Choose timezone
Your profile timezone:
Fiducial/STXS subgroup meeting
→
Europe/Zurich
4/3-001 (CERN)
Intro:
- Please keep announcing the meeting on the full WG2 mailing list
STXS Bin definitions:
- ATLAS stage-1 bin merging:
-> So far ATLAS merging bins with >~80% correlation, some question on whether correlations are well-reproduced.
-> Will try to push a bit more in the future, if numerical problems don't show up.
-> Useful to check model interpretions both in context of covariance matrices and with full-workspace implementations, to see if the former agrees well.
- ggF/bbH : fully merge or keep bbH separated within the ggF categorization ? Up to experiments...
- Low-pT bin:
-> Problem of compatibility with the jet binning, difficult to control theory uncertainties.
-> Differential pT distributions in 4l/2γ would be more sensitive anyway; however hard to combine this with other measurements....
-> Could split 0-jet into <>10 or 15 GeV.
-> Some discussion about "touching" also 1-jet bin (as 1-jet events with jet out of acceptance contribute to 0-jet bin).
- Boosted ggF and ttH:
-> ggF bin boundary at 500 GeV to be adjusted to value where experiments starts their analyses - expect this value to be similar between ATLAS & CMS
(currently 450 GeV is where the CMS analysis starts, need to check where ATLAS will be sensitive). However sensitivity starts at ~600 GeV.
-> The exact ttH bin boundary need still to be understood (not clear if 200 GeV is the best choice).
STXS VBF/VH uncertainties
- How to choose between options 1 and 2 ?
-> Option 1 for VH, option 2 for VBF ?
-> Option 1 better for the case where we know well the inclusive XS, option 2 better when we know better a specific region (e.g. high-mjj VBF)
-> For VBF, option 1 would make the uncertainty much larger
- Discussion with A Karlsberg (related to the uncertainty correlations in mjj, and weather we should go with option 1 or 2 for VBF]:
-> What is needed from theorists ? e.g. know how correlated are uncertainties across mjj. First -- correlated or uncorrelated across for high/low mjj ?
-> First guess: should be highly correlated across mjj (A Karlsberg).
-> Delta_hard: anything not related to Sudakov, so not universal across VH/VBF.
- Need to interact with the VBF and VH experts for feedback on proposal and conclusions...
Presentation of results:
- How do deal with non-Gaussian behavior ?
-> Need to check how much non-Gaussian effects actually affect the interpretations.
Helpful to try out some interpretation exercises, comparing cov matrix with full workspace.
-> Should still follow up on "simplified likelihoods" to parameterize non-Gaussian effects directly
- Timescale for tools e.g. for kappa framework ? next 3-4 months, to match release of Stage-1 results.
- Report all (prod,decay) STXS (30*5), or just STXS + BR ratios (30+5)?
The latter is more compact, but more correlated and probably less Gaussian...
There are minutes attached to this event.
Show them.
