# (analysis) note on filtering in the ttbar samples ## Heavy flavour filtering ntuple generation **SOLARBv2**: - the event overlap between the ttbar nonallhad samples (410470) and the HF enhanced samples is taken into account by removing the corresponding events in 410470 - this is done in at Level 2, by setting the flag in the config: `DoTTbarBfilter true` - this feature was added to the `run_level2.py` script on March 15th, in this [commit](https://gitlab.cern.ch/ls2multib/ntupler/-/commit/c72411eaf6f383cb22f769482fc2a20261451ed5) - **test this**: Confirmed: when looking into the merged root files, the `ttbar_lep` sample only includes events that have the `TopHeavyFlavorFilterFlag == 0` ![](https://i.imgur.com/eV3lva5.png =x300) --- ntuple generation **SOLARB2121440-pc1**: - here the event overlap removal is not applied - **test this**: Confirmed: the events in the `ttbar_lep` root file have event with different `TopHeavyFlavorFilterFlag` ![](https://i.imgur.com/KoyW8QX.png =x300) ## Dilepton filtering - Dilepton filtering is not applied in either SOLARBv2 or SOLARB2121440-pc1. - **test this** on SOLARBv2: by checking the variable `truth_top_dilep_filter` on the `ttbar_dil` sample (some events do indeed have value 1): ![](https://i.imgur.com/FiH0Sc0.png =x300) ## Notes on ntuple-generations (add to README): ### for SOLARBv3p1 - Event overlap removal is applied considering both HF filtering and dilep, i.e. to get a fully inclusive sample, all ttbar samples must be used (similar for 1l): - ttbar_lep_mc16*_SOLARBv3p1_2l.root - ttbar_dil_mc16*_SOLARBv3p1_2l.root - ttbar_ljets_BBFilt_mc16*_SOLARBv3p1_2l.root - ttbar_ljets_BFiltBBVeto_mc16*_SOLARBv3p1_2l.root - ttbar_ljets_CFiltBVeto_mc16*_SOLARBv3p1_2l.root - ttbar_dil_BBFilt_mc16*_SOLARBv3p1_2l.root - ttbar_dil_BFiltBBVeto_mc16*_SOLARBv3p1_2l.root - ttbar_dil_CFiltBVeto_mc16*_SOLARBv3p1_2l.root ### for SOLARBv2 - Event overlap removal is applied only considering HF filtering, i.e. the sample called "lep" includes both ljets+dil events (complement to the filtered events): - do not use the `ttbar_dil` sample without manually removing the overlap (can be done with `truth_top_dilep_filter` flag) - recommended set of samples (similar for 1l): - ttbar_lep_mc16*_SOLARBv2.1_2l.root - ttbar_ljets_BBFilt_mc16*_SOLARBv2.1_2l.root - ttbar_ljets_BFiltBBVeto_mc16*_SOLARBv2.1_2l.root - ttbar_ljets_CFiltBVeto_mc16*_SOLARBv2.1_2l.root - ttbar_dil_BBFilt_mc16*_SOLARBv2.1_2l.root - ttbar_dil_BFiltBBVeto_mc16*_SOLARBv2.1_2l.root - ttbar_dil_CFiltBVeto_mc16*_SOLARBv2.1_2l.root ### for SOLARB2121440-pc1 - No event overlap removal is applied, i.e. the "lep" sample is a fully inclusive sample - only use the other samples when manually removing the overlap - recommended set of samples (similar for 1l): - ttbar_lep_mc16*_SOLARB2121440-pc1_2l.root