First version of BPHY19 and BPhysConversionFinder
This new derivation format (BPHY19) reconstructs radiative chi_c/b -> J/psi/Y(nS)(mu+mu-) + gamma decays using photon conversions. This is channel has effectively been closed in Run 2 due to technical problems with re-running the track based photon conversion finder on the xAOD format.
The problem originated from the removal of the first measurement from the track object from the xAOD, a piece of information that the standard track based conversion finding code relies upon (though only from a technical point of view). I have written a new conversion finding wrapper algorithm "BPhysConversionFinder" (loosely based on “Reco_V0Finder”) which is essentially includes a simplified clone of the original top level conversion finding code (InnerDetector/InDetRecAlgs/InDetConversionFinder) with all reliance on the first measurement removed. The first measurement was only used to estimate the pre-fit vertex position and apply a loose requirement on its radius (which was not used by default in any case). To be clear, all I’ve done is re-implement the main “track loop” which calls the existing “official” track pre-selection, vertex position estimation and fitter tools but I’ve removed the few hard coded requirements on the presence of the first measurement. As a result, this modified version performs almost identically to the original and exhibits resolution (i.e. deltaM for chi_c/b) and reconstruction efficiency very comparable to the Run 1 studies in these channels. An additional cascade fit is also performed with the di-muon and conversion vertices (constraining their masses to the PDG values) to reconstruct the chi_c/b 4-vector.
To facilitate some first Run 2 studies of radiative chi_c/b decays, the BPHY19 format runs this conversion finding algorithm then look for chi_c/b > mu+mu + gamma (conversion) candidates and only save events which include a candidate (i.e. m(mumugamma) - m(mumu) < 2 GeV), keeping the format very slim and avoiding a significant overlap with BPHY1.
The output is around 0.8% of the input AOD (for full 2017 physics_BphysLS data run, 340030).
Thanks,
Andy
JIRA documenting development: https://its.cern.ch/jira/browse/ATLBPHYS-103