Libmtx 0.5.0

This version adds support for reordering and partitioning sparse matrices using the SCOTCH graph partitioner.

Libmtx 0.4.0

This version adds support for reordering sparse matrices by first partitioning the adjacency graph using METIS and then grouping rows and columns by the parts that were assigned by the partitioner.

This version fixes an issue with the METIS partitioning and nested dissection reordering of non-square matrices. The fix ensures that the data structure representing the (undirected) bipartite graph of the matrix rows and columns is correctly passed to METIS.

Libmtx 0.3.0

This version adds support for reordering sparse matrices based on the nested dissection algorithm as implemented by the METIS graph partitioner.

It is also now easier to build with a user-defined BLAS library.

mtxrand was added to generate random matrices and vectors in Matrix Market format.

mtxgemm was added to compute matrix-matrix products. This can be used to benchmark matrix-matrix multiplication (gemm) in BLAS libraries.

Libmtx 0.2.0

MPI support is added for many operations, such as sorting and vector operations. There is work in progress to implement distributed sparse matrix-vector multiplication.

Several implementations for vector operations now provided, including a serial implementation, a BLAS-accelerated implementation, and an OpenMP shared-memory parallel implementation.

Matrices can now be reordered using Reverse Cuthill-McKee and partitioned using METIS.

libmtx 0.1.0

This is the initial release of libmtx, which supports reading
and writing Matrix Market files, sorting sparse matrices and basic
BLAS operations, such as matrix-vector multiplication.