Remove CPU HLT1 luminosity line

Closes #477 (closed)

To be tested together with Rec!3108 (merged) Panoramix!121

added Luminosity legacy labels

marked this merge request as draft

added RTA label

added 1 commit

• d3f53f85 - commented out HLT lumi line from Run 2

Compare with previous version

added 1 commit

• 40879979 - Fixed formatting

Compare with previous version

marked this merge request as ready

In fact, eventually, this should be tested together with Rec!3108 (merged), but I did not figure out how to create this dependency in gitlab between Rec and Moore. Test with Rec master could also be useful, however, but the final test with Rec!3108 (merged) should be stricter.

assigned to @balagura and @msaur

changed the description

mentioned in merge request Rec!3108 (merged)

• [2022-09-12 12:57] Validation started with lhcb-master-mr#5624
• [2022-10-10 13:21] Validation started with lhcb-master-mr#5871

added cleanup label

changed the description

mentioned in merge request Panoramix!121

Hello @balagura, with @rmatev was discussed that it is better to remove obsolete code than just commenting it out. So better to do it as in the case of Rec!3108 (merged).

@balagura let me clarify that this is mostly removing the calculation of counters and filling the HltLumiSummary bank in the CPU HLT1 (kept as the backup solution in case Allen faced some showstopper). This is probably okay by now, however, it might be worth keeping the relevant bits in Rec (and here) in case we want to create and write lumi counters in HLT2 (or in monitoring tasks). Note that this is decoupled from the really obsolete FSR-related algorithms and tools for offline luminosity.

What do you think?

mentioned in merge request LHCb!3518 (merged)

marked this merge request as draft

Sorry for my late reply, I was a bit busy with other things (BGI is resuming in Glasgow and Bogota, PLUME-MUON lumi discrepancy discussed at the lumi meeting today) while this might require changes deeper than planned. And sorry for my general ignorance of why we need Moore even when we have Allen... Indeed, we may probably rely on GPU now for HLT1 (I hope), but RICH counters might be calculated at HLT2 because RICH is not reconstructed in Allen (as I understand). This is under discussion, maybe а simple RICH hit counting can still be done in Allen, but this requires decoding of the RICH banks etc.

On the other hand, to make things work in Moore (even just pass-through in HLT2) we'll need a new HltLumiSummary which is under development. So, this creates more dependencies on things which I know only generally. I'll try to learn them and then propose something.