Exotica session notes



Bloc 1: Implications of recent results / Piecing together a signal


=== 9 May 2016 ===

11:00 - 12:00 Existing excesses to be discussed
  1. Diphoton channel bump with mass of 750 GeV, possible consequences are conventional, others not, outline possibilities
  2. 2 TeV diboson
  3. ATLAS SS dilepton with 2b or 3b
  4. ATLAS Z+jets+ETmiss
  5. CMS H → tau mu
  6. B Factories/LHCb B → D(*) tau nu
  7. CMS ttH (multilepton channel)

Discuss whether these anomalies point to certain types of models and require further emphasis on existing or new experimental signatures.


2:00 - 5:00
Diphoton bump at 750 GeV experimental overview [Chou]
  • CMS: Improvements by inclusion of B=0 data + recalibration ==> p-value 3.4 sigma, 1.6 sigma global with narrow width Gamma/M of 1.4 x 10^-4, cross section = 4-5 fb, all for combined 8+13 TeV data
  • Bill Murray toy exercise: possible to have wide resonance reconstructed as narrow but not other way around; this could reconcile preference for wide resonance at ATLAS, but narrow at CMS
  • ATLAS: spin-0 analysis has 3.9 sigma local + 2.0 sigma global; spin-2 analysis has 3.6 sigma local, 1.8 sigma global
  • ATLAS compatibility with 8 TeV re-analysis: 1.9 sigma local for spin-0 analysis, nothing for spin-2 analysis (all 6% width)
  • Open question: What is probability of an excess in CMS given the excess in ATLAS?


Diphoton bump theory discussion [Craig & Strassler]
  • Background studies, CMS only fit to the diphoton mass distribution from the low end; bkg comp is gamma gamma (94%), gam jet and jet jet
  • ATLAS bkg from diphox MC + data estimate for gamma+gamma spin-2 analysis, data-derived estimate of gamma+jet background
  • NNLO calculations by MCFM team including top quark now available for 5f. This shows structure in cross section that is gamma-gamma mass dependent; top quark box adds at most 3% to cross section above mass of ~600 GeV; calculations are in good agreement with background parameterization from ATLAS and CMS (at least checking the dominant gamma+gamma contribution).
  • Signal could be due to gamma+gamma or NOT. For example, could be two pairs of nearby photons from the decay of a low-mass resonance. Could explain differences between ATLAS and CMS.
  • There are 3 non-photon possibilities: consider X → aa with a → gamma gamma, where a is prompt and light or a yields a displaced signature and is not so light (reduces angle in calorimeter and probability to convert), or a yields displaced signature and is so light that it can only decay to e+e-. There are probably holes in experimental coverage here. For example, ATLAS has already searched for 2 displaced jets but not for 1 displaced jet + 1 photon.
  • Production of X must proceed by gluon fusion to explain difference in rates between sqrt(s) of 8 and 13 TeV. Production involves either X-h mixing or occurs through loops. Both scenarios can reproduce observed cross section, not difficult to get a range of different widths.
  • X decay can be through loop again to gamma gamma or direct X → aa and a → gamma gamma, here do not expect large BR of X to gamma gamma directly.
  • X decay to gamma gamma: top loop cannot be involved because BR to gamma gamma would be 10^-5 of BR to ttbar (model independent) ⇒ expect loop to contain new states with SM quantum numbers and mass greater than 375 GeV.
  • Mediator can be X particle or a bound state of two new fermions (bound either by a new interaction or by QCD)
  • Observation of 750 GeV diphoton bump motivates searches in other channels:
    • Zgamma, WW, ZZ, gg with rates relative to gamma gamma that are 0.61 for Zgamma for d_R, u_R singlet representations and 4.5 for Q doublet → this is close to current limits so strong motivation to look for Zgamma resonances; ZZ/gamma gamma of 0.093 and 7.4; WW/gamma gamma = 26 for Q doublet; gamma gamma/gg = 4 x 10^-4 for d_R, 6 x 10^-3 for u_R, 3 x 10^-3 for Q, all compatible with limits on dijet resonances.
    • Mixing of 750 GeV spin 0 (scalar) particle with Higgs -> BR(W_L W_L) / BR(gamma gamma) = 10^5, this forces mixing angle < 10^-3 which is already quite small compared with expectation; this is not a limitation if X is pseudo scalar; note CMS has slight data excess in WW consistent with 750 GeV particle.
    • Interpretations: SUSY, CHM, extra dim; sneutrino + QLD RPV a possibility but quark initiated so not so consistent with cross section at 8 TeV or Dirac gauginos are other possibility but predicts Zgamma/gammagamma = 6 which is under stress
    • spin 2 and ED interpretations are also viable
  • Bump at 750 GeV could be due to:
    • new VLQ pair bound by QCD with VLQ mass of 375 GeV, those VLQ would decay to tc or tu without violating existing bounds.
    • if bound state of scalars with charge = -4/3 (from a triplet), then expect scalar to decay to jj final state
    • If bound state of scalars with charge = 5/3 (from a sextet), then expect scalar decay to jjjj
    • If bound state of scalars with charge -4/3 (from a sextet), decay can be to jj but would violate bounds, however OK to decay to jjjj
    • If bound state of fermions with charge -4/3, then decay to jjj
    • Above cases motivate searches for 3j and 4j resonances that are produced in pairs; all these resonances are narrow
  • Open point: would be interesting to write down all possible representations and corresponding resonance searches compatible with 750 GeV resonance.



=== 10 May 2016 ===

11:00 - 12:00
Diphoton bump spin-2 interpretation [Giddings]
Paper by Steve Giddings and Hao Zhang__http://arxiv.org/abs/1602.02793__.
Consider warped extra-dimensional geometries, as arise generally in higher-dimensional theories (Randall-Sundrum is a model for these).
  • Consider coupling to lowest KK excitation mode of graviton. Parameterize in terms of “simplified model,” with parameters mass and inverse coupling Lambda. In simplest model with universal couplings expect cross section to gamma/gamma = cross section to ee+mumu.
  • Comparison with dilepton cross section limits indicates some tension with measured 750 GeV diphoton cross section but not violent disagreement.
  • If cross section assumed to be 9 fb (between ATLAS and CMS measured cross sections) ⇒ interaction scale Lambda = 60 TeV and extra-dimensional Planck mass ~ 30 TeV.
  • Can extend minimal model to separate KK graviton interactions with fermions and gauge bosons; this would eliminate dilepton tension. Could also separate based on flavor for increased complexity. Coupling 1/Lambda_i varies linearly with wavefunction of KK graviton at higher-dimensional location of the field to which it couples.
  • Predictions for gauge universal spin 2: rates relative to gamma+gamma = 1 for ZZ, 2 for WW, 0 for Zgamma. Predict gg/gamma+gamma = 8, i.e. rather low predicted cross sections in these other modes; Zgamma = 0 is a hard prediction for minimal model, though may be circumvented in some non-minimal models. [these numbers for gg, i.e. not including quarks].
  • Direct determination of spin 2 vs 0: expect to be able to make from angular distributions, using 25-50/fb of data.

2:00 - 4:00
Diphoton bump spin-0 interpretation in extra dimensions [Gunion]
See paper by J. Gunion et al. __https://arxiv.org/abs/1512.05771__.
Consider Randall-Sundrum model, spin-0 radion assumed to have mass of 750 GeV, need to choose where to localize SM particles, all on TeV brane not viable due to FCNC and EWPT, so need SM gauge bosons in bulk and have Higgs, t and b on TeV brane to avoid those problems; radion related to volume fluctuations of extra dimensional space.
  • Crucial parameter of model is ratio of Higgs vev to radion vev = 246 GeV / Lambda.
  • Small mixing between Higgs boson and radion is critical to get right gamma+gamma rate.
  • Near conformal limit, all couplings vanish except coupling to gamma+gamma and gg, e.g. tt, bb and hh rate are approaching 0, WW and ZZ are also low but do not vanish since W and Z fields are in the bulk.
  • Lambda required to be in the range 1.5 to 2.5 TeV to fit gamma+gamma signal, rate very sensitive to parameter xi, highest rate near xi = ⅙, close to conformal limit, rate to non gamma+gamma and non gg becomes very small near xi = 1/6
  • Limit on KK gluon mass is > 3 TeV from LHC8, here expect KK gluon mass between 3 and 5 TeV.
  • Radion width is ~1 GeV in this model, predicted cross section consistent with ATLAS and CMS diphoton bump and dijet limits (expect 2 pb at 13 TeV), predict very small decay rate to Zgamma.

4:00 - 4:30
Status report [Curtin]
  • Roadmap to interpretations of 750 GeV bump:
    • Spin 2 (KK graviton): predict rate to Zgamma = 0, WW = 2 gamma+gamma, ZZ = gamma+gamma, gg = 8 gamma+gamma. Also, expect KK tower of resonances. Will require about 300 fb^-1 to observe ZZ and/or WW. Decay to dileptons already at the edge of viability, can be observed or ruled out with ~10 fb^-1.
    • Spin-0 elementary particle: production and decay via loop couplings. Approximate expected rates are gamma+gamma = WW = ZZ = Zgamma and gg = 100 x gamma+gamma. Latter implies that dijet resonance could be observed with 30-100 fb^-1.
    • Spin-0 onium bound state or quirk:
      • if binding via new interaction, expect dilepton resonance to be observed soon;
      • if binding via QCD, expect large charge resulting in smaller rate to dileptons and dijet but open production would produce multijet resonances.
    • Spin-0 composite particle:
      • if extra-dim radion provides expectations similar to elementary scalar but with addition of KK tower and absence of decay to Zgamma.
      • If composite (vectorlike case), expect light state similar to pion with other states considerably heavier; similar phenomenology as elementary scalar.
      • If glueball, difficult to observe, must look for other states.
  • Important signatures to disentangle different interpretations of the 750 GeV bump are: dilepton, dijet, Zgamma, also new heavy fermions. These become relevant with 10-30 fb^-1 of collisions at 13 TeV.
  • Discrimination between spin 0 and spin 2 possible with angular distributions. Need at least 30 fb-1.
  • ZZ and WW signatures also useful but require higher integrated luminosity like 300 fb^-1.
  • Search for resonance in dijet channel near 750 GeV is affected by trigger turn-on. Would be beneficial to lower thresholds. This requires increased bandwidth or additional requirements on dijet system like delta eta cut to select more central dijets in this mass region. Comparison between potential signals in 2- and 3-jet events may also help study impact of trigger turn-on.
  • Possibility to lower backgrounds in dijet search with quark-gluon discrimination.


=== 11 May 2016 ===

11:00 - 12:00
ATLAS same-sign dilepton+trilepton with 2 or 3 b-tags excess from Run 1 [Willocq]
Analysis documented in __http://arxiv.org/abs/1504.04605__.
  • Excess of data in events with H_T > 700 GeV and ETmiss > 100 GeV. Significance of 2.5 sigmas considering all four-top signal regions (five such regions).
  • Most of the events are ee or emu, lack of like-sign mumu events in signal regions SR6 and SR7 seems anomalous, perhaps indicates underestimated electron fake/non-prompt or charge misID.
  • Similar analysis by CMS (__http://arxiv.org/abs/1311.6736__) does not see excess in SR28 with similar kinematic requirements, observe 2 events in good agreement with background ⇒ supports background fluctuation interpretation of ATLAS excess.

1:30 - 2:30
BaBar/Belle/LHCb measurements of B → D(*) tau nu / B → D(*) l nu [Renner]
  • Measured values of ratio for D and D* is larger values than SM by about 30%, significance of combined deviation from SM greater than 3 sigmas; BaBar paper at __https://arxiv.org/abs/1205.5442__ which is 3.4 sigmas away from SM.
  • Interpretation in __http://arxiv.org/abs/1507.07567__, requires 2HDM Type III with large tan beta (~ 50) and m(H+) = 200 GeV or 3rd generation leptoquarks with mass near 1 TeV.
  • Consequence for LHC: look for t —> h c and look for t —> H c with heavy Higgs H —> tau tau or tau mu; other interesting channel to look at Z—> tau mu at the LHC.
  • Large tan beta implications for Higgs physics at LHC: continue program to search for heavy scalars that are favored by large tan beta, especially H+ search in tau nu or W H final states, which is challenging in mass region near top quark mass; major production mode is gg —> b H+ —> H+ t (forward b) which is difficult; check this by performing Wt with W->tau nu and then search for H+ t.

2:30 - 3:30
ttH multilepton CMS 8 TeV excess [Chou]
  • CMS combination paper at__https://arxiv.org/abs/1408.1682__, data excess primarily is same-sign dimuon: SS ll + >= 2 b, >= 4 jets, pT(l1) > 20 GeV, pT(l2,l3) > 10 GeV, deviation at 3.4 sigma level.
  • Interpretation of this excess in connection with ATLAS same-sign dilepton excess requires pair production of top partners or stop with charm-jet mistagged as b.
  • Open question: Could this excess be due to new particle with displaced vertex? Need to know efficiency for a long-lived particle to pass b-tagging requirement to explain large number of b-tags in these multilepton searches.




Bloc 2: Dark matter signatures and models


Mono-Z’ signature

Mono-displaced signature

Mono-lepton-jet signature

Black Hole + MET signature
  • motivation
    • Dark Matter may communicate to the SM entirely through gravity, which may itself become strong at the TeV scale, leading to black hole production.
  • possible existing papers

Mono-tau signature


Mono-X+Y signature
Mono-X + soft leptons (for example, mono-jet + soft leptons). Also, Mono-X+soft lepton + soft photon (all prompt)
  • motivation
    • Happens in classical wimp models where dark matter carries electroweak charges. Soft photon can come from radiative decay of heavy neutralinos to light neutralinos
  • possible existing papers
    • Bino/Higgsino mixed dark matter: a lot of discussions; for instance, __https://arxiv.org/pdf/1404.0682.pdf__ (Section 5);
    • see also arXiv:1312.7350, 1307.5952, 1401.1235, 1408.6530, 1409.4533, 1412.4789

Mono-jet + soft displaced lepton or photon signature
  • motivation
  • For a slightly more complicated dark sector with multi-states, additional displaced soft lepton or photon could be emitted from dark transitions. Hadronic decays have larger branching fractions and can be considered if backgrounds can be kept small.
  • possible existing papers

Multi-X signature
Dijet resonance + X

Multi-W (Z) + MET signature
  • motivation
    • Diboson final state recoiling against MET. Kinematic region different from those in SUSY chargino searches? No concrete models yet


T-channel


A t-channel interpretation of mono-jet is important for several reasons:
  • s-channel models are not the whole story (not even the main part of the story until recently): t-channel provides breadth to the picture of LHC contributions to our understanding of dark matter.
  • dijet resonance constraints do not apply (no q-q-mediator vertex)
  • Monojet vs Dijet: gQ^4 scaling vs gDM gQ scaling
  • the jet pT and MET distributions are qualitatively different than MSSM or s-channel DM simplified models: for couplings between MSSM-like (<~0.3) and the non-particle limit (>>1), direct production of single squarks can produce Jacobian peaks

Literature:

Comments:
  • One should not ignore the squark pair production diagram in the analysis. This can still be large, leading to multijets+MET constraints.

Electroweak Backgrounds to Mono-X Analyses

Mono-jet-like searches are limited by their ability to precisely predict the background from Z+jets processes at Z pTs of ~200 GeV–1 TeV. This is the dominant uncertainty in the mono-jet search. Signals involving light mediators (invisible Higgs, for example) predict MET distributions that are very similar to the Z background which already require very small background uncertainties to observe. At the higher end of the Z pT range, the background uncertainty will set the ultimate sensitivity of the analysis for Run 2 and beyond.

Present approach

Both ATLAS and CMS estimate the Z->vv backgrounds with MC that has been reweighted to reproduce the jet and MET distributions in control samples of Z->ll, W->lv, or photon+jet events. For a given Z pT, Z->ll control regions always provide ~3x fewer statistics than the corresponding signal region, leading to the use of the W or photon samples. Extrapolation of the shapes observed in the W and photon samples to Z samples uses MC predictions of the ratio of W/Z or photon/Z.

ATLAS uses parton-matched Sherpa MC (__http://arxiv.org/abs/1604.07773__ 13 TeV, __http://arxiv.org/abs/1502.01518__ 8 TeV) with electroweak corrections, and data in W and Z control regions. Systematic uncertainties on the ratio of W/Z are derived with commonly used procedures (fixed choice of scales varied by factors of ½ or 2, etc.) ATLAS and CMS quote a 2-3% total uncertainty on the Z->vv prediction at 250 GeV in Z pT, increasing to order 20% at higher MET.

Are these uncertainties realistic?
  • Electroweak corrections are O(20%) getting larger at higher boson pT, but these are Sudakov logs and well-understood, so their uncertainty is not large. However, the electroweak Sudakov factors for W and Z are different and may not cancel entirely.
  • The scale uncertainties are likely to be much larger than ATLAS or CMS currently quote. These ad hoc variations (½ or 2) are not like PDF uncertainties. They are attempting to capture higher order corrections, but
    • It is not obvious that the variations should be correlated in the numerator and the denominator of the V/Z ratio---the higher order corrections may have different signs. A conservative approach would be to take the maximal variation of numerator and denominator as an estimate. It is expected that the resulting uncertainty would be on the order of 7–15%.
  • The scale choice itself (the central value) may not be appropriate for the analysis phase space. This requires more careful discussion with QCD theorists on a case by case basis. In addition, one should take care that a dynamic scale choice is not inadvertently converted to a fixed scale choice by the generation procedure.
  • Given the ad hoc nature of the scale uncertainties, the current situation where they dominate the experimental sensitivity is undesirable...

How can the estimate be improved?
  • Better calculations are available for small numbers of additional partons (<=3, which fits well with the usual veto after 2-3 additional jets). A ratio calculation for the mono-jet signal cuts was discussed.
  • The next generation of precision is at least several years away. Sub 1% precision seems very difficult at present.
  • One should further explore the use of photon+jets. While this provides higher statistics, it is not obvious whether additional uncertainties (such as on photon fragmentation) negate this gain. The present Z/photon predictions do not describe the data as well as hoped, only to 15–20%.
  • One could also explore absolute predictions of Z+jets, normalizing the prediction in control regions (for example, away from the Jacobian peak of t-channel signals)
  • Once a better picture of the true uncertainties is available, one could also explore changing the analysis selection to reduce the sensitivity to them. For example, “Z->vv with pT=1 TeV plus no more than a few jets” may sample an unusual corner of phase space.

Improving the central uncertainty on many DM searches seems a very fruitful area for collaboration between DM searches and precision QCD theorists like those present for the neighboring workshop. Discussion in the LHC DM WG could be one way to advance this.

List of ideas that were discussed but not yet pursued
  • Are the assumptions we are applying in reinterpretation (e.g. relic density) useful and applicable?
  • Bottom-up approach to Dark Matter: look for all objects in addition to MET (in the direction of general searches). Thinking about a SUSY simplified model approach: are there models that are matched to a wide range of topologies?
  • What measurements of SM can be sensitive to DM?
  • What about replacing MET with soft objects? Do we need a search program for that, adjacent to a compressed scenario?
  • Other kinds of DM (WIMPonium, self-interacting, asymmetric, dynamical DM)
  • Interplay between DM and Higgs models
  • Systematic review of more out-of-the-box DM: displaced, hidden valley, multiple soft things...


Bloc 3: Trigger opportunities

Preliminary information:
Trigger menus:

Plenary topics:

- Motivation for looking for dijet resonances at low masses

Topics for this part:
• DM very simplified models pointing the way
• TLA done in ATLAS and CMS: numbers and performance

Points for discussion
• what can we do that dileptons can't
  • are there sensible models without leptons?
  • FCC question: when are dileptons more expensive than dijets? (muon resolution)
• are you convinced by the quality of a TLA?
• how low to go in couplings: in principle, leave no stone unturned. in practice, effort vs outcome: need more compelling cases
  • light RPV stops (link to SUSY session) for substructure triggers, can we calculate granularity needed from first principles given this model?
• what can you do with FTK and TLA? track information at L1. Lepton/photon separation?

- Motivation for looking for other scoutable things

Points for discussion:
• Necessary to find further physics cases for this
• Not much MET, jets are not hard enough, lepton is soft
• what other corners can we experimentally cut?
  • leptons - isolation, 5 GeV
▪ what are the motivations to do this? multilepton?
• what can we do with muons?
  • L1_MU15/20, impose FTK, can probably go down to L1_Mu6? (careful about semileptonic B decays), record jets in the lower and have events coming out at L1.5
• photons
  • what are the fake rates for photons when lower threshold? can't beat pi0 fakes with this
  • can you push down the single photon trigger?
• electrons

- Differently timed LHC analyses

Topics for this part:
• real-time analysis (mostly detector calibration)
• unblinded monitoring analysis
• delayed stream/data parking
• accessory collision experiments (ATLAS/LHCf example)

Points for discussion:
• what do we want to park and why?
• blind analyses vs automatic analyses
• LHCb turbo stream and buffering: why software (100% software) vs hardware triggers (FTK-like)
• what other things can we do with auxiliary detectors using ATLAS as trigger

- Triggering outside the box

Topics for this part:
• Exotic Higgs as a use case
• What is done in ATLAS and CMS in terms of unusual triggers for LLPs

Points for discussion:
• How to motivate these searches, links between DM and composite Higgs and LLPs
• Are anti-triggers useful (e.g. track or pixel veto for FTK, trackless jets)?
• Are jet+X with X soft useful?
• Are we missing something? review ATLAS/CMS trigger menu (public info) and think of signatures that we are neglecting, make a list and put on twiki
• Where in the repertoire of standard searches are we losing ground due to rising thresholds?
• What could be newly feasible with the improvements at levels 1 and 1.5?
• Higgs ExoDK, DM, Compressed Spectra good benchmarks for low-HT triggers. Are there others?
• Triggers for boosted objects; Need? Status? Improvements?
• Triggers for long-lived particles; Optimized? Worth additional effort right now?
• Any places where moderate improvements in processing power could make a big difference?

First trigger session: what are we scouting / what else can we scout?


- What is being scouted:

Ideal use case: particle without known mass, produced in pairs, hidden in a large background. Can use sidebands/fit to find it.

Sociological/difficulty issue: these searches have to be very well motivated, as those require dedicated experimental studies.

Jets: use cases and models
  • dijets
  • plain: done at CMS/ATLAS
  • + ISR
    • may need substructure since ISR boosts fat jet
    • models:
      • vector/axial Z', but lower masses needs to have lepton signatures for theoretical reasons, so can use dilepton signatures
      • pseudoscalar particle, but would need to have contrived loops on vertex
      • single production of RPV squark, ~100 GeV
  • paired dijets/4 jets
    • light RPV stops, pair produced
    • may need substructure -> see third session
    • double parton scattering: qq->jj x2 within the same pp collision
  • triplets
    • gluino coverage gap of 40 to 100 GeV (LEP/CDF)

CMS: Use of parked data in case there is something. The rate limitation of 1 kHz still applies, but only 300 Hz is processed right away.
ATLAS: Record 1kHz, no parked data so far in 2015.

Muons: use cases and models
• single muons (dimuons)
• CMS: invariant mass of 10 GeV and above, HLT muon threshold: pT>3 GeV
• low dimuon resonances, dark photons

- What else to scout?

Photons
• standard photons: could calculate photon fake rate in sidebands, parameterise for trigger jets even though tracking isn't there
• photon-jets, pure calorimetric objects, could be interesting if we manage to remove noisy background. A CalRatio trigger like the ATLAS one may work here.
• no compelling physics cases yet

[Here the session was interrupted by a whale sighting https://goo.gl/photos/DEnSe6yc4S1Wao649]

Taus:
• see problems with photons

Trigger thresholds are high for single photons and taus, but no compelling physics cases so far.
Additional complication: need photon and tau identification with tracking -> see second session

MET:
• would be interesting for VBF signatures and soft objects + lower MET
• Example: "exploding Higgs": Higgs decaying into many soft particles
• however, need to do all objects well online in order to avoid saving them (small data format is the point of data scouting)
• this may be a problem in the forward region
• CMS has a chance with PFlow jets (HT>450 GeV, also parked)

Displaced triggers:
• CMS has VBF+displaced jet triggers
• ATLAS has displaced triggers
• Can we do trackless jets in scouting and outside scouting?

Second trigger session: what can we do with FTK?


Added value of a track trigger:
• for scouting: have more track info calculated already - can scout low pT b-jets and taus, have MET
• for offline: decrease thresholds because background is removed thanks to extra track info

Object improvements:
• online MET
• online pileup rejection
• better online jet resolution
• isolated tracks: are there one, zero, or many non isolated tracks?
• could improve lepton/photon identification
• displacement information

Possible use cases for FTK:
• Z+gamma, through a scalar

• H-tautau, H-bb (VBF)
• H -> hh -> 4b, but low momentum S/B is prohibitive
• h/H->AA (< 50 GeV)->4tau, not decaying to muons
• VBF + trackless jets
• jet + MET + resonant X
• jet + MET + dE/dx: with FTK, choose events with a high pT track
• VBF + MET + pair of leptons or photons, with kinematic structure, but too soft to trigger on their own - could be a signature of H->DM DM -> pairs of dark/non dark photons

but also cascade of chi2 chi1 radiating photons (edge)

Third trigger session: triggering on clutter

Theoretical motivation: see N. Craig's talk

What is clutter? Hard process (annihilation) + soft process (spindown).
• Expecting a hard resonance in di-things + soft objects.
• How does this fragment? If many radiative objects, see “signals of new physics in underlying event”: http://arxiv.org/abs/0810.3948
• Natural place to start: add a cut to the dijet/dilepton/diphoton search (e.g. large track multiplicity or lack of tracks) to reduce the background, then bumphunt again
• Equivalent to: look under the lamppost, then move the lamppost a little
• Crucial to make sure that cut does not shape the bumphunted distribution

Theory question: do we have models for this kind of searches? Once we have a model we start benchmarking searches and seeing if current triggers are sufficient (depends on how soft things are, it would be good to find a limiting case).

Fourth trigger session: triggering on substructure

State of the art: CMS can already recluster (=rerun jet finder) for pflow jets at the trigger level as all inputs are there, ATLAS would need more information.

CMS uses trimmed mass in trigger, brings the rates of the HT down slightly. After a certain point in mass (roughly 30 GeV), the trimmed mass is not usable anymore. Other substructure quantities are fairly correlated with mass, so not used directly at the trigger level.

ATLAS has plans to add components that allow reading entire calorimeter at L1, implementing some ‘subjettiness’ variables (see upgrade TDR: https://cds.cern.ch/record/1602235). See also https://twiki.cern.ch/twiki/bin/view/AtlasPublic/JetTriggerPublicResults#Phase_I_Upgrade_Performance_Plot

Idea: trackless jets with substructure. Could potentially catch topologies like the one below:
TracklessJetsWithSubstructure.png
Experimentalists would like to have a benchmark for this.


Bloc 5: Long-lived Particles



Bloc 6: Collimated Objects / Hadronic Resonances

Goals:

1) Maximize theory input to experiment
2) SM Opportunities
3) New Ideas for BSM searches

1) Maximize theory input to experiment

A lot has been done but we are missing connections - should be improved.
a) Improved theory input can improve uncertainties:
Example: decorrelation for N_subjettiness variables (tau_DDT),
https://arxiv.org/abs/1603.00027
b) calculational tools can be used to constrain experimental uncertainties
Example:
Machine learning vs. groomers - soft-residual function is not currently exploited in tagging algorithms. This is information that helps machine learning algorithms performing a bit better. Should be explored.

2) SM Opportunities

Discussed 3 main areas where we can improve by utilizing substructure for SM measurements:
a) precision measurements
b) constraining systematics and improving simulations (eg parton showers, pdf's etc)
c) extrapolating SM measurements to higher pt and applications to searches
low-lying fruit:
- boosted SM dibosons, ttbar
- can gain knowledge by considering semileptonic channels
- study the internal structure of gluon jets more: track multiplicity is used a lot but is infrared unsafe;
number of subjets is better to look at infrared safe vars
- q/g discrimination currently in context of searches but could have some made headway for SM information