Brian

Currently looking into the normalization of the first hit probability distribution. In the plots shown during the group meeting yesterday, several parameters turned out to be fixed (e.g. the direction angles). This forms part of the reason for the tiny probability values that were reported.

Also provided the derivation for the hit arrival times, assuming a muon Cherenkov hypothesis (as part of improving track reconstruction documentation).

Jordan

Currently busy comparing some of the details in the toy mc and JSirene mc. There seems to be a bigger variance in the arrival times of the hits in toy mc. Cause is under investigation

(You can find the toy mc code in the AAnet/rec directory)

Thijs

Last time, showed that the toy mc prefers negative arrival times. This was found to be caused by the light contribution originating very close to the vertex (previously there were no sampling points in the elongation PDF, to account for this).

This time, account for the light contribution close to the vertex using two different methods:

1. Assigning a fraction of the total emitted energy to the vertex
2. Sampling more frequently between the vertex and the 10th percentile of the elongation PDF.

Assigning a 1% energy emission fraction to the vertex yields the best perpendicular resolution for the vertex, regardless of how often the elongation PDF is sampled in the region up to the 10th percentile. But doing so moves the maximum likelihood estimate for the arrival time to a negative value with respect to the MC truth. So a balance may need to be struck.