Removing m_allCells from CaloTopoClusterFCCee algorithm members

[kjvbrt]

The m_allCells unordered map is to my mind unnecessary as it does not provide any boost in performance, and introduces compulsory zeroing of the map before every event.
Little benchmark on 100 evt:

Runtime:
before [s]: 8.456, 8.603, 8.489, 8.452, 8.661
after [s]: 8.492, 8.578, 8.548, 8.398, 8.568
average before[s]: 8.5322
average after[s]: 8.5168

Also, fixing issue/crash when running over ddsim generated sample(s).

[BrieucF]

Thanks Juraj, tagging also @giovannimarchiori and @gartrog

[giovannimarchiori]

Hi @kjvbrt ,
since you're touching this code, could you please have a look at why the positions of topoclustered cells in the output root file are in cm while those of other hits/cells/cells after SW clustering are in mm?
I suspect it's this line in CaloTopoClusterFCCee.cpp
newCell.setPosition(edm4hep::Vector3f(posCell.X(), posCell.Y(), posCell.Z()));
where one should divide the 3 coordinates by edm4hep::mm (which is = 0.1)?

Thanks a lot!
Giovanni

[kjvbrt]

Hi @giovannimarchiori,
good point, the cell position comes from DD4hep function, which uses cm as base unit. Using dd4hep::mm ...

What about the position of the cluster itself? Is it OK to have it in cm or should I change that as well?

Best,
Juraj

[giovannimarchiori]

right, also the cluster position is in different units for topoclusters (cm) and sw clusters (mm) so would probably a good idea to harmonise and divide cluster position by /mm here

[giovannimarchiori]

please divide all 3 coordinates by dd4hep::mm

[giovannimarchiori]

Hello! May I ask what's the status of the review of this PR?

Thanks, Giovanni

[BrieucF]

After homogenization of the units it looks good to me!