Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
SparsityPattern -> ColumnColoring
Partition sparse matrix columns into independent non-overlapping sets to compute the entire Jacobian in a minimal number of function evaluations.
Problem it solves
Computing a sparse Jacobian via standard finite differences requires as many function evaluations as there are dimensions, which is highly inefficient.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.