Alg - A set of linear solvers for sparse matrices

Version 1.4.0

Alg is a library to handle sparse matrices involved in some finite element problems. It is NOT a general purpose library, only the minimal set of elementary algebraic functions is implemented in order to give access to gradient conjugate and bi-conjugate gradient algorithms.

The static library libalg.a is created from CMakeLists.txt in alg/src

Available algorithms: conjugate gradient : standard and directional biconjugate stabilized gradient : standard and directional LU(for precond matrix) diagonal preconditioner is implemented in most routines.

All cg and related algorithms are using diagonal pre-conditioner.

installation

sudo make install to copy the static library libalg.a to usr/local/lib and headers to usr/local/include/alg

Dependencies :

C++ 17 and STL

libtbb-dev has to be installed because STL depends on TBB for the implementation of parallel execution policies with gcc compiler. how to install : sudo apt-get install libtbb-dev

Options :

1. Some unit tests are available passing the following option to cmake (OFF by default): cmake . -DENABLE_UTESTS=ON

2. Some GPU executable (dev only) are available using the dedicated option (OFF by default): cmake . -DENABLE_GPU=ON These small executables are here for future development purpose only, they are useless to the alg library at the moment.