Development build. This page describes main, not a released package. The latest published Lodestar.Extensions.MathNet is 0.1.0 — read its documentation.

Home › ExtensionsMathNet › Matrix conversion

Converts a Math.NET matrix into a CsrMatrix.

public static CsrMatrix ToCsrMatrix(Matrix<double> matrix)

Parameters — matrix is the matrix to convert; sparse or dense.

Returns — a CsrMatrix of the same shape holding the same values, sharing no array with the source.

Exceptions — ArgumentNullException when matrix is null. ArgumentException when it has more rows than one array of row pointers holds, which a diagonal matrix of int.MaxValue rows reaches in a few bytes (#1529); when a compressed-row storage's public arrays were left inconsistent — row pointers not starting at 0, decreasing or ending past the values or the column indices, or a column outside the matrix (#1608); and when a caller's own storage reports an entry outside the matrix, more non-zero values than one array holds, or different entries on the two walks the conversion makes (#1547).

Example — a dense matrix keeps only what it actually stores.

using Lodestar.Abstractions;
using Lodestar.Extensions.MathNet;
using MathNet.Numerics.LinearAlgebra;
using MathNet.Numerics.LinearAlgebra.Double;

Matrix<double> dense = DenseMatrix.OfRowArrays(
    [0.0, 3.0, 0.0],
    [0.0, 0.0, 0.0],
    [4.0, 0.0, 5.0]);

CsrMatrix csr = MathNetInterop.ToCsrMatrix(dense);

int stored = csr.NonZeroCount;   // => 3
int rows = csr.RowCount;         // => 3

Remarks — a matrix already in compressed-row form hands over its three arrays, checked and copied so neither side can mutate the other's, a null array reading as empty. Any other storage is read over what it stores. A dense matrix is read from its own column-major array and a diagonal one from its array of diagonal values; either way the non-zero ones are counted per row and then placed, so a diagonal matrix costs its diagonal, not its square (#1220), and a dense one its cells, which is the honest cost of changing layout. On those two paths a stored NaN is kept and a stored zero dropped; the compressed-row path copies explicit zeros with the rest, so a CsrMatrix with sorted rows and no repeated column round-trips unchanged, zeros included (#1403).

Any storage other than those three — a caller's own subclass of MatrixStorage<double> included — is walked the same way, and may enumerate in any order: each entry is placed by its row (#1551).

An empty row is a row: it contributes no stored value and its row pointer repeats the previous one, which is what the example's middle row shows.

Applies to — net10.0, netstandard2.0.

See also — MathNetInterop, MathNetInterop.ToSparseMatrix.