Preparatory ATLAS-CMS discussion on MC items related to Higgs boson pair production
ATLAS+CMS meeting on MC HH, 06.02.2019
* Xiaohu (ATLAS):
1. Resonant ggF
Spin-0: narrow-width so far, consider larger widths
Spin-2: theoretical input needed, c=2 dropped + possibly consider model-independent generation (narrow or fixed-width signals)
2. Non-resonant ggF
Current: NLO FTapprox and mHH reweighting. Target: NLO FT (validation in progress) in MG5_aMC@NLO and Powheg, compare both generators for uncertainty.
Non-resonant: vary kappa_V, kappa_2V and kappa_lambda in LO MG5_aMC@NLO
Resonant: NLO Powheg with narrow width
4. Parton shower
ggF non-resonant: Herwig 7.0
ggF resonant: Herwig 7.1 for spin-0, Pythia8 for spin-2
VBF non-resonant: Herwig 7.0
VBF resonant: Pythia8 with dipoleRecoil=on
5. X to SH/SS via ggF
For now using CP-odd A to mimic CP-even heavy S in LO Pythia8 => check at LHE level the differences w.r.t. a fully CP-even set of scalars. No impact in the decay kinematics of the CP-odd object?
6. Kappa_t and kappa_lambda scans
Current: LO MG5_aMC@NLO in which BSM couplings are off + varied kappa_t and kappa_lambda, then linear combination of three samples and mHH re-weighting of the NLO SM sample.
EoR2: dedicated NLO non-resonant package with varied kappa_t and kappa_lambda.
7. EFT frameworks
Need a manageable plan to probe EFT couplings up to dimension 6, an event generator and a grid to scan or benchmark point tests?
8. Questions from CMS:
- VBF non-resonant showering: Pythia8 dipoleRecoil=on improves the VBF jet modelling, studies done in single-Higgs in ATLAS and significant impact on the third jet => CMS wants a reference…
- Choice of parton showers? Herwig 7.1 is used, Herwig 7.0 was used because of time constraints. Pythia8 has too strong pT(HH) tails at LO=> CMS asks about the tune.
- EW NLO corrections? No
- Which NLO top-mass generator will be used, how will they be compared? Negative-weight issues favour Powheg.
* Alexandra (CMS):
1. Resonant - ggF and VBF
narrow resonances: spin-0 -> H and radion, spin-2 -> KK-graviton
otherwise, scan the width (to be done) => agree with ATLAS on mass x width grid?
problem: interference with non-resonant SM HH production, line shapes are model-dependent => simplified model to start with?
2. Resonant X -> SH/SS
Only SS not considered yet => use same as HH, SH within a NMSSM model
3. Non-resonant ggF
EFT ggF privately provided by E. Vryonidou, shape benchmarks, LO and can vary the 5 EFT couplings.
EFT NLO + top mass: private version of Powheg allowing varied kappa_lambda and kappa_t, but not all EFT operators.
SM ggF : LO samples produced so far. NLO samples are so far integrated but done in the previous Powheg-box version. For the latest theory model, working on the Powheg integration and only recently received the MG5 model
4. Non-resonant BSM:
Two methods with pros and cons, can be used for NLO with top-mass. Reweighting method favoured for NLO BSM scans, but can use the same method as ATLAS for klambda-kt scans.
5. Non-resonant VBF:
Both SM and EFT with aMC@NLO in LO mode
Six random points chosen to scan the parameter space, shape benchmarks? common set of points?
What is the target of VBF non-resonant analyses (i.e., which interpretation / results) and modelling of samples? ATLAS and CMS both plan to have samples for different coupling values, but for finer coupling scans a procedure is needed to model a possible coupling combination (as for klambda in ggF), but the problem is made harder by the presence of 6 components instead of just 3.
6. Parton shower:
7. EFT frameworks:
- shape benchmarks
- specific scans, but where to look, what (2D) scans to make? kappa_lambda and kappa_t for sure, but we could agree on others
- single-Higgs also affected by the same EFT parameters and how to consider VBF in scans?
- suggestion to deliver complete likelihood in terms of HH
9. Questions from ATLAS:
- why Pythia8, which is a bit harder? nothing else was really tested… PS uncertainty not assessed in general, but ATLAS sees a non-negligible impact. CMS claims it’s not needed in resonant (still in search mode). Not tested in HH, but the analysers got in contact with the VBF single H experts where this has been studied to understand what needs to be used for VBF showering.
- CMS: in general, tune and matching should be documented. LO can be made look like NLO by changing the tune…
- spin-2 model seems to be different between ATLAS and CMS: CMS only considered narrow spin 2 signals (they can vary the width but did not make any interpretation on that), ATLAS adjusts c. We need to converge on which model to target… RS gravitons not recommended by theorists (on hold by ATLAS).
- SS/SH generation: ATLAS uses a CP-odd as a CP-even but the kinematics should be roughly the same, CMS uses a different (private) model. Compare LHE files?
- Modelling on HH in non-resonant, different techniques: to be addressed
* For next round: share more technical details, even job-options? Here is a short list with the key differences between the two experiments and open topics to be followed up:
- spin 2 resonant: narrow vs model-dependent, non-narrow width
- SS/SH : full cp-even set of scalar vs cp-odd intermediate scalar
- modelling of HH for klambda variations (re-weighting vs components sum) -> at least compare their pros/cons and accuracy when using the latest NLO simulation
- modelling of VBF for coupling variations
- EFT couplings -> both experiments looking for recommendations in their exploration
- document the precise choices of tunes / configurations / models for comparison