Definition of the 1-jet selection

Some sensitivity, probably coming mostly from the high eta region. In CMS the selection uses a ME discriminant, in ATLAS a BDT, both should favor the high-eta region: we could either extrapolate to the full phase space or define separate bins. Modeling ggF at high eta could be problematic, which would point towards minimizing extrapolation. No definite numbers yet on where we have sensitivity, so no discussion of the binning today.

Binning in mjj:

• Options A and B: focusing on the [350, 700] bin (Option A) and the [250, 500] bin (Option B) : the latter allows to measure a wider range of mjj, however a) most analyses have sensitivity starting at ~400 GeV or above, so more extrapolation if we measure down to 250 GeV and b) no overriding interest in measuring in the [250,350] interval. => Go for Option A, with first mjj boundaries at 350 and 700 GeV.
• 2j/3j splits: could be less useful at very high mjj, since ggF contamination becomes smaller. Also the fraction of 3j events in VBF seems to decrease with mjj, whereas it increases for ggF. However keep the 2j/3j splits everywhere at least for theory uncertainty computation.
• Revisit the pTHjj cut for the 2j/3j separation (currently 25 GeV) ? Current value is a compromise between the Δφ_{γγ,jj} cut used in H->γγ (which corresponded to lower values of pTHjj) and the pTHjj and pTjet cuts used in H->WW, H->ZZ. The optimal value at lower mjj (~400 GeV) is probably at 25 GeV or even lower, but since resummation effects in ggF go as log^{^2}(pTHjj/mjj), higher pTHjj values could work better at higher mjj. Could have mjj-dependent cut (e.g. cut on pTHjj/mjj) but seems to complex and not trivial to include in analyses. So stick to the current 25 GeV value.

pT(H/j) cut

• Where to apply the cut ? Can be done at the top of the selection as for the pTj1 cut now, but then strongly correlated with the high-pTH selection in ggF => strong correlation in the measurements, and conceptually does not bring much new information. Can instead apply the cut within the VBF-topology region (mjj > 350 GeV), where correlation with ggF is smaller and we have an independently interesting measurement (high-pT together with high mjj). This also matches well an existing signal region in the ATLAS H->ττ analysis (pTττ > 140 GeV and mjj > 400 GeV). In Powheg, (pTH > 200 GeV && mjj > 350 GeV) selects ~4% of total VBF, whereas pTH > 200 GeV is ~6% so this does capture the bulk of high-pTH VBF. However it is already small enough that further bin splittings are probably not useful, so implement as a top-level selection within the mjj > 350 GeV region, without further splits in mjj or 2j/3j.
• pTH or pTj1 ? For HVV couplings, studies from both ATLAS and CMS show that both variables give similar sensitivity. ATLAS also shows that the cHq3 operator of the SMEFT basis (coupling to an isotriplet quark current) is better constrained by pTj1l. However this operator also has a large impact mjj, which probably provides a larger constraint. Using pTH has the advantage that the VBF selection can match exactly the VBF-topology region defined in ggF (where pTH is used) => in conclusion, use pTH.
• pTH cut in the ggF selection: Currently, separate BSM bins in 1j and >=2j regions, but large theory uncertainties on jet multiplicity in high-pTH region so the pTH cut will be moved to the top level. Do we then need a separate pTH > 200 GeV selection in the mjj > 350 GeV region of ggF ?

Timescale

• A first document will be prepared in the next few days, documenting the new binning only (can be cited in Winter 2019 results)
• Another document on the Summer timescale documenting also the theory uncertainties