Notes on the 5 March 2015 Meeting of the ATLAS/CMS Dark Matter Forum
===========================================================

Agenda: https://indico.cern.ch/event/378495/

As a reminder, agendas and minutes of previous meetings can be found here: https://twiki.cern.ch/twiki/bin/view/LHCDMF/WebHome

A Roadmap for Simplified models of mono-X - Alessandro Vichi
===========================================================

This talk presents a survey of simple models, which have EW or H emission (not as ISR), with sizeable cross-sections. 

1- mono-H models (arXiv: )
2- Z'+2HDM+DM (arXiv: )
3- Z'+inelastic DM (Not available)
4- squark/sbottom + neutralino (MSSM with generic couplings) -> covered by standard SUSY searches

The LHC sensitivity is studied for 8 TeV (done) and 14 TeV (in progress)

1-Vector and scalar mediators, separately (mono-Higgs signatures). 

For both models, the sensitivity is dominated by Direct Detection (DD) if mDM > mZ'/2.
In the mDM<mH/2 range, colliders can provide constraints with invisible H width measurements. 

2-Both a vector (Z_H) and a scalar mediator (S)

If the Z' mixes with Z and with new a scalar, and with SM higgs, it will lead to both mono-H and mono-Z signatures, with branching ratios depending on the dark sector (see 1402.7074). 
In this case, one can restrict the parameter space and consider points that are:
- not constrained by dijets/precision (rho)
- not constrained from direct detection
Mono-Z is generally more powerful than mono-Higgs.

3-Z' vector mediator + inelastic DM

Inelastic DM would evade DD constraints, but a heavy vector resonance would have a sizeable cross-section at colliders. 

The model contains the Z', Dirac fermion chi/chi' with a mass splitting, and an additional scalar mixing with the Higgs boson, connecting the SM and DM sectors. 
The main signatures are mono-Z and mono-H, mono-W is possible but less important. 

Parameters of the model include the mixing angle Z/Z', the gauge coupling. 
Some benchmark points are investigated, giving the exclusion limit vs the cross-section, with different color showing the coupling ratios: the decreasing ratio gDM/gqq determines the exclusion limit. Mono-Z searches are already good at 8 TeV. 

Q: what is the upper bound on the couplings of these models?
A: order unity, dictated mainly by EW constraints and direct searches (see 1402.7074)

Q: Why do you want to have the DM coupling smaller than the qq coupling? 
A: whenever DM coupling is larger than 0.5, model is excluded. But one could relax the assumptions and scan the couplings. 

4-t-channel models in which h and Z are radiated from the squark. 

Mono-Z underperforms wrt dijets+MET, except in the compressed region. If the splitting squark/DM is below 10 GeV, the squark on-shell will decay to invisible and the mono-Z excludes the model for masses < 100 GeV. 

The same analysis could be done for the 3rd generation (sbottom + neutralino) and the Higgs would become more important due to the coupling with mass, but the conclusions on the sensitivity would be the same. 

Q: Why can't we use the 'standard' searches for ISR, when in the compressed region (mono-jet, mono-photon) instead? They may be more powerful. 
A: Agree. 

General question and follow-up: 
If possible, the authors should make the MET distributions available to the Forum, so that extra signal regions can be thought about to avoid missing the models that are uncovered. 

Choice of EW models (summary of last meeting) - Caterina Doglioni
===========================================================

Model categories: 

1. EFTs
    • V via ISR
        • Possible choices: D1, D5, D9 (D8 can be rescaled from D5)
           • Q: is D1 worth investigating? A: Since the model is the same as the monojet but much less sensitive in all V cases,  it can be mentioned in the write-up and left to the analysis to decide whether to use or not. 
        • D5 for mono-W: generate two interference values, rescale the third one
           • Q: Is there a version of Powheg implementing models with ISR boson? A: Not available, not foreseen in the near future.
    • V from direct DM-V interaction
        • collider kinematic differences for Dirac DM operators are small (1307.5604), can generate only one type and rescale
        • in case of H+MET, kinematics are different for different operators (1312.2592)
2. Simplified models
    • V via ISR
        • could have colored scalar t-channel mediator, but monojet still more sensitive --> leave this for later
        • vector boson mediator (with constructive interference) for mono-W seems most promising in terms of sensitivity, may include model for other final states as well for comparison
    • V from specific models
        • Z'_B, Z'_D, Scalar mediator with mono-Higgs/mono-Z signatures - see next talk

