RichRayTracing - Turn off the beam pipe intersection checks with raytracing photons for the CK rings.

To see the changes w.r.t. !888 (merged)

https://gitlab.cern.ch/lhcb/Rec/compare/RichPhotonQuarticReco-TurnOffUnambigPhotonAndBeamChecks...RichCKRingRayTrace-OptRingPointsNoBeamCheck

This MR turns off the beam pipe intersection checks for the ray traced photons used for the CK rings. Not only does it save CPU, but also (a little to my surprise) improves the PID performance a little. See the attached plot, for the Upgrade simulations.

RichDLLk-RichDLLpi_Long_P3-100_Pt0.5-100.pdf

added 1 commit

• 8d16f445 - fix some settings

Compare with previous version

added 1 commit

• cbf38319 - Turn off beampipe check in ray tracing

Compare with previous version

added 1 commit

• 7cdbceca - Make unambiguous photon check UNLIKELY

Compare with previous version

added 1 commit

• bfd427f6 - Remove now incorrect UNLIKELY

Compare with previous version

added 1 commit

• 2dd631e9 - explicitly assign float types

Compare with previous version

changed the description

changed the description

@seaso @fkeizer FYI. Note the small improvement in upgrade PID performance.

changed the description

added 1 commit

• c2c7206d - Add some LIKELY compiler hints

Compare with previous version

@jonrob is your plan to include this in the next round of testing tomorrow? It would be nice if by Thursday we could be in a position to merge the Rich MRs and the Brunel references updates the following day.

My plan was to test my last update to the refs in Brunel!357 (merged) in the nightlies over night, and if OK label this for testing in lhcb-gaudi-head tomorrow. So yes all going to plan on Thursday I will have the last updates for the refs ready to be tested and merged Friday. Is this OK ?

Yes, my thinking is that the Rich MRs can be merged (or at least unWiped) on Thursday so the last references can be tested in lhcb-head on Friday and I can move to another set of MRs for lhcb-gaudi-head on Thursday

mentioned in merge request Brunel!357 (merged)

• [2018-02-09 00:19] Validation started with lhcb-gaudi-head#1753
• [2018-02-09 13:31] Validation started with lhcb-gaudi-head#1754
• [2018-02-09 17:23] Validation started with lhcb-gaudi-head#1755
• [2018-02-10 00:11] Validation started with lhcb-clang-test#857
• [2018-02-10 00:12] Validation started with lhcb-gaudi-head-py3#60
• [2018-02-10 00:14] Validation started with lhcb-tdr-test#55
• [2018-02-10 00:15] Validation started with lhcb-head#1750
• [2018-02-10 00:22] Validation started with lhcb-gaudi-head#1756
• [2018-02-10 12:51] Validation started with lhcb-gaudi-merge#412

@jonrob could you please check that the Brunel test diffs in today's lhcb-gaudi-head are as you would expect from this MR? If so please update the Brunel references one last time, and I will merge everything tomorrow.

Yes, all seems fine in the nightlies to me. This MR causes a small increase in the number of tracks per event that get selected,and a corresponding increase in the photon counts as well. The PID table changes look roughly, given the stats, as I would expect. I will update the brunel logs shortly.

unmarked as a Work In Progress