Closed tomaskontrimas closed 5 months ago
Hi @martwo, could you take a look?
I'm wondering whether it's worth to optimize it by introducing a bit more complex check:
I could set a variable like self.search_centers_failed
in the except ValueError:
clause, and reset it when initializing the new trial. This would save some time, as following likelihood minimization steps after the first one would not try to call search_centers
again.
In case the photospline
search_centers
call fails (wheneventdata
is outside photospline boundaries) we can fall back to the slower photospline evaluation (evaluate_simple
) which handles this case by returning 0 for the event.It is relevant for the northern tracks
v005p00
andv005p01
datasets.