using pi() in the spreadsheet as well as unit conversions). On smaller systems, the applied margin had been enough. On this, much larger than typical, system, these little differences, throughout the calculation, compounded such that it was more than the applied margin and assumptions. As I started to modify and existing spreadsheets, adding documentation to the changes (e.g. using pi() and other higher precision correction), the spreadsheets and the hand calculation fell in line with my MathCAD/SMath worksheet.