SEAS2016

Research visit at Harvard SEAS with Dr Ba

Building Python C extension of Luca's algorithm for computing inverse of a banded matrix

Use Makefile to generate bandinverse.o (object file corresponding to C version of Luca's code)
The Python C extension will build a module _bandinverse that can be imported as one would any Python module

Compile the C extension as follows python setup.py build_ext --inplace, which will generate the module
The _bandinverse module has one function bandinv that takes in arguments in the same format as Luca's C code.
Test the function as follows

Testing the code on a simple 3 by 3 matrix with bandwidth 2 (only one off diagonal):

import _bandinverse
U = np.array([0,1.0,0.7,1.0,0.7,1.0])
M, N = 2, 3
U.reshape(M,N,order='Fortran')
P =  _bandinverse.bandinv(U,M,N)

Name		Name	Last commit message	Last commit date
Latest commit History 27 Commits
.gitignore		.gitignore
ClusteringKalmanBasedOnSigma2e.ipynb		ClusteringKalmanBasedOnSigma2e.ipynb
ClusteringLaplaceBasedOnSigma2e.ipynb		ClusteringLaplaceBasedOnSigma2e.ipynb
ClusteringNotation.ipynb		ClusteringNotation.ipynb
Convergence_block_coordinate_MM.ipynb		Convergence_block_coordinate_MM.ipynb
Makefile		Makefile
README.md		README.md
Test_Kalman.ipynb		Test_Kalman.ipynb
Test_LikelihoodBernoulli.ipynb		Test_LikelihoodBernoulli.ipynb
Test_LikelihoodBernoulliAndGaussian.ipynb		Test_LikelihoodBernoulliAndGaussian.ipynb
Test_LikelihoodGaussian.ipynb		Test_LikelihoodGaussian.ipynb
_bandinverse.c		_bandinverse.c
bandinverse.h		bandinverse.h
bandinverse2.c		bandinverse2.c
cluster_train_data.csv		cluster_train_data.csv
setup.py		setup.py
testbandinv.c		testbandinv.c
thaldata.csv		thaldata.csv