Skip to content

Feat/v7 - #6

Open
SimonSPD wants to merge 28 commits into
mainfrom
feat/v7-run2
Open

SimonSPD wants to merge 28 commits into
mainfrom
feat/v7-run2

Conversation

@SimonSPD

Copy link
Copy Markdown
Contributor

V7 -> adaptations for full run2

…ions turn Up/Down variations into lnN yield ratios
… with rescans, singles as cross-check, refit from synced shapes and a continuous ES across 40 GeV
…on from W MC, three-process fractions), payload variants by name, placeholder edges and the luminosity share of a sample in a measurement over several eras
… with a nuisance report, then the limits or the background-only fit
…ss/fail fits with the efficiency as POI, counting cross-check, payload) and equal-data edges from a given cut
…ff_data and eff_mc references, and a grid comparison of two payloads
…e refuses a missing booked class, and the plots draw several reference payloads
…y when the embedded sample never fires the monitoring path, as the 2018 mu tau embedding
… a profile with fewer than five points inside its interval is rescanned on a narrower range, and a best value outside its interval is a problem, as the coarse 2D grid of 2016preVFP Medium/Tight DM0 gave whole-range scans and an unresolved r interval
…ame, bit-identical and 3.6 times faster for the tau-ID MC bookings with about 1000 columns
…g instead of ending the run, as in 2016postVFP Medium/Tight
…minima with --freeze, as PostFitShapesFromWorkspace takes only the floating parameters of the fit result and drew r = 1 and ES = 0
…ing instead of ending the run, as in 2016preVFP Medium and 2017 VVVLoose_not_Medium
…is window, takes the error limit absolute or relative and never lets a counted efficiency outside (0, 1) qualify, as the whole mT < 65 GeV region without a window inflated the counting error 3 to 6 times over that of the fit and merged every Medium decay mode into one bin
…ofile scan with points without likelihood is a problem, as negative MC bins left 2018 Medium DM1 and DM10 above 120 GeV with NaN scan points and a closure fit at 0.67 instead of 0.91
… own base and records that base in the skim contract of the kind, so that the MC and embedding of a re-production are skimmed beside the data of the first production while a configuration with a single base behaves and is recorded as before
…tegory key of only one payload instead of failing, so that a fake-factor payload with W+jets fractions compares with one without
…at is not positive and records the measured one, as a negative factor from few data in the scale regions made every ttbar fake factor and non-closure 0 (2016preVFP et of the v7b measurement), while a positive one cancels in the corrected fake factor
…of every process and symmetrise the pairs of Analysis.symmetrised_variations that are one-sided or more asymmetric than Analysis.asymmetry_ratio, bin by bin around the nominal, with every change logged and listed in template_changes.json
… parameter robustHesse dropped from the covariance) is written as null and reported as undefined in background_fit.json and .md and in the log, never as 0
… a site whose page switches between the galleries of its subdirectories with an era switch that keeps the selection where the other era has it, puts the era into the link and defaults to the most recent era, from file:// and over HTTP
… run filled them, also when the yield variable is not plotted, and the shown era clicked while another era loads stays shown
@SimonSPD SimonSPD self-assigned this Oct 25, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant