[PWGHF] Fix vector error in candidateSelectorOmegac0ToOmegaPi.cxx

[FDUEnrich]

No description provided.

[github-actions]

O2 linter results: ❌ 1 errors, ⚠️ 0 warnings, 🔕 0 disabled

[vkucera]

Can you please explain the error that is fixed by this PR?

[FDUEnrich]

Can you please explain the error that is fixed by this PR?

Hi Vit. The reason is that if we don't apply ML, std::vector outputMlOmegac would be an empty vector. If I fill outputMlOmegac[0] with other variables(such as Invmass). This will result in out-of-bounds access on a null vector.

[vkucera]

Can you please explain the error that is fixed by this PR?

Hi Vit. The reason is that if we don't apply ML, std::vector outputMlOmegac would be an empty vector. If I fill outputMlOmegac[0] with other variables(such as Invmass). This will result in out-of-bounds access on a null vector.

1. It is not clear to me why you need to store outputMlOmegac[0] in the HfSelToOmegaPi table when the full outputMlOmegac vector is already stored in the HfMlSelOmegacToOmegaPi table.
2. If you really need to store it, you can simply replace outputMlOmegac[0] with outputMlOmegac.size() > 0 ? outputMlOmegac[0] : -1.f.

[fgrosa]

Hi @FDUEnrich @vkucera I would suggest to store directly the vector with BDT scores as done for the other channels (see

O2Physics/PWGHF/DataModel/CandidateSelectionTables.h

Lines 47 to 61 in 4ec9c7a

	DECLARE_SOA_COLUMN(MlProbD0, mlProbD0, std::vector<float>); //!
	DECLARE_SOA_COLUMN(MlProbD0bar, mlProbD0bar, std::vector<float>); //!
	} // namespace hf_sel_candidate_d0
	DECLARE_SOA_TABLE(HfSelD0, "AOD", "HFSELD0", //!
	hf_sel_candidate_d0::IsSelD0,
	hf_sel_candidate_d0::IsSelD0bar,
	hf_sel_candidate_d0::IsRecoHfFlag,
	hf_sel_candidate_d0::IsRecoTopol,
	hf_sel_candidate_d0::IsRecoCand,
	hf_sel_candidate_d0::IsRecoPid);
	DECLARE_SOA_TABLE(HfMlD0, "AOD", "HFMLD0", //!
	hf_sel_candidate_d0::MlProbD0,
	hf_sel_candidate_d0::MlProbD0bar);

) in a separated joinable table. First of all, this gives the flexibility to use binary or multi class classification without changing data model, then it makes easier the inclusion of the OmegaC in already existing common code (i.e. flow task). In this way one does not even care about the size of the vector, even if empty.

[fgrosa]

Hi @FDUEnrich do the last commits include everything you wanted to implement, or will you still add the ML table?

[FDUEnrich]

Hi @FDUEnrich do the last commits include everything you wanted to implement, or will you still add the ML table?
Hi Fabrizio, yes. I'm still testing. I think I can fix this bug first. And add centrality and ML-joined table next time. When the test done I will open Ready for review. Tks