Enable option of fitting v1 tracks with VeloKalman

We persist PVs (RecVertex) and their tracks (v1 tracks), but refitting them is currently only possible with the PrKalman or TrackMasterFitter. This breaks the reproducibility of what happens in HLT2, which uses tracks fitted with the VeloKalman. The functions in the VeloKalmanHelpers are unfortunately specific to the SoA containrs.

Ideally we fix this and make it also available for a TrackStateInit-like tool to re-initialise VELO tracks.

added RTA RTA WP2 Tracking labels

@ahennequ you have any suggestions on how to proceed here? if possible, to look into this?

The way to do this, is to keep as much common code as possible. Looking at VeloKalmanHelpers.h, the core of the computation (the filterWithMomentum function) is actually templated and should be reusable. So what need to be done is to write a new fitBackwardWithMomentum that takes track v1, and a corresponding VeloKalman algorithm (maybe the current one can be templated, since its just calling the fit function and copying all other parameters.

Do we really call those functions on VELO tracks? I'm a bit surprised.

We call them on the Velo part of the long track (only considering velo hits)

The case here is about the tracks we use to make the PV in HLT2, so VELO tracks, not long tracks.

(originally PrVelo tracks I think)

There are 2 versions of the fitting that are both defined in VeloKalmanHelper.h:

• The one called from the VeloKalman alg (fitBackwardWithMomentum) requires a momentum estimate.
• The one called from VeloClusterTracking (right after the track get created: fitBackward / fitForward) that doesn't require a qop.

I guess its the second version you are interested in.

Indeed, I don't think we actually use the former in practice.

@ahennequ are you working on this implementation? If not, I can write something, but it could require some refactoring...

I haven't started yet, but it is on my list of things to do. If you need it soon, I can work on it this week.

OK - that would be great! Indeed, the sooner the better.

mentioned in issue #578 (closed)

mentioned in merge request !4113 (merged)

closed with merge request !4113 (merged)

mentioned in commit 7ba7e142

mentioned in commit 514d78b7

mentioned in commit ee7a0bcc

mentioned in commit a2e9ecc7

mentioned in commit a06d47e9

mentioned in commit 4f9936ca

mentioned in commit 009c1d3f

mentioned in commit 9e12fc05