Just noticed in PR #62 that @HironaoMiyatake has a different way of making formatted arrays (well, recarrays) than the way we currently do it: the function numpy.rec.fromarrays. I see here that the major difference between formatted numpy arrays and recarrays is that access in recarrays is a little slower, so if you loop a lot it can be a time sink.
This is a reminder to myself to investigate (once the higher-priority items are done) exactly how much it slows us down to switch to recarrays, because I'd rather use numpy code if we can, but since it would be a change in how our data is stored, it could potentially slow us down quite a bit.
Just noticed in PR #62 that @HironaoMiyatake has a different way of making formatted arrays (well, recarrays) than the way we currently do it: the function
numpy.rec.fromarrays. I see here that the major difference between formatted numpy arrays and recarrays is that access in recarrays is a little slower, so if you loop a lot it can be a time sink.This is a reminder to myself to investigate (once the higher-priority items are done) exactly how much it slows us down to switch to recarrays, because I'd rather use numpy code if we can, but since it would be a change in how our data is stored, it could potentially slow us down quite a bit.