Common parameters for mono-photon/W/Z models - Andrew Nelson, Marie-Helene Genest
===========================================================

This is a talk presenting the proposal on the mailing list and twiki. 

##Implementation 
  * *Suggested generator*: MadGraph2.2.2
  * *PDF set*: NNPDF2.3LO (eventually free for the experiment to choose)

##Grid scan

  * Parameters for all models (DM mass) / simplified models (mediator mass/DM mass)
     * mChi/mMed*: start with 7 points in mChi/mMed plane, including one mass point close to the H mass and one high mMed mass point to emulate EFT. We need to understand whether it is more convenient to have them close to the diagonal mMed=2mChi, and whether we want an off-shell mediator (mMed<2*mChi). 

  * Parameters for EFT models:
     * *Lambda scale*: 3 TeV (common choice, but it affects mono-Higgs models)
     * *Operator choices*:
        * D5: low MET, interference for mono-W [check: I thought this was high MET]
        * D9: middle MET range
        * D1: low MET range, even if the sensitivity is low it is complementary to the others

  * Parameters for EWKEWKchichi models:
     * *interference Z/gamma*: no effect on kinematics, choose k1=k2=1
     * Q: correlation with mono-H -> follow thread with Linda, but it should be a different model. 

  * Parameters for completions of EWKEWKchichi models: 
     * there is a heavy H as completion (Follow up: on SVN?)
     * choosing minimum width on mediator being investigated

  * Models with no EWK interaction with the mediator (boson from ISR): 
     * Agreement to concentrate on D5 / vector mediator comparison as mono-W, using the same model and settings for other final states (gamma/Z) for comparison.  


Higgs simplified models - Xiangyan Yu
===========================================================

This talk covers the vector (Z_B) and scalar (S) mediators from [monoH paper], specifically requiring the ZZ decaying leptonically as this is the final state of interest of the group. Other final states are also being investigated. 

A parameter scan is performed on
- couplings
- med mass
- dm mass

Observations: in these models the BR is not consistent with HDecay anymore, as the H can decay to DM as well. This has an impact on the (very small) cross-section. This is also happening when H->DM decays are forbidden, as the final state has a very small branching ratio and we may not be sensitive to this model.

Generator: Madgraph, exclude the photon for simplification. It would be easier and faster if Pythia could be used for the Higgs decays. 

Observation for mDM/mMed: only request one point for mDM and rescale the rest, until w get to mDM>2mMed. This statement needs to be updated after cross-checking the pdgid of the particles involved. 

Scalar mediator updates - Priscilla Pani
=================

Models in MG2.2.2 are now very close (previously: DM was Yukawa for the CMS case), current (not relevant) differences:
- UFO provenance
- automatic calculation of gammaMin in MG

After resolving the coupling difference, ATLAS and CMS analysers are in agreement on mediator pT, DM pT, minimal width and x-sec calculation for both scalar and pseudoscalar mediator. The width of the mediator does not influence the kinematics until very small couplings (gDM~1/veV) and mMed>800 GeV, not in reach for the first year of data -> proposal of not scanning width for the very early data. 
However the width does affect near-threshold on-shell models, with smaller widths giving a less hard MET spectrum -> proposal to use the minimal width everywhere, ignoring effects on threshold. 

Q&A:

Q: Is the narrow width / large mediator mass effect in the MET understood?
A: Yes, it is PDF effects on the mediator pushing events to lower invariant mass, which is reflected on the MET. This can be tested reweighting according to the PDFs. It might be worth studying the onset of this effect with Mmed for future analyses.