wangyu9 / suitesparse.py

Cython/Python interface to SuiteSparse

SuiteSparse.py

Cython/Python interface to SuiteSparse.

You can use any Python/Cython libraries that implement sparse matrices but SuiteSparse.py is really well integrated with CySparse.

SuiteSparse is a huge project written mainly in C and C++. We try to adhere as much as we can to the original libraries and provide both the common and the advanced usages of each sub libraries. Python not being C nor C++, some choices had to be made. See the documentation on how to use this Cython/Python wrapper.

Our whish: if you know Python and know the SuiteSparse library, you'll very quickly be able to use SuiteSparse.py to its fullest.

Announcements

Working on UMFPACK. Not urgent but "easy" and a good start!

Nikolaj

Dependencies

You need to install SuiteSparse. Pay attention to the fact that you need to install the dynamic libraries, not the static ones.

Althought SuiteSparse.py has been optimized to use sparse matrices from CySparse, you don't have to install CySparse! It's strongly recommended but not mandatory.

• NumPy

Installation

Python version

Cython version

Run tests

Run performance tests

What is implemented?

• UMFPACK Versions 5.7.1 (Oct 10, 2014)
• CHOLMOD Versions 3.0.3 (Oct 23, 2014)
• SPQR
• KLU
• BTF
• AMD
• CAMD
• COLAMD
• CCOLAMD

Release history

• Version 0.1.0 released on Jan 10, 2016

  Added tests (48) for basic UMFPACK get_LU(). Introduced CHOLMOD.

• Version 0.0.1 released on Dec 31, 2015

  Added basic UMFPACK for CySparse LLSparseMatrix matrices.