Choose timezone
Your profile timezone:
BOOST13 report: WG1 meeting
→
Europe/Zurich
CERN
CERN
Description
Second meeting of WG1 for the Boost13 report.
Minutes of the 2nd WG1 meeting
Present: Ben Cooper, Andrew Larkoski, Roman Kogler, Steve Ellis, James Ferrando, Nhan Viet Tran, Andreas Hinzmann
- Andrew presented a review of the 2012 substructure measurements from ATLAS and CMS
- Slides added with a list of all the paper references.
- 2012 Measurements include pile-up subtraction, Top tagging performance, jet charge, Q-jets, q/g tagging (see list of papers).
- None of these recent 2012 measurements have been unfolded, and won't be on the timescale of the 2013 report.
- Can we explore the consistency of the ATLAS and CMS measurements, make statements on how they differ, and why this might be?
- Related to the above: study at generator-level how differences in e.g. the jet algorithm or kinematic regime used affect the observables. Does the MC modelling depend on the choice of jet radius/kinematics? Is the discrimination better or worse with differing jet radius? How do the ROC curves change?
- Roman gave a quick introduction to the current Rivet-based tools - how to obtain the code, use the repository etc. See Romans slides and email to boost13-report@cern.ch.
- Andrew presented some slides for discussion on the study areas of WG1
- ROC curves can be central to WG1 studies. Can use these to test:
- Dependence of discrimination power on jet algorithms/kinematics (see above).
- Correlations: does the ROC curve improve when we combine observables/taggers? If not, significant correlations exist.
- MC modelling. We don't have data measurements of ROC curves. But differences in ROC curves for different MC can tell us how sensitive we are to sub-leading physics. Can these be used as an MC tuning tool?
- Potentially we can produce a wish-list of unfolded measurements to aid in MC tuning in the future.
- Quark/gluon tagging is a good place to start with these studies.
- Produce two sets of samples: one pure gluon-initiated jets and one pure quark-initiated, for a range of q/g pT.
- Reconstruct the q/g discrominating variables and ROC curves.
TASKS:
- Ben, James to follow up with Deepak, Karl and Andy B as to what substructure observables are already implemented into Rivet.
- Compile a list of what variables we want that are not already available (for the q/g studies initially) and implement in the github code.
- Andrew and Marat to follow up in generating the necessary QCD samples for q/g studies.
Present: Ben Cooper, Andrew Larkoski, Roman Kogler, Steve Ellis, James Ferrando, Nhan Viet Tran, Andreas Hinzmann
- Andrew presented a review of the 2012 substructure measurements from ATLAS and CMS
- Slides added with a list of all the paper references.
- 2012 Measurements include pile-up subtraction, Top tagging performance, jet charge, Q-jets, q/g tagging (see list of papers).
- None of these recent 2012 measurements have been unfolded, and won't be on the timescale of the 2013 report.
- Can we explore the consistency of the ATLAS and CMS measurements, make statements on how they differ, and why this might be?
- Related to the above: study at generator-level how differences in e.g. the jet algorithm or kinematic regime used affect the observables. Does the MC modelling depend on the choice of jet radius/kinematics? Is the discrimination better or worse with differing jet radius? How do the ROC curves change?
- Roman gave a quick introduction to the current Rivet-based tools - how to obtain the code, use the repository etc. See Romans slides and email to boost13-report@cern.ch.
- Andrew presented some slides for discussion on the study areas of WG1
- ROC curves can be central to WG1 studies. Can use these to test:
- Dependence of discrimination power on jet algorithms/kinematics (see above).
- Correlations: does the ROC curve improve when we combine observables/taggers? If not, significant correlations exist.
- MC modelling. We don't have data measurements of ROC curves. But differences in ROC curves for different MC can tell us how sensitive we are to sub-leading physics. Can these be used as an MC tuning tool?
- Potentially we can produce a wish-list of unfolded measurements to aid in MC tuning in the future.
- Quark/gluon tagging is a good place to start with these studies.
- Produce two sets of samples: one pure gluon-initiated jets and one pure quark-initiated, for a range of q/g pT.
- Reconstruct the q/g discrominating variables and ROC curves.
TASKS:
- Ben, James to follow up with Deepak, Karl and Andy B as to what substructure observables are already implemented into Rivet.
- Compile a list of what variables we want that are not already available (for the q/g studies initially) and implement in the github code.
- Andrew and Marat to follow up in generating the necessary QCD samples for q/g studies.
There are minutes attached to this event.
Show them.
