Lazarus

Programming => Packages and Libraries => Topic started by: stephanweber on March 19, 2019, 01:27:48 pm

Title: Math package for nonlinear constrained optimization, low-discrepancy sampling?
Post by: stephanweber on March 19, 2019, 01:27:48 pm
Hi,

in the internet you can find many great math libs, but for these special topics I for nothing yet for Pascal. Does anybody has a hint? Or maybe a really good C++ (or c#, python, java, etc.) to Pascal converter?

Bye Stephan
Title: Re: Math package for nonlinear constrained optimization, low-discrepancy sampling?
Post by: sstvmaster on March 19, 2019, 02:09:36 pm
"Compatible with FreePascal" -> http://www.alglib.net/download.php#delphi
Title: Re: Math package for nonlinear constrained optimization, low-discrepancy sampling?
Post by: apeoperaio on March 20, 2019, 08:26:32 pm
Check mrmath library, it works on Lazarus and it is superfast.
https://github.com/mikerabat/mrmath
Title: Re: Math package for nonlinear constrained optimization, low-discrepancy sampling?
Post by: stephanweber on March 28, 2019, 12:42:26 pm
Thanks! These units work quite fine.
However, the most complete unit seems to be AMath, BUT I cannot compile it under freepascal, there are some asm statements, unfortunately.
Has anybody made AMath working under Lazarus?

Bye Stephan
Title: Re: Math package for nonlinear constrained optimization, low-discrepancy sampling?
Post by: MathMan on March 28, 2019, 01:48:20 pm
Hi Stephan,

I assume you mean the AMath unit of Wolfgang Ehrhardt (http://www.wolfgang-ehrhardt.de/misc_en.html#amath)?

If so, what's your target environment? The website explicitly states

"The archive amath_2018-11-27.zip contains units for accurate mathematical methods without using multi precision arithmetic. Please note that the high accuracy can only be achieved with the rmNearest rounding mode; it decreases if other modes are used. AMath is designed for the 80-bit extended data type, and therefore cannot be used with 64-bit code on 64-bit system (for these use the corresponding DAMath package). More information can be found on the separate introduction page."

Maybe you need to take a look at DAMath instead.

Cheers,
MathMan
Title: Re: Math package for nonlinear constrained optimization, low-discrepancy sampling?
Post by: glorfin on April 29, 2019, 11:25:35 am
Hi Stefan!

I have quite recently implememnted in Pascal the COBYLA algothithm for LMath library (https://sourceforge.net/projects/lmath-library/).
It is not yes officially released, but is already in SVN repository. So, have a look at it. See uCobyla unit in uOptimum package.

Best regards!
Viatcheslav