Does PMIS coarsening run in parallel?

[learning-chip]

I was looking at the internal code of PMIS, the maximal_independent_set_parallel function inside amg_core/graph.h:

pyamg/pyamg/amg_core/graph.h

Lines 97 to 107 in e9f2218

	template<class I, class T, class R>
	I maximal_independent_set_parallel(const I num_rows,
	const I Ap[], const int Ap_size,
	const I Aj[], const int Aj_size,
	const T active,
	const T C,
	const T F,
	T x[], const int x_size,
	const R y[], const int y_size,
	const I max_iters)
	{

However I did not find code related to parallelization / multi-threading / openmp / etc. Am I missing something, or parallelism is not implemented yet? The algorithm itself is indeed highly parallelizable.

Thanks!

[lukeolson]

no parallelism yet. but this helped nudge an old branch and we now have a PR:
#324

This would add OpenMP to the solve phase ... adding it to the setup phase is more or less straightforward. Let's see how this PR goes.