Numerical Analysis/Methods Packages

  • Thread starter Thread starter neurocomp2003
  • Start date Start date
  • Tags Tags
    Numerical
AI Thread Summary
The discussion centers around the search for effective numerical analysis packages in C/C++, with a specific interest in alternatives to LAPACK. Users mention resources like Netlib, which hosts a variety of numerical analysis functions, and the GNU Scientific Library (GSL), although GSL is noted to be primarily for Linux, posing a challenge for Windows users. There is acknowledgment of "Numerical Recipes in C" and an interest in its C++ version. Users share experiences with Netlib packages, particularly LAPACK, noting their reliability and performance. Optimization of these packages is discussed, highlighting that optimization can vary based on the intended processing type, such as vector or parallel processing. The conversation also references LAPACK++ and the Template Numerical Toolkit as evolving options. Additional resources from neighboring organizations are provided for further exploration of numerical software.
neurocomp2003
Messages
1,359
Reaction score
4
Any body use or know of any good numerical analysis packages in C/C++.
something like LAPACK. and are they easy to use.
 
Computer science news on Phys.org
Here is a good library of numerical analysis functions:

http://www.gnu.org/software/gsl/

If you want to learn how numberical analysis is done in C this is a good resource:

http://www.library.cornell.edu/nr/cbookcpdf.html
 
Last edited by a moderator:
thanks i alredy have thle numerical recipesin C(waiting on C++) just wondering if there were some other ones, or if people have used cLAPACK. I'll look into gsl but it seems only for linux =[ and I'm coding in windows platform(like the vc6 interface too much.

i'll also look at the netlib site...thanks...thats the one my compphys text(landau) suggested...have you guys used these packages...are they pretty optimized?
 
Yes, I've used/use several programs from netlib and have never had any problems with any of them (including several routines from LAPACK-fortran version). As for optimization it depends on what you mean by "optimized". The code can be optimized for vector processing, parallel processing, or scalar processing. Check the header of the given code you want for specifics.
 
You may know of this, but there's a version of LAPACK for C++ called LAPACK++, but this is being superseeded by the Template Numerical Toolkit, available at http://math.nist.gov/tnt/.
 
With respect to degree of "optimization" of netlib packages typically exceed my own workings by a fair margin, as problem free as it comes.

Usually access via our neighboring organization who have their things pretty good straightened and sorted out:

http://www.csc.fi/math_topics/FTP/index.html
http://www.csc.fi/math_topics/Software.html
 
Last edited by a moderator:
Back
Top