PWGHF: Ds-h correlation, effciency processes modified

[scattaru]

Dear all,
I have modified the processes for evaluating both the candidate and associated particle efficiencies in order to consider at the kinematic level only those collisions which are reconstructed with the same selections applied on data.

Let me know for any comments,
Cheers,
Samuele

[alibuild]

Error while checking build/O2Physics/o2 for 74cc15c at 2024-07-31 14:50:

## sw/BUILD/O2Physics-latest/log
/sw/SOURCES/O2Physics/7089-slc7_x86-64/0/PWGHF/HFC/Tasks/taskCorrelationDsHadrons.cxx:559:13: error: unused variable 'posZReco' [-Werror=unused-variable]
/sw/SOURCES/O2Physics/7089-slc7_x86-64/0/PWGHF/HFC/Tasks/taskCorrelationDsHadrons.cxx:561:13: error: unused variable 'posZGen' [-Werror=unused-variable]
ninja: build stopped: subcommand failed.

Full log here.

[fgrosa]

Hi @scattaru, apart from some minor comments, I have a substantial one about the logic of your implementation. You first loop over reconstructed collisions, you get the corresponding MC collision and then you loop over the generated particles associated to the MC collision. However, in case of split vertices you will have double counts (because you will have multiple times the same MC collision and hence you will loop multiple times on the same generated particles). I would rather reverse the logic and loop over the MC collisions, check if they have at least a reco collision, and in this case loop over generated collisions.

[scattaru]

Hi @scattaru, apart from some minor comments, I have a substantial one about the logic of your implementation. You first loop over reconstructed collisions, you get the corresponding MC collision and then you loop over the generated particles associated to the MC collision. However, in case of split vertices you will have double counts (because you will have multiple times the same MC collision and hence you will loop multiple times on the same generated particles). I would rather reverse the logic and loop over the MC collisions, check if they have at least a reco collision, and in this case loop over generated collisions.

Hi @fgrosa, thanks a lot for the suggestions! You are right, I have not considered this case of the split vertices. I have reversed the logic in order to loop on the generated collision and then take the reconstructed collision. I have defined a configurable removeCollWSplitVtxto have also the possibility to explicitly remove the case in which for one generated collision there are more than 1 associated reconstructed collisions, so we can check what happen if we consider or not also those split collisions.

[fgrosa]

Hi @scattaru, apart from some minor comments, I have a substantial one about the logic of your implementation. You first loop over reconstructed collisions, you get the corresponding MC collision and then you loop over the generated particles associated to the MC collision. However, in case of split vertices you will have double counts (because you will have multiple times the same MC collision and hence you will loop multiple times on the same generated particles). I would rather reverse the logic and loop over the MC collisions, check if they have at least a reco collision, and in this case loop over generated collisions.

Hi @fgrosa, thanks a lot for the suggestions! You are right, I have not considered this case of the split vertices. I have reversed the logic in order to loop on the generated collision and then take the reconstructed collision. I have defined a configurable removeCollWSplitVtxto have also the possibility to explicitly remove the case in which for one generated collision there are more than 1 associated reconstructed collisions, so we can check what happen if we consider or not also those split collisions.

Thanks @scattaru!