Implements VeloKalman for V1 tracks

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.

This MR adds a VeloKalmanV1 algorithm that is able to fit v1 tracks.

Closes #587 (closed)

NOTE: Once merged to master, a cherrypick MR should be opened back porting this to 2024-patches. See discussion below.

FYI @ldufour @mveghel

added 1 commit

• b6eca748 - Implements VeloKalman for V1 tracks

Compare with previous version

@ldufour do you have an example sequence where this could be tested ?

added RTA label

I suspect something is wrong - when I use this fitter instead of the PrKalman on my PV tracks, I see the PV z position change by 130 microns on average, with a spread of 442 microns.

For the PrKalman this is much smaller (no bias, and a spread of 6 microns). If anything, we expect this fitter to be better, not worse.

added 1 commit

• 868628ac - Fix xy fractions magnitude

Compare with previous version

added 1 commit

• 1f15631c - Refactor and input range

Compare with previous version

added 74 commits

• 1f15631c...9a7e26f2 - 71 commits from branch master
• bfa32db1 - Implements VeloKalman for V1 tracks
• 6dbd2d7a - Fix xy fractions magnitude
• 30704079 - Refactor and input range

Compare with previous version

Do we want this in 2024-patches, or master?

added 1 commit

• 36996502 - Add option to reset initial conditions of the fit

Compare with previous version

added 1 commit

• ac7c8d13 - Add option to reset initial conditions of the fit

Compare with previous version

changed the description

mentioned in merge request !4081 (merged)

mentioned in merge request Moore!3921 (merged)

What is the status of this? @ahennequ @mveghel

marked this merge request as ready