Choose timezone
Your profile timezone:
LHC-SM WG: CMS-ATLAS Jet sys correlations
→
Europe/Zurich
LHC-EW WG: Jets and EW bosons
Link to kick-off discussion: https://indico.cern.ch/event/1681040/
Persons connected: Federico Sforza, Valentina Guglielmi, Mikko Voutilanen, Patrick Louis Connor, Petr Jacka, Mario Campanelli, Francesco Giuli
Very optimistic timeline would be to have a result for September PDF conference -> too optimistic but would be great
Would be important to keep involved also new ATLAS experts (Ota, Anastasia?)
Item 1) Make list of what changed from ATLAS and CMS
Discussion about work done in previous note:
Bogdan and others: going one by one on the uncertainties, this was done by 7 and 8 TeV. It took a lot of effort. There are many components. We don't have the same number of components. We have much more on ATLAS side than CMS side. We went through detector, model, pileup, etc. (5-6 categories). It was impossible to say what is 0% or 100% correlated. Things are similar but not identical. There were some loose cases where we provided a range of values.
We should see what has changed since then. Example, from ATLAS side, we have changed the e/p method for ALTAS JES, this should impact the correlation scheme.
We also switched to PF, not sure if this matter. This impacts the prescriptions. Some recommendations we made in Run 1 may be hard to generalise.
Second remark: the correlations that we derive work for single quantities. What we now aim for is to have correlations for differential quantities.
Valentina: for the next meeting, we should target separately (CMS / ATLAS) to have a list of what is new and what has not changed
Mikko: very small changes from CMS, maybe none
Item 2) most important change from the past correlation note is moving from inclusive observables to differential observables
Discussion
Mikko: use inclusive jet cross section as a test case.
It's a well defined process: we can do a Rivet for both to correct for pt binning and code sizes and see if we can git them at the same time.
Federico: I think that the main goal of this effort is this, as it was triggered by the PDF community
Patrick: we should be careful with this exercise, bc we would be assuming that other effects are understood (and Gaussian), e.g. NP or model
Bogdan: agree, we should do this exercise at the very end as a test. A PDF fit should not be the way to derive the correlations.
Francesco: we tested the impact of correlations on gluon distributions and didn't see much effect from changes.
E.g. 7,8,13 TeV in ATLAS PDFs fit: nominal fit some measurement at R=0.6 and R=0.4, correlation impact was not completely taken into account, but impact was very small (for specific observable)
Bogdan and Petr: mentioned that this could be the final goal of the work, because we could get misleading information about correlation from what we like or not.
Mikko: coming back to the original point, I also don't plan a PDF fit. I plan to only compare the data.
Federico: this goal of this discussion is to provide a benchmark indeed
Patrick: this might be sth for future measurements, where we could try to converge to a common binning, common samples, perhaps even common method
Proposal for (near) future development:
try to run a simultaneous fit of 13 TeV di-jet data of ATLAS and CMS with common theo model and check correlatoin model on chi2 (?)
Discussion could lead to a benchmark of correlation and measurements at the end of the work.
Could lead to improved future measurements
Item 3) some discussion about jet cone studies and QCD analysis
Valentina: in the ATLAS PDF fit, which inclusive jet did you have?
Francesco: we tested 7 (AK6), 8, 13 (AK4) TeV. We varied a few cases, and are now working on a combined fit with AK4 and several observables.
Valentina: going to higher radii, we don't expect large changes of methods.
Mikko: indeed, in fact we don't rederive everything, we mostly recycle the AK4 corrections to AK8, adapting minor effects like pileup and residual corrections
Patrick: we may want to check if there is any significant change related to the choice of the model
CMS perform QCD analysis with R=0.8 to reduce out-of-cone impact , CMS also performs R=0.4 measurements
NB: JES calib is done with R=0.4 with further correction for PU and other MC related corrections.
CMS will provide list of JES uncertainties in both jet-R
Federico: trying to conclude -- we have to get the updated list of JES procedure, then we want to move from inclusive to differential observables. We should start with the same jet radius.
Also need to discuss about generators:
- e.g for phase correction proposal of NLO+PS
- Powheg+Py8, used by ATLAS for measurement, pushing it also for CMS, but tune will be different
- for upcoming analysis that can be used for the future
Timeline: Having list in the coming weeks from ATLAS and CMS and then try to have a common meeting at the end of June/beginning July.
There are minutes attached to this event.
Show them.