Track finding based on idealised seeding should be implemented

[stephenswat]

In order to better understand the performance of traccc, we need to disentangle the performance issues related to 1) the number of seeds produced in the seed finding stage; and 2) the performance of the track finding stage. An important step in this disentanglement process is the construction of an idealised seeder, i.e. a seeder that employs truth information to provide the smallest possible number of the highest quality track. A brief overview of the algorithm of such a seeder is as follows: for each particle in the truth data, find the first three measurements. Then convert those measurements to spacepoints and form a seed from them.

The idealised seeder will always return one and only one seed for each charged truth particle with non-trivial lifetime.

[beomki-yeo]

If I understand correctly, Track_finding_based_on_idealised_seeding will collect the truth triplets by combining the first three measurements of all truth particles and run track parameter estimation on them. is that correct?

truth_track_finding collects the seeds (bound track parameter) from the truth trajectory information at the first measurement of all truth particles. So I think it already does what you need 🤔

[stephenswat]

Fabulous! We can use that then! 😄

[stephenswat]

Okay - it seems that the truth tracking examples currently don't work on Geant4 simulated data, but I think that can be fixed easily enough.