Draft: Supernodal Cholesky

Reference issue

What does this implement/fix?

This implements the Supernodal Cholesky, an algorithm which divides the matrix into supernodes, then computes each supernode as a dense panel using dense linear algebra algorithms. Credit to Tim Davis + his collaborators for the bulk of the algorithm used.

Among other things one of the advantages of this algorithm, is that it has a lot of potential for parallelism.

Additional information

Todo

• Tests
• Base sparse Cholesky class
• Benchmarks with CHOLMOD and the simple solver.
• Allow independent super nodes to be ran in parallel