A project that attempts to do the following:
- Wrap the Eigen sparse (and conjugate gradient) Cholesky solver for Python.
- Test speed against the scikits-spase wrapper for CHOLMOD on multiple benchmarks.
- Create a Tensorflow op for sparse Cholesky solver.
- Create the Tensorflow forward and backward differentation equations.
To install Eigen, simply download the source from here and copy the header files into a location that is in the PATH e.g.
cd ~/Downloads
tar -xvf ~/eigen-eigen-<build-info>
cd eigen-eigen-<build-info>
sudo cp -r Eigen /usr/local/include
For example
python3 test.py 10000 x x cholesky_dense numpy