search

joshspeagle / brutus

Modeling stellar photometry with "brute force" methods
MIT License
33 stars 6 forks source link

Replace astropy coordinate transform #39

Closed gusbeane closed 3 years ago

gusbeane commented 4 years ago

Currently takes up ~3.5% of the runtime, possibly can be shaved with a homemade solution

joshspeagle commented 4 years ago

This turns out to be a negligible portion of the runtime per object using BruteForce, but would probably have a non-negligible amount of overhead if we ever wanted to use brutus in "sampling" mode (since I do actually have model generation utilities using isochrone and SEDmaker). I think we can consider this lower priority for now but should still plan to implement it just to prepare for this.

gusbeane commented 4 years ago

With preliminary performance tests of 578b241 I was finding it still takes ~0.1s/obj, so I think it should still be targeted before the next flagship run