Multigrid.jl

A multigrid package in Julia. Uses shared memory parallelism using OMP and [ParSpMatVec] (https://github.com/JuliaInv/ParSpMatVec.jl) .

Includes:

1. Geometric multigrid on a regular mesh.

2. Smoothed Aggregation AMG Multigrid, based on the following paper (please cite if you use this code):

   Eran Treister and Irad Yavneh, Non-Galerkin Multigrid based on Sparsified Smoothed Aggregation. SIAM Journal on Scientific Computing, 37 (1), A30-A54, 2015.

Options for V,F,W and K cycles.

Includes a block version of multigrid. Most effective for using as a preconditioner for Block Krylov methods (see KrylovMethods.jl). Coarsest Grid can be solved using [MUMPS] (https://github.com/JuliaSparse/MUMPS.jl) or using Julia's backslash.

Requirements

This package is intended to use with julia versions 0.7-1.0.

This package is an add-on for jInv, which needs to be installed (mostly for regular mesh module).

Installation

Pkg.clone("https://github.com/JuliaInv/jInv.jl","jInv")
Pkg.clone("https://github.com/JuliaInv/Multigrid.jl","Multigrid")
Pkg.clone("https://github.com/JuliaInv/ParSpMatVec.jl","ParSpMatVec")
Pkg.build("ParSpMatVec");

Pkg.test("Multigrid")