Photon Conversion Reconstruction in hadronic Tau Lepton Decays
Introduction
To identified the hadronic lepton decay mode correctly it is important to reconstruct all hadronic lepton decay products. Due to interactions between photons from hadronic decay products of the and detector material electron-positron pairs (photon conversions cp. fig a) below) may be produced. These particles lead to additional charged tracks which are reconstructed as tracks. The changed track multiplicity will then either be interpreted as another decay mode or due to inefficiencies in the track reconstruction unphysical track multiplicities will be observed and the candidate will be removed. To avoid such misidentifications an explicit photon conversion identification in the very dense lepton decay cone needs to be investiagted.
Implementation in ATHENA rel 15
Based on the results worked out in my Diplomathesis an explicit photon conversion tagging tool for tracks has been implemented in ATHENA.
First a independent photon conversion identification tool checks for each VxCandidate of the Conversion Candidate Container (these VxCandidates are provided by the default egamma conversion finder tool), if the electron probability of each track is larger than a certain value (def: eProb > 0.9). Then it stores the identified conversions in an additional VxCandidate Container.
Afterwards the second tool (tauConversionFinder) compares the tracks of each identified photon conversion with the associated tracks of the candidate. Due to a refitting procedure which optimises the track fit of the VxCandidate tracks with constraints of the secondary vertex, it is important to use the original track (before refitting) for the comparison with the tracks. If a track belongs to both an identified photon conversion and a candidate, the track is tagged as conversion track.
The corrected track multiplicity, the number of tagged conversion tracks per , and a flag for each track, are stored in the new TauEDM. These variables are stored once for TauJet tracks and once for associated loose tracks.
Results
The first tool which identifies the conversion candidates with a cut on the electron probability (def: eProb > 0.9) achieves an electron purity of 92 % The second tool, tauConversionFinder, stores the overlap of identified photon conversions and candidate tracks.
Trk ID |
Conversion Candidates |
Conversions after ID |
e± |
77.78 ± 0.61 % |
92.41 ± 0.89 % |
p± |
20.19 ± 0.31 % |
7.00 ± 0.24 % |
K± |
2.03 ± 0.10 % |
0.59 ± 0.07 % |
This figure shows the number of tagged conversion tracks per t candidate for Track Seed candidates (a) and for Calo Seed candidates (b). The histogram showing the Track Seeds, has fewer entries than the Calo Seeds. This is due to harder cuts of the Track Seed based reconstruction algorithm and independent from conversion tagging. That feature is in agreement with results from the Tau Working Group. Apart from the number of entries a lower number of overlaps with conversion tracks can be observed, which is evidence that candidates reconstructed from Track Seed algorithm are more robust against conversion tracks than the Calo Seeds. This can be investigated in more detail, analysing the overlap of Track Seed and Calo Seed candidates with conversion candidates and identified photon conversions (see below). 56 % of Calo Seed candidate tracks are also photon conversion candidate tracks (magenta fraction) and 30% are also identified as photon conversion (blue fraction).
Another variable stored in the new TauEDM is the track multiplicity after substracting the tagged conversion tracks. Both the “old“ track multiplicity before the conversion tagging and afterwards are stored. The following plots shows a comparison of the track multiplicities before and after conversion tagging for Track Seeds and Calo Seeds. The black solid lines mark the reconstructed multiplicities without conversion tagging, the black dashed lines the multiplicities after substracting the conversion tracks and in red both cases are shown matched to truth hadronic leptons.
As expected the Calo Seedcandidates seem to benefit much more from the explicit conversion track reconstruction. But to construe the shown plots a more sophisticated analysis of the track multiplicity is needed. Therefore the track multiplicities are matched to true 1 Prong and true 3 Prong t decays and to QCD background (following plot, first row). For a more detailed investigation a differentiation between Prongs with and without neutral pions is done. This is plotted in the second and third row for Track Seeds (left) and Calo Seeds (right).
In general it has to be mentioned that all improvements of the Calo Seed candidates can also be seen for Track Seeds but only as very small improvement. First it is important that a QCD background correction does not accumulate background in 1 or 3 track candidates. This means that the background is not formed into more signal-like candidates due to the conversion tagging. The histograms in the second and third row clearly show that only in case of decay modes with additional neutral pions tracks are tagged.
Talks: Photon Conversion Reconstruction in hadronic Tau Lepton Decays
Documentation
Diploma Thesis: Reconstruction of Photon Conversions in Tau LeptonDecays in the ATLAS Experiment
ATL-PHYS-INT-2009-056; ATL-COM-PHYS-2009-186: Explicit Photon Conversion Reconstruction in Hadronic Tau Lepton Decays