WG2: Higgs CP properties through EFT fits


Higgs WG2 ttH CP discussion


ATLAS (Hongtao Yang)


  • List of CP sensitive analysis:

    • H→ττ: probe CP of V-H interaction in VBF (definition of “optimal observable”)

    • H→WW: probe CP mixing in effective g-H interaction using ggF+2-jet events

    • Reinterpret simplified template x-section (STXS) measurements in SMEFT framework: fitting CP-odd Wilson coefficients to data

      • No observables sensitive to CP-even/CP-odd interference are included in

H->ZZ analysis. Sensitivity through CP-odd^2 operators, fixing CP-even contribution to SM. Also in the ttH->gg analysis, no observables sensitive to interference.

  • From: ttH/tH, H→γγ CP analysis

    • Use “Higgs Characterization model” https://arxiv.org/pdf/1407.5089.pdf

    • Fit always in terms of (kappa_t, alpha), but include ggF/Hgg in some cases to help constrain (for instance when kappa_g is parameterized using kappa_t and alpha)

    • ttH with unresolved loops: other kappas (kappa_V, kappa_b) fixed to SM. Doesn’t affect the measurement directly, but does through the total H width (kappa_b in particular). However this is an hypothesis that is generally needed in the kappa framework.

    • Use κγ vs κg results from 80 fb-1 Higgs coupling combination removing the tH/ttH(γγ) categories in the fir  to constrain H→γγ and ggF rates.

    • Fit resolved Hgg/ggF loops: assume potential new physics in H→γγ/ggF is only in t-H coupling, and can be parameterized as function of α and κt (Ellis et. al. JHEP 04 (2014) 004) such that H→γγ/ggF rates could be used to constrain α together with ttH and tH.


CMS (Meng Xiao)


  • ttHgg:

    • D0- is a BDT discriminant, similar to an optimal observable for the CP-odd^2 vs. CP-even^2 terms. Measurement regions are 3 bins of D0- values, with different CP-even purities.

    • DCP is not fit (it would need to tag the charge of the tops, OK in dilept but too small BR, not easy in semilept, impossible in hadronic (need the charge of light jets) dropped DCP in this analysis only use D0-

  • ttH (4l + gg) [HIG-19-009]:

-   fCP and k/k~ scans

-   ttHgg all other prod modes fixed to SM

-   ttH4l  all prod mech floated

      -   ttH (4l and gg) [HIG-19-009]

-   when combine ttH and ggF in gg + 4l all loops are resolved and kb = kt floated in the H total width, Hgg loop expressed in terms of k, k~

-   couplings Higgs to gauge bosons floated


Theory (Henning Bahl)


  • The main sensitivity in the analysis of 2007.08542 comes from the relative rates of ttH and tH. This explains why the limits are weaker than those set by ATLAS or CMS analyses, which also include kinematic information for tH and ttH. 

  • 2D likelihood are model dependent, but CMS measurement of fCP is not significantly sensitive to c_V or the ggF rate, so could be included in principle.

  • Could report a likelihood of (ct~, ct,, cV) (or (ct~, ct) while profiling cV) but not clear if this is useful given that assumptions about e.g. kappa_b are also included.

  • Would be useful if the scenarios explored by ATLAS and CMS (which parameters fixed, and to what) would be coordinated. This was not done in this round since these were new analyses, but can be worked on for future iterations. Anyway the main assumptions (e.g. c_V, kappa_b) are already common.

  • The recent ATLAS ttH STXS measurement in ptH bins could bring interesting information, in particular if ttH and tH components are separated (or even ttH, tHqb and tHW). (STXS of course loose in sensitivity with respect to what you can get with a BDT)

  • Additional measurements sensitive to CP-odd (optimal observables, etc.) would also be useful. However difficult since one needs to tag the charge of the top quarks. Should some follow-up discussions be held how to move forward with this ?

  • Would be useful to investigate the interplay with EFT efforts, which is currently limited to CP-even but can in principle also be extended to cover CP-odd interactions. Could be generally useful to design Higgs analyses taking account CP-odd sensitivity, in particular tH.



There are minutes attached to this event. Show them.