Skip to content

CaloPID line improvements

Maximilien Chefdeville requested to merge chefdevi_calib_improvements into run2-patches

Modification of 2 calo lines. Before/after rate comparison done without re-running the calo reco on 2012 data (i.e. commenting out #importOptions("$STRIPPINGSELECTIONSROOT/tests/users/DV-RedoCaloPID-Stripping21.py") in user test script to allow running over more stat in a shorter time. Test with re-running the calo-reco will follow, hence the WIP).

Ds* -> Ds gamma from StrippingCaloPID.py had too high timing due to use of StdLooseDplus2KKPi. Now use StdVeryTightDsplus2KKPi which implement P & PID cut on daughters. These cuts were present at a later stage anyway in my selections, so no signal is lost.

Rate test over 50k with StdLooseDplus2KKPi:

StrippingReport                                                INFO Event 50000, Good event 50000
 |                                              *Decision name*|*Rate,%*|*Accepted*| *Mult*|*ms/evt*|
 |!StrippingCaloPIDDsst2DsGammaLine                            |  0.0280|        14|  1.143|  15.584|
 |!StrippingCaloPIDDsst2DsGammaLine_TIMING                     |  0.0280|        14|  1.143|   0.036|

Rate test over 50k with StdVeryTightDsplus2KKPi:

StrippingReport                                                INFO Event 50000, Good event 50000
 |                                              *Decision name*|*Rate,%*|*Accepted*| *Mult*|*ms/evt*|
 |!StrippingCaloPIDDsst2DsGammaLine                            |  0.0280|        14|  1.143|   2.769|
 |!StrippingCaloPIDDsst2DsGammaLine_TIMING                     |  0.0280|        14|  1.143|   0.035|

-> factor of 5 improvement on timing for same physics output.

D0 -> K pi gamma from StrippingD02KPiGamma.py. This line was heavily pre-scaled (5%) due to the large retention rate (despite a large 2 GeV PT cut on the photon). To reduce the rate and remove the prescale, the D* -> D0 pi+ decay was reconstructed by adding a slow pion. A new stripping line is created: StrippingDst2D0Pi.py, the previous D02KPiGamma line can be ignored in the upcoming stripping campaigns.

Rate test over 50k without reco of the slow pion and 5% prescale:

StrippingReport                                                INFO Event 50000, Good event 50000
 |                                              *Decision name*|*Rate,%*|*Accepted*| *Mult*|*ms/evt*|
 |_StrippingGlobal_                                            |  0.0300|        15|       |   9.801|
 |_StrippingSequenceStreamCalibration_                         |  0.0300|        15|       |   9.787|
 |!StrippingD02KPiGamma                                        |  0.0140|         7|  1.143|   1.679|
 |!StrippingD02KPiGamma_TIMING                                 |  0.0160|         8|  1.250|   1.687|

Rate test over 50k with reco of the slow pion and no prescale:

StrippingReport                                                INFO Event 50000, Good event 50000
 |                                              *Decision name*|*Rate,%*|*Accepted*| *Mult*|*ms/evt*|
 |_StrippingGlobal_                                            |  0.0040|         2|       |  15.793|
 |_StrippingSequenceStreamCalibration_                         |  0.0040|         2|       |  15.777|
 |!StrippingDst2D0Pi                                           |  0.0040|         2|  1.000|   9.276|
 |!StrippingDst2D0P_TIMING                                     |  0.0040|         2|  1.000|   0.040|

-> factor of 10 improvement on retention rate at the cost of increased timing (not sure why). If necessary a prescale of 0.5 could be introduced.

Rate tests with re-running the calo on 2011 & 2012 data will follow.

Edited by Maximilien Chefdeville

Merge request reports