Cholesky Decomposition is slow #69
Description
The current implementation of the cholesky decomposition involves lots of copying, like L = L.set(i, j, value). The whole thing would be a little bit more efficient if we operated on a S[][] instead of a Matrix<S>.