gomm

gomm: a MatrixMarket parser in Go.

Parses matrices from the MatrixMarket.

The Matrix Market provides convenient access to a repository of test data for use in comparative studies of algorithms for numerical linear algebra. Matrices as well as matrix generation software and services, from linear systems, least squares, and eigenvalue computations in a wide variety of scientific and engineering disciplines are provided. Tools for browsing through the collection or for searching for matrices with special properties are included. -- MatrixMarket.

Usage

Matrices can be downloaded directly as a mat.Matrix interface using the GetMatrix routine. For example, to download a sparse matrix and obtain its *sparse.CSR representation:

mm, err := GetMatrix("Harwell-Boeing", "bcsstruct1", "bcsstk01")
csr, ok := mm.(*sparse.CSR)

n, m := csr.Dims()
nnz := csr.NNZ()
fmt.Printf("type: %T, (rows,cols): (%d,%d), nzz: %d", csr, n, m, nnz)

// Output: type: *sparse.CSR, (rows,cols): (48,48), nzz: 400

Alternatively, the matrices can be dowloaded to disk first in as .mtx.gz and parsed from there:

matrix := gomm.Matrix{
  collection: "Harwell-Boeing",
  set: "bcsstruct1",
  name: "bcsstk01",
}

// downloads to bcsstk01.mtx.gz 
err := matrix.Download()

// open and decompress
f, _ := os.Open(file)
defer f.Close()

rd, _ := gzip.NewReader(f)

// parse matrix, returs mat.Matrix interface 
mm, _ := matrix.Parse(rd) 

// sparse matrices can be retrieved
csr, _ := mm.(*sparse.CSR) 

n, m := csr.Dims()
nnz := csr.NNZ()
fmt.Printf("type: %T, (rows,cols): (%d,%d), nzz: %d", csr, n, m, nnz)

// Output: type: *sparse.CSR, (rows,cols): (48,48), nzz: 400

Install

go get github.com/maxvdkolk/gomm

References

• The MatrixMarket format: https://math.nist.gov/MatrixMarket/
• Dense matrices and the mat.Matrix interface: https://godoc.org/gonum.org/v1/gonum/mat
• Sparse matrices *sparse.CSR: https://github.com/james-bowman/sparse