#REDUNDANT MATRIX MULTIPLICATION

this library provide a faster approach to multiply matrices: A += B*C, where B has a lot of redundancies. For example, when the matrix is huge, but the number of distinct entries is very limited.

##TEST run make test

##BENCHMARKING ON CNNs benchmark_sparse.c include a few methods to benchmark different convolutional layers in Cifar and German

the example for 2nd cifar layer has with it a real input layer. The other examples simply uses all-zero-entry matrices.

Note, for A += B*C, only B is implemented as the redundant matrix. A and C are still a flattened 1D array.

Currently, a python script is used to convert matrix from dense to redundant. The c program only does the loading and the multiplication