- From Aleksander Filip ?arnecki to Everyone: (11:12 AM)
?The main problem for PDF extraction at LHC is that one can not use the same data for PDF fit and then for BSM searches. The combined approach is needed, when you constrain BSM effects within the PDF fit. This was done by ZEUS in CI analysis of HERA data...?
- From Joey W Huston to Everyone: (11:16 AM)
?It was also done for the inclusive jet cross section at the Tevatron by CTEQ. Note that the PDF fits themselves can tell you if there may be new physics at high pT, as data sets like jet production cover a wide range in x and Q^2 in regions that may contain new physics and regions which probably will not (i.e. low pT and high y).
- From Alessandro Tricoli to Everyone: (11:17 AM)
?To decrease tensions in data sets we also need to work on theory uncertainties in MC generators uses in assessing experimental uncertainties, e.g. jet energy scales
From Maria Ubiali to Everyone: (11:25 AM)
?The study done by ZEUS is similar to the one we did in this paper (https://arxiv.org/abs/1905.05215) and show that indeed a combined approach is needed.
From Joey W Huston to Everyone: (11:26 AM)
?Two-point uncertainties, for example, the difference between Pythia and Herwig in ATLAS, can be dangerous.?ItÕs better to understand the source of the differences, and also to compare to an additional MC, such as Sherpa.
From Alessandro Tricoli to Everyone: (11:27 AM)
?@Joey, exactly this is mostly what I am referring to. Understanding these differences is very important, and needs time and effort
From Mandy to Everyone: (11:29 AM)
?A huge amount of effort was put into correlations/decorrelations of systematic uncertainties for the 8 TeV jet data of ATLAS--- with the end result that nobody (outside ATLAS) actually uses this information!!
From Pavel Nadolsky to Everyone: (11:44 AM)
?@Aleksander, @Alessandro, the in-situ PDF+BSM fits quickly become too unwieldy. In some cases, a feasible workaround approximates the "PDF part" of the fit by a fast technique such as PDF reweighting which retains the correlations between the BSM parameters and PDF-introduced nuisance parameters. ÊSuch approximate BSM+PDF fits are quite fast, and in fact the computer tools for such fast fits (xFitter, ePump, PDFSense) are already available. What is needed for the success of such approaches is developing a consistent methodology for error propagation from the PDF side to the BSM side that does not lose accuracy as a result of approximations made within the fast fitting techniques.
From Maria Ubiali to Everyone: (11:49 AM)
?@pavel @alessandro @aleksander I agree with Pavel that one must be careful with fast fitting techniques, indeed I do not think that reweighting would work. A simultaneous fit ˆ la alphas+PDFs would be more appropriate.
From Pavel Nadolsky to Everyone: (11:57 AM)
?@Maria, indeed, it is easy to get a wrong result with a fast fitting technique if it is not consistent with the full implementation in the global fit. On the other hand, groups like JAM essentially do that by analyzing combined distributions of MC replicas sampled over unpolarized PDFs, polarized PDFs, FFs simultaneously. So it may be interesting to see how much the JAM-like approach can be pushed in other applications.
From Pavel Nadolsky to Everyone: (12:00 PM)
?@Mandy, would it make sense for us to learn how to get the best use of the full information about the systemic errors in the ATLAS 8 TeV jet production? Can it be encapsulated in a few particularly useful results?