LHC-HH Subgroup Meeting

160/R-009 (CERN)



Show room on map
Arnaud Ferrari (Uppsala University (SE)), Javier Mazzitelli (Max-Planck-Institut für Physik, München), Luca Cadamuro (University of Florida (US)), Margarete Mühlleitner (KIT), Ramona Groeber (Università di Padova and INFN, Sezione di Padova)

Discussion on Higgs boson pair production within the LHC Higgs Cross-Section Working Group

Videoconference Rooms
HH Discussion
Magdalena Slawinska
Auto-join URL
Useful links
Phone numbers

Minutes HH meeting January 2020

Introduction -- Arnaud --------------------------------------------------------------------------------------

Next meeting April 27

LHC-HH workshop --> date to be set, feedback from community (specially from ATLAS and CMS)? Spring 2021?

Milestones 2020:

- Les Houches, NNLO vs kappa_lambda, top scheme uncertainties
- VBF studies, HH+2j full mt dependence
- How to make EFT interpretations of single and double H searches, needs to be done with the full HXSWG
- Collect input from benchmarks, HH/SH/SS. Need to decide on input format, and then will ask X->SS/SH BMs
- Review new th models with HH signatures
- Prepare full Run-2 ATLAS+CMS non-resonant HH combination


ggF + 2jets --> sharing MEs samples between ATLAS and CMS to avoid doubling CPU usage?
This could be discussed in the next meeting. Also, do we want to do a VBF combination as well? <---------

Light Yukawa couplings in HH production -- Lina -------------------------------------------------------------

Current bounds on 2nd and 1st gen of quarks are very weak

We consider new fermion coupling from linear SMEFT

With enhanced couplings to light quarks, the qqb->HH channel becomes relevant, becoming larger than ggF even
with enhancements within the current experimental bounds for the yukawas

Shape of Mhh distribution different in qqb wrt ggF.

BRs also change significantly.
In bbyy we get enhancement with the yukawas in the production, but some suppression in the BR

At the HL-LHC for the d quark the bounds could be better than current bounds

In non-linear EFT context, we get independent c_q and c_qq couplings, which can only be probed in HH (or HHj)
In this case we get more sensitivity, also to 2nd generation

2nd generation: possible to prove ccyy without full c-tagging (Kim et al., Perez et al.)
In this way we get better sensitivity than with the current bounds in optimistic scenarios

Could be extended to bbbb, bbcc, cccc final states


Javier: how do the bounds compare to bounds obtained from other measurements in the HL-LHC
Lina: not fair comparison, current searches are model dependent. Also no sensitivity to non-linear effects

Spira: Modification to HDECAY: how did you define the light quark mass for H->yy?
Lina: We used the MSbar scheme (PDG values).
Spira: but HDECAY used a hybrid scheme, so to be consistent you need to modify the NLO corrections as well
Lina: I beleive it was done

Arnaud: is the fact that the peak in mhh is in low values of mhh taken into account in the analysis?
Lina: yes

HH in composite Higgs models -- Maggie ----------------------------------------------------------------------

Composite models: H-V coupling modified ==> unitarity now warranted ==> VV->VV and VV->HH grow
with energy, smoking gun for composite models

Higgs self-couplings are model dependent, important for HH prodcution

Implementation of BRs in eHDECAY

In composite models new couplings HHff~, which contributes largely HH production
Also, new fermions in the loop need to be taken into account

MCHM4 and 5, mostly sensitive to deviations wrt SM in ggF

Are we sensitive to non-zero lambda_hhh?
In general yes, but for MCHM5 and xi=0.5 there is no sensitivity

Experimental measurements challenging (see slide 27)

New physics first in HH? It might be possible, mainly due to the new coupling HHff~

NLO QCD corrections basically independent of xi (but this is in the HTL!)


Lina: can one have non-minimal CHM with another vev with higher scale?
Maggie: yes, you can have non-minimal Higgs sectors (slide 5, composite 2HDM)
Arnaud: sensitive studies mostly just scaling SM XS, and see if you are 5sigma away from it.
For something more advanced, kinematics, acceptance, etc., we need a MC where we can play with
the parameters.
Maggie: we didn't write a MC, we only look at total XS, don't know if anyone working on this
Xanda: you could use tools available with anomalous couplings
Spira: need to work on the proper matching of the Wilson coefficient to the composite scenario.
This has to be worked out explicitly, don't know if it has been done.
Ramona: there is a work from Sally on matching for heavy fermions
Spira: in summary it would be possible to build an approximate solution, it requires some careful work
Arnaud: experimentally we can put limits on anomalous couplings
Follow up on the matching between anomalous couplings and composite Higgs models <-------

HH MC update -- Xanda ---------------------------------------------------------------------------------------

NLO MC kl-kt scan in advanced status.
We can construct any combination of them with 3 samples.
This could be improved (additional sample), not yet done.

Other anomalous couplings: no clear idea of direct interpretation to be done (which
scans/correlations to be done)

Showering: different between CMS and ATLAS. Presciption to correlate (or not) uncertainties? (slide 5)

VBF non-resonant: next level of precision, adding full loop-induced ggF HH+jj
Combination with ggF HH including kl-kt needes to be done

Resonant: X->HH settled, setting up X->SH and X->SS

Th unc due to single-H + heavy flavor (background to bbXX)
Currently, ATLAS HF single H assigns 100% unc


Tania: in resonant production, is the top diagram included with full dependence?
Xanda: we are not doing the interference with the SM
Tania: might be good to take it into account, several studies on interference (see e.g. WP)
Is it important for narrow width? Was it decided it wasn't relevant in the Fermilab workshop?

Arnaud: we should identify what is necessary to make the combination.
No resonant combination planned, waste of time to validate. And will we do a VBF combination?
If we do, what do we need to do? Are there any show-stoppers? <-------

Do we need to harmonize the 3 values of kl used for the MC samples? kl=5 vs kl=20?

Looking back at the HL-LHC HH combination -- Elisabeth ------------------------------------------------------

To keep in mind: HH statistically dominated (even at HL-LHC), especially in the 2 main channels.
Different situation compared to single H!

Assumed no correlation between channels, nor between experiments.
Theory and lumi unc negligible impact.

Significance: Combination vs Sum in quadrature gave very similar results

Summary: simple methods seemed accurate enough for combined projections.

Technical details: see slides


Are there people from other communitie that we should invite?
Arnaud: we were thinking about inviting people from single-H.
Arnaud: do we want to do H+HH CMS+ATLAS combination as well? And ggF+VBF?
We need to probe the interest in both collaborations, if we want to do the
combination, what we want to include, etc. <---------

========== Lunch break ======================================================================================

Single Higgs EFT fits and possible implications for di-Higgs -- Anke ----------------------------------------

Fit in SMEFT context with H + di-boson + EWP data

Inclusion of fermionic operators in the fit loosen the constraints on bosonic operator O_B.

27TeV: adding here as well di-Higgs production (no chromomagnetic operator), and the hhh coupling
Specifically, 10 operators + O_phi3 (no fermionic operators)

95% CL limit \Lambda/\sqrt{f_phi3} > 250GeV (vs 700GeV from single parameter fit), slide 11
Almost a factor of 3 worse


Did you extrapolate 8TeV to 13TeV, and compared to actual limits, to see if it agrees?
Anke: yes, it agrees well. For 13TeV we haven't included any HH data yet.
Julien: do you plan to include NLO corrections?
Anke: yes, we need to see how.
Arnaud: timescale for 13TeV results?
Anke: probably a matter of months.

HH in EFTs at NLO -- Gudrun ---------------------------------------------------------------------------------

Clarification: both SMEFT and HEFT respect SM gauge symmetries.
In HEFT physical Higgs SU(2)xU(1) singlet.

Difference for di-Higgs: in SMEFT there are relations between the couplings (see slide 7)

Chromomagnetic operator: loop suppressed (1/16pi^2 factor)

MC including lambda variation public, HEFT MC within POWHEG in progress

Overview of NLO results in HEFT, see slides.

Obs: the 'dip' characteristich in the mhh distribution for lambda=2.4 (max interference) can
dissappear if other couplings are modified, of can appear for other values of lambda as well.


Spira: don't understand why in a strong interacting theory the chromo would be suppressed?
Gudrun: agree.
(someone knocked my door, I missed the comments, Eleni was talking when I was back
arguing against the argument on the suppression of the chromomagnetic operator)

Spira: the large K-factors are related to strong destructive interference (change in the
position of the minimum in the mhh spectrum when going from LO to NLO), not unexpected.
Gudrun: agree
Arnaud: shape benchmarks are valid at LO. If we go to NLO I guess we need new BMs.
Gudrun: yes, that's in our agenda. We would come up with lower number of BMs.
XX: Do you plan to extend to include other variables (e.g. cos(tita*))
Gudrun: no, the cos(tita*) is extremely flat, we only use mhh

Mass scheme uncertainties in HH production -- Stephen -------------------------------------------------------

Goal of the study to highlight this effect, not to solve it.

Introduction to renormalization schemes, see slides.

HH unc @ LO:
Mhh, peak shifted, big suppression in the tail in the MSbar scheme (slide 10)
From LO to NLO, factor 2 uncertainty reduction in the tail

How much of the effects comes from the yukawa, how much from the mass in the loops? (slide 13)
LO: low mhh, dominated by mass, tail, dominated by yukawa.

Rest of the presentation on uncertainties beyond HH: H*, H+jet

Going forward, possible options:
- Keep computing
- Understand log structure
- Choose one scheme over the other?


Ramona: is the procedure in the paper by Baglio et al. to estimate uncertainties agreed in LH?
Stephen: not really
Spira: what are the arguments against this treatment?
There seem to be no strong arguments

Uncertainties at NNLO FTapprox vs lambda -- Javier <---------------------------------------------------------

Review of the status on QCD corrections for lambda variations and EFT.
New: combination of full NLO with B-i NNLO, in a bin-by-bin reweighting.
A table with XS vs lambda, including scale uncertainties, is provided.


Spira: how exactly is the B-i reweighting defined?
Javier: the full form factors F_triang and F_box are used. For the topology with 2 effective
ggH vertices, the F_triang is used evaluated at a dynamic scale (see paper for details)
Arnaud: are you planning to have this published in a journal besides the LH proceedings? In
order to have it peer-reviewed.
Javier: don't know
Arnaud: is it possible to add -10, -5 and + 10 for the twiki?
Javier: yes
Spira: noted that uncertainties related to missing EW corrections can be large (5-10%),
regardless of the value of lambda (i.e. also for the SM prediction). The corrections
for instance are large in the case of heavy single Higgs production

Input for X to HH benchmarks in models with an extra singlet -- Tania ---------------------------------------

Review of models with extra singlet (similar to previous presentations), see slides

Several 2D plots of the hh XS vs two parameter of the model (sin[alpha], mX, widthX, tan[beta]).
Maximal width ~ 0.02mH, so it's always narrow

hh can be dominant up to mH<600GeV


Arnaud: we have these .tsv files, we should see in the collaboration if they
are useful for interpretation. We try on this model and see if we are missing information,
and provide feedback. <--------



There are minutes attached to this event. Show them.
    • 09:30 09:45
      Introduction 15m 160/R-009



      Show room on map
      Speakers: Arnaud Ferrari (Uppsala University (SE)), Javier Mazzitelli, Luca Cadamuro (University of Florida (US)), Margarete Mühlleitner (KIT), Ramona Groeber (Università di Padova and INFN, Sezione di Padova)
    • 09:45 10:15
      Light Yukawa couplings in HH production (20+10) 30m 160/R-009



      Show room on map
      Speaker: Lina Alasfar (HU-Berlin)
    • 10:15 10:45
      HH in composite Higgs models (20+10) 30m 160/R-009



      Show room on map
      Speaker: Prof. Margarete Muehlleitner (KIT)
    • 10:45 11:15
      HH MC update (20+10) 30m 160/R-009



      Show room on map
      Speaker: Alexandra Carvalho Antunes De Oliveira (Tallinn University (KBFI))
    • 11:15 11:45
      Looking back at the HL-LHC HH combination (20+10) 30m 160/R-009



      Show room on map
      Speaker: Elisabeth Petit (CPPM-Marseille (FR))
    • 14:00 14:30
      Single Higgs EFT fits and possible implications for di-Higgs (20+10) 30m 160/R-009



      Show room on map
      Speaker: Anke Biekoetter (Uni Heidelberg)
    • 14:30 15:00
      HH in EFTs at NLO (20+10) 30m 160/R-009



      Show room on map
      Speaker: Gudrun Heinrich (Max Planck Institute for Physics)
    • 15:00 15:30
      Uncertainties in HH production related to the scheme and scale of the top quark mass (20+10) 30m 160/R-009



      Show room on map
      Speaker: Stephen Philip Jones (CERN)
    • 15:30 16:00
      Uncertainties at NNLO FTapprox vs lambda (20+10) 30m 160/R-009



      Show room on map
      Speaker: Javier Mazzitelli
    • 16:00 16:30
      Input for X to HH benchmarks in models with an extra singlet (20+10) 30m 160/R-009



      Show room on map
      Speaker: Tania Natalie Robens (Rudjer Boskovic Institute (HR))