Rhys Goodall

Seems like a much bigger refactor to be instantiating new PRNGs everywhere and then potentially runs the risk of users using the same seed in everything and ending up be...

I think this also suffers from correlated PRNGs as you're global seed is fixed meaning that every PRNG you spawn will start at the same point in the sequence. You...

@blankjul Any thoughts on this PR as stands? The only tests that currently fail are the hash assert tests that check there haven't been any changes in output. Due to...

So my belief is that the way this is implemented here doesn't introduce any new issues that aren't present in the current numpy global state. A strict singleton would be...

It's set to allow maintainer edits and so I'm not sure why it doesn't work. I am not sure it makes sense to give you access to my fork so...

I think there would need to be further logic to ensure that the threading lock and the thread singletons were still deterministically set according to the seed.

Can I bump this? I don't see the need to wait for 2.7.3 to be cut before we get an official 2.7.2 release that we can include in deps files....

thanks!

> Any objections to `numba` as a dependency (@ardunn for `matbench` and @janosh for `pymatviz`)? If so, that's fine, especially for Matbench based on the constraints @ardunn mentioned. @janosh if...

> @CompRhys I haven't compared to `elemd`. I think it's ~100-200x speedup relative to the original ElMD calculations. For reference, a 10,000×10,000 distance matrix can be computed in ~10-20s parallelized...