You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe. MapDatasetEventSampler is quite slow when sampling a very large number of sky-models (hundreds of sources).
Describe the solution you'd like
Evaluate IRF values for all models just once?
The text was updated successfully, but these errors were encountered:
It would be good to do some profiling here and actually find out what the bottleneck is.
In general I think this is a bit tricky: when you have a large number of low flux sources, most of them do not create any event. However in general you cannot know which one. So you have to evaluate the model, sample the flux and see if you get an event or not. So you still have to evaluate the predicted counts for all source models.
The sampling of the IRFs then scales with the number of events, not the number of sources. So if you have large number of low flux sources, the IRF sampling should not significantly increase in runtime. The runtime should be completely dominated by the high flux sources, which create most of the events.
Is your feature request related to a problem? Please describe.
MapDatasetEventSampler
is quite slow when sampling a very large number of sky-models (hundreds of sources).Describe the solution you'd like
Evaluate IRF values for all models just once?
The text was updated successfully, but these errors were encountered: