@@ -9,7 +9,7 @@ pymatsolver
99 :target: https://github.com/simpeg/pymatsolver/blob/master/LICENSE
1010 :alt: MIT license.
1111
12- .. image :: https://codecov.io/gh/simpeg/pymatsolver/branch/master /graph/badge.svg
12+ .. image :: https://codecov.io/gh/simpeg/pymatsolver/branch/main /graph/badge.svg?token=8uQoxzxf3r
1313 :target: https://codecov.io/gh/simpeg/pymatsolver
1414 :alt: Coverage status
1515
@@ -32,68 +32,42 @@ All solvers work with :code:`scipy.sparse` matricies, and a single or multiple r
3232
3333* L/U Triangular Solves
3434* Wrapping of SciPy matrix solvers (direct and indirect)
35- * Pardiso solvers now that MKL comes with conda!
36- * Mumps solver with nice error messages
35+ * Pardiso solvers
36+ * Mumps solvers
3737
3838
39- Installing Mumps
40- ================
39+ Installing Solvers
40+ ==================
41+ Often, there are faster solvers available for your system than the default scipy factorizations available.
42+ pymatsolver provides a consistent interface to both MKL's ``Pardiso `` routines and the ``MUMPS `` solver package. To
43+ make use of these we use intermediate wrappers for the libraries that must be installed separately.
4144
42- We have not been able to get the pip install to work because of multiple dependencies on fortran libraries.
43- However, the linux and mac installs are relatively easy. Note that you must have mumps pre-installed,
44- currently we have only got this working for the sequential version, so when you are installing,
45- you will need to point to that one. You can also look at the `.travis.yml ` file for how to get it working on TravisCI.
45+ Pardiso
46+ -------
47+ The Pardiso interface is recommended for Intel processor based systems. The interface is enabled by
48+ the `` pydiso `` python package, which can be installed through conda-forge as:
4649
47- Linux
48- -----
49-
50- From a clean install on Ubuntu:
51-
52- .. code-block :: bash
53-
54- apt-get update
55- apt-get -y install gcc gfortran git libopenmpi-dev libmumps-seq-dev libblas-dev liblapack-dev
56-
57- # Install all the python you need!
58- wget http://repo.continuum.io/miniconda/Miniconda-3.8.3-Linux-x86_64.sh -O miniconda.sh;
59- chmod +x miniconda.sh
60- ./miniconda.sh -b
61- export PATH=/root/anaconda/bin:/root/miniconda/bin:$PATH
62- conda update --yes conda
63- conda install --yes numpy scipy matplotlib cython ipython pytest coverage
64-
65- git clone https://github.com/rowanc1/pymatsolver.git
66- cd pymatsolver
67- make mumps
50+ .. code ::
6851
69- Mac
70- ---
52+ conda install -c conda-forge pydiso
7153
72- This assumes that you have Brew and some python installed (numpy, scipy):
73-
74- .. code-block :: bash
75-
76- brew install mumps --with-scotch5 --without-mpi
77-
78- git clone https://github.com/rowanc1/pymatsolver.git
79- cd pymatsolver
80- make mumps_mac
81-
82- If you have problems you may have to go into the Makefile and update the pointers to Lib and Include for the various libraries.
54+ Mumps
55+ -----
56+ Mumps is available for all platforms. The mumps interface is enabled by installing the ``python-mumps ``
57+ wrapper package. This can easily be installed through conda-forge with:
8358
84- This command is helpful for finding dependencies. You should also take note of have happens when brew installs mumps.
59+ .. code ::
8560
86- .. code-block :: bash
61+ conda install -c conda-forge python-mumps
8762
88- mpicc --showme
8963
9064
9165 Code:
9266https://github.com/simpeg/pymatsolver
9367
9468
9569Tests:
96- https://travis-ci.org /simpeg/pymatsolver
70+ https://github.com /simpeg/pymatsolver/actions
9771
9872
9973Bugs & Issues:
0 commit comments