#matlib matlib is a small package for the creation and testing of matrices for openLCA. It uses the high performance math library OpenBLAS.

Usage

Checkout this repository:

git clone https://github.com/GreenDelta/matlib.git

Build the native library for native library for your system as described in native/Build_<your_os>.md.

File format

This package provides methods for loading (goblapack.Load) and saving (goblapack.Save) matrices from files in a simple binary format:

header 8 bytes
4 bytes: uint32, number of rows, little endian order
4 bytes: uint32, number of columns, little endian order

content, rows * columns * 8 bytes:
matrix data, float64, little endian and column major order

Here is a small Python script for writing a Numpy matrix in this format:

    m = numpy.load('path/to/file.npy')
    rows, cols = m.shape
    with open('path/to/file.bin', 'wb') as f:
      f.write(struct.pack("<i", rows))
      f.write(struct.pack("<i", cols))
      for col in range(0, cols):
        for row in range(0, rows):
          val = m[row, col]
          f.write(struct.pack("<d", val))