".
To replicate this behaviour try this simple program
static void Main(string[] args)
{
Matrix<double> pippo = CreateMatrix.Dense<double>(2, 2, 1.0);
pippo[0, 1] = 2.0; pippo[1, 0] = 3.0; pippo[1, 1] = 4.5;
MatlabWriter.Write("prova2.mat", p, "pippo");
}
Perhaps the problem is in octave, because I can read using MatlabReader a Matrix exported by octave in compressed format.
So I added to MatlabWriter a function WritePlain to export matrices using not compressed format.
In MatlabWriter.cs
public static void WritePlain<T>(string filePath, Matrix<T> matrix, string matrixName)
where T : struct, IEquatable<T>, IFormattable
{
using (var stream = File.OpenWrite(filePath))
{
Formatter.FormatFilePlain(stream, new[] { Pack(matrix, matrixName) });
}
}
and a function FormatFilePlain in Formatter.cs
/// <summary>
/// Writes all matrix blocks to a stream not using compress format.
/// </summary>
internal static void FormatFilePlain(Stream stream, IEnumerable<MatlabMatrix> matrices)
{
using (var buffer = new BufferedStream(stream))
using (var writer = new BinaryWriter(buffer))
{
// write header and subsystem data offset (116+8 bytes)
var header = Encoding.ASCII.GetBytes(HeaderText + DateTime.Now.ToString(Resources.MatlabDateHeaderFormat));
writer.Write(header);
Pad(writer, 116 - header.Length + 8, 32);
// write version (2 bytes)
writer.Write((short)0x100);
// write little endian indicator (2 bytes)
writer.Write((byte)0x49);
writer.Write((byte)0x4D);
foreach (var matrix in matrices)
{
// write data type
writer.Write((int)DataType.Matrix);
writer.Write(matrix.Data.Length);
writer.Write(matrix.Data);
}
writer.Flush();
writer.Close();
}
}
Using this code, octave can read matrices generated in Mathnet.Numerics.
Perhaps this code could be useful to someone.
Thanks for all.
Lalo.
.