Update on Conversions SCIPP ATLAS Meeting February Bruce Schumm, UCSC/SCIPP
Step I: Reconstruction Have migrated from with two patches to standard (confirmed with Thomas Koffas that this is current best) Reconstruct 50 events of GMSB2 for now – will need to increase by x10 or more soon. Large GMSB1 samples reconstructed in , which (I’m fairly sure) is not appropriate for conversions.
Step II: Find the Conversions Am now using the conversion in the standard EGamma list rather than those in the Conversion container much more filtered. Number of reconstructed conversions down by factor of Step III: Truth Matching Have worked through “track” truth matching for conversions (thanks to code snippet from Thomas). Truth match only if an electron or positron from a photon conversion (1-track conversions) or one electron and one positron from a photon conversion (2-track conversions). 52/85 ~ 60% are truth matched, but…
Step III: Rconstruction Comparison for Truth- Matched Conversions (compare true photons with EGamma reconstructed photon ) Define momentum difference P as P = (p reco – p truth )/p truth Let R be the three-dimensional angle between the true and reconstructed photon. Let O between the difference in the true and reconstructed radial position of the conversion.
3D Angular Separation Closest True Conversion Truth-Matched Conversion GMSB2 Low Statistics
Momentum Difference Closest True Conversion Truth-Matched Conversion GMSB2 Low Statistics
Difference in Radius of Origin Truth-Matched Conversion Next, look with higher stats (150 events 150 events), and separate between 1- and 2-track conversions GMSB2 Low Statistics
GMSB2 3D Angular Separation GMSB2 Higher Statistics All Truth- matched Conversions Two-track Conversions One-track Conversions
GMSB2 Momentum Difference GMSB2 Higher Statistics All Truth- matched Conversions Two-track Conversions One-track Conversions
And try GMSB1… 3D Angular Separation GMSB2 (300 Events) All Truth- matched Conversions Two-track Conversions One-track Conversions
GMSB1 Momentum Difference GMSB1 (300 Events) All Truth- matched Conversions Two-track Conversions One-track Conversions
Some ideas for what to do next… *) Just look at GMSB conversions (almost all set up to do that *) Look at the individual truth-matched electron/positron tracks *) Look at truth-matched pions to see if bremsstrahlung is causing trouble…