fix: propagate parameters from reco_flags to cpp plugins

[wdconinc]

Briefly, what does this PR introduce?

This addresses a lot of the discrepancies in https://eic.github.io/EICrecon/#/table_flags/flags_view

What kind of change does this PR introduce?

• Bug fix (issue: reco_flags duplicates parameters)
• New feature (issue #__)
• Documentation update
• Other: __

Please check if this PR fulfills the following:

• Tests for the changes have been added
• Documentation has been added / updated
• Changes have been communicated to collaborators

Does this PR introduce breaking changes? What changes might users need to make to their code?

No.

Does this PR change default behavior?

No.

[wdconinc]

Use the pion flags file for diffs to have higher probability that Hcal factories are initialized.

[c-dilks]

There are still some differences at https://wdconinc.github.io/EICrecon/#/table_flags/flags_view, is this the latest version? Also, I will take care of the DRICH ones in a separate PR.

[wdconinc]

I think my main is pointing to a different commit by now.

[wdconinc]

Ah, no, my main is pointing to a but the last commit here. The remaining differences in protoclusters is still under investigation. But I think this is still ready for merging.

[wdconinc]

There is also the issue that the gh-pages can only show either arches or brycecanyon, and EICrecon only initializes a factory when needed. So part of the chain is just not exercised.

[c-dilks]

I believe these are the remaining differences:

#
#key                                                           => reco_flags.py as is   VS reco_flags.py with no overrides (empty `eicrecon_reco_flags` list)
#
BEMC:EcalBarrelImagingRawHits:readoutClass                     => pixel                 VS
BEMC:EcalBarrelScFiRawHits:energyResolutions                   =>                       VS 0
BEMC:EcalBarrelScFiRawHits:readoutClass                        => light_guide           VS
EEMC:EcalEndcapPInsertIslandProtoClusters:dimScaledLocalDistXY => 1.5,1.5               VS 1.8,1.8
EEMC:EcalEndcapPInsertIslandProtoClusters:localDistXY          => 10,10                 VS
EEMC:EcalEndcapPInsertIslandProtoClusters:minClusterCenterEdep => 0.01                  VS 0.03
EEMC:EcalEndcapPInsertIslandProtoClusters:minClusterHitEdep    => 0                     VS 0.001
EEMC:EcalEndcapPInsertRecHits:capacityADC                      => 16384                 VS 8096
EEMC:EcalEndcapPInsertRecHits:dynamicRangeADC                  => 3.0                   VS 0.1
EEMC:EcalEndcapPInsertRecHits:pedestalMean                     => 100                   VS 400
EEMC:EcalEndcapPInsertRecHits:pedestalSigma                    => 0.7                   VS 3.2
EEMC:EcalEndcapPInsertRecHits:samplingFraction                 => 0.03                  VS 0.998
EEMC:EcalEndcapPInsertRecHits:thresholdFactor                  => 5.0                   VS 4
EEMC:EcalEndcapPInsertRecHits:thresholdValue                   => 2                     VS 0
EEMC:EcalEndcapPInsertTruthClusters:enableEtaBounds            => 1                     VS 0
EEMC:EcalEndcapPInsertTruthClusters:logWeightBase              => 6.2                   VS 3.6
EEMC:EcalEndcapPIslandProtoClusters:dimScaledLocalDistXY       => 1.5,1.5               VS 1.8,1.8
EEMC:EcalEndcapPIslandProtoClusters:localDistXY                => 10,10                 VS
EEMC:EcalEndcapPIslandProtoClusters:minClusterCenterEdep       => 0.01                  VS 0.03
EEMC:EcalEndcapPIslandProtoClusters:minClusterHitEdep          => 0                     VS 0.001
HCAL:HcalBarrelIslandProtoClusters:localDistXY                 => 150,150               VS 1.5,1.5
HCAL:HcalEndcapNIslandProtoClusters:localDistXY                => 150,150               VS 1.5,1.5
HCAL:HcalEndcapPInsertIslandProtoClusters:localDistXY          =>                       VS 1.5,1.5
HCAL:HcalEndcapPInsertRecHits:samplingFraction                 => 0.998                 VS 1
HCAL:HcalEndcapPIslandProtoClusters:localDistXY                => 150,150               VS 1.5,1.5
tracking:CentralCKFTrajectories:Chi2CutOff                     => 15                    VS 50

New issue+PR or fix them here?

[wdconinc]

HCAL:HcalBarrelIslandProtoClusters:localDistXY                 => 150,150               VS 1.5,1.5
HCAL:HcalEndcapNIslandProtoClusters:localDistXY                => 150,150               VS 1.5,1.5
HCAL:HcalEndcapPIslandProtoClusters:localDistXY                => 150,150               VS 1.5,1.5

That just can't be right. Do we really want localDistXY to be 150 cm?

[wdconinc]

Remaining (along with the HCAL localDistXY above) are

BEMC:EcalBarrelImagingRawHits:readoutClass                     => pixel                 VS
BEMC:EcalBarrelScFiRawHits:energyResolutions                   =>                       VS 0
BEMC:EcalBarrelScFiRawHits:readoutClass                        => light_guide           VS

[c-dilks]

Now we are down to:

BEMC:EcalBarrelImagingRawHits:readoutClass            => pixel                 VS
BEMC:EcalBarrelScFiRawHits:energyResolutions          =>                       VS 0
BEMC:EcalBarrelScFiRawHits:readoutClass               => light_guide           VS

EEMC:EcalEndcapPInsertIslandProtoClusters:localDistXY => 10,10                 VS 1,1
EEMC:EcalEndcapPIslandProtoClusters:localDistXY       => 10,10                 VS 1,1

HCAL:HcalBarrelIslandProtoClusters:localDistXY        => 150,150               VS 1.5,1.5
HCAL:HcalEndcapNIslandProtoClusters:localDistXY       => 150,150               VS 1.5,1.5
HCAL:HcalEndcapPInsertIslandProtoClusters:localDistXY =>                       VS 1.5,1.5
HCAL:HcalEndcapPIslandProtoClusters:localDistXY       => 150,150               VS 1.5,1.5

I suppose EEMC and HCAL differences are caused by #433.

[wdconinc]

Now we are down to:

 BEMC:EcalBarrelImagingRawHits:readoutClass            => pixel                 VS
BEMC:EcalBarrelScFiRawHits:readoutClass               => light_guide           VS

Afaik this value is unused, and pixel or light_guide are not even classes. As introduced by e6f1f93, @faustus123, does this do something new in EICrecon?

BEMC:EcalBarrelScFiRawHits:energyResolutions          =>                       VS 0

I'd argue this is not right. Every other calorimeter has a non-zero energy resolution term, so this may need to be {0,0.02,0} to allow fair comparison with the other barrel calorimeter. @mariakzurek?

EEMC:EcalEndcapPInsertIslandProtoClusters:localDistXY => 10,10                 VS 1,1
EEMC:EcalEndcapPIslandProtoClusters:localDistXY       => 10,10                 VS 1,1

✓ I'll fix those. I keep thinking in mm, not cm, as the basic unit.

HCAL:HcalBarrelIslandProtoClusters:localDistXY        => 150,150               VS 1.5,1.5
HCAL:HcalEndcapNIslandProtoClusters:localDistXY       => 150,150               VS 1.5,1.5
HCAL:HcalEndcapPInsertIslandProtoClusters:localDistXY =>                       VS 1.5,1.5
HCAL:HcalEndcapPIslandProtoClusters:localDistXY       => 150,150               VS 1.5,1.5

I suppose EEMC and HCAL differences are caused by #433.

There is even more wrong here than meets the eye. The intention is that only one of the localDistXY etc pairs is supposed to be set. If multiple are set, then the last one in the arbitrary list in the algorithm is used, in this case dimScaledLocalDistXY, and that one should not have dimensions. So, I'm going to punt on fixing Hcal clustering for this PR.

[wdconinc]

Juggler didn't have pixel either, https://eicweb.phy.anl.gov/EIC/juggler/-/blob/main/JugReco/src/components/ImagingPixelReco.cpp#L45.

[faustus123]

Afaik this value is unused, and pixel or light_guide are not even classes. As introduced by e6f1f93, @faustus123, does this do something new in EICrecon?

I cannot recall where those came from. I just poked around in my old work directories, but am not seeing "pixel" show up in this context. It is a mystery to me. So is this something that should be left as an empty string?

[wdconinc]

So is this something that should be left as an empty string?

Hits refer to CellIDs. The CellID includes bits that define the detector. The detector can have multiple readout collections, but typically has only one. When there are multiple readout collections for a detector, then the readoutClass field is used to choose which readout collection to use. That does not apply here, and I think an empty string should be used.

[faustus123]

OK. Since they are already empty strings in the library code, then I take it no action is needed. You just wanted verification that there was not a reason why they should actually be set to "pixel" and "light_guide". Let me know if this is incorrect and you need more from me.

[mariakzurek]

I'd argue this is not right. Every other calorimeter has a non-zero energy resolution term, so this may need to be {0,0.02,0} to allow fair comparison with the other barrel calorimeter. @mariakzurek?

This is right only for the Imaging Calorimetr; I guess it has been copied for other calorimeters from the imaging layers. It's 2% flat resolution on single pixel readout for AstroPix. @Chao1009 @sly2j

[wdconinc]

Verified with @mariakzurek that BEMC:EcalBarrelScFiRawHits:energyResolutions should be 0 as in the cpp file right now, not empty string '' as in the reco_flags (since that is not actually a specification of a value, just a deferral to the default in the cpp file).

[wdconinc]

So, only remaining issue is:

HCAL:HcalBarrelIslandProtoClusters:localDistXY        => 150,150               VS 1.5,1.5
HCAL:HcalEndcapNIslandProtoClusters:localDistXY       => 150,150               VS 1.5,1.5
HCAL:HcalEndcapPInsertIslandProtoClusters:localDistXY =>                       VS 1.5,1.5
HCAL:HcalEndcapPIslandProtoClusters:localDistXY       => 150,150               VS 1.5,1.5