Denis Demidov

Hi, did you check that the preconditioned solution really converged with the required tolerance? That is, is ||f - Ax|| / ||f|| < 1e-8 (or however small did you set...

amgcl::io::mm_reader does not require the rows in the input file to be ordered, so you can use that (see usage examples in examples/solver.cpp). The reader will sort the rows as...

I guess you would have to implement a custom reader in this case. If you want to solve the system also using three cpus, then maybe you could read the...

Not that I know of. I never worked with vcpkg, but if you know how to set it up, then I'll gladly accept a pull request.

Found an interesting discussion here: https://github.com/microsoft/vcpkg/issues/995, which contains a link to the list of header-only vcpkg libraries. Maybe you could get some examples/ideas there.

Tested successfully in https://github.com/KratosMultiphysics/Kratos/pull/3831.

There are no such plans currently, but I would accept a PR for this :). Note that the python package here is more of an example, as it does not...

One could also solve for `y` in `My = F(x)` using an iterative solver. Depending on `M` this could be more memory efficient.

That was just a suggestion in case the system is too large to be solved by a direct solver (as compute cost in general grows as `O(n^3)` for a matrix...

10000 could be too big for a dense LU decomposition, but you could use a sparse LU solver, from [SuperLU](http://crd-legacy.lbl.gov/~xiaoye/SuperLU/) or [PaStiX](http://pastix.gforge.inria.fr/files/README-txt.html). You could also borrow an implementation of SkylineLU...