Searching for "jacobi method" finds passing mention in PDLA::MatrixOps, but if you perhaps asked a more specific question with sample code showing what you're attempting to do you might get a more germane (or possibly even useful) reply.

As for your second question, yes; see perlxs or Inline::C for example. In general you'd want to search CPAN first because if you're wanting to use an open source C library there's a good chance someone's written a module wrapping it already.

Thank you. What I have to do is find eigenvalues and eigenvectors using that method but I have been asked to use certain routine from the book Numerical Recipes in C.

I think https://metacpan.org/pod/PDL::MatrixOps#eigens looks like what you want. Please don't use PDLA, it's effectively deprecated (I should update it accordingly).

An additional PDL module that uses Jacobians within a Levenberg-Marquardt approach is PDL::Fit::Levmar.

FWIR, jacobi (and SOR, Gauss-Seidel) are basically just averaging values over time steps. How many dimensions are you dealing with and what is the PDE you're solving? PDL almost surely can help you out; but even if not it'd be pretty easy to implement this in pure Perl^tm.

If you are interested in running this on the GPU (for example, here is a random CUDA implementation https://github.com/MMichel/CudaJacobi) via Perl, I am looking for testers with a windows machine and an nvidia CUDA-capable GPU for https://github.com/hadjiprocopis/perl-inline-cuda . I could try and port the algorithm as an Inline::CUDA test, in exchange.

bw, bliako

Until tonight I didn’t know anything about this issue - except that an old friend of mine is named Jacoby. I guess this is what you are looking for. Probably you find a solution in PDL. I don’t know.

But you might take a look at this. If I’m not totally wrong about what you want.

If I guessed right you might take a look at Is this a way to Go Perl #1.

And probably you find a "simple solution" for your problem or at least some further inspiration. And a serious use case to Go Perl.

Best regards, Karl

Jacobi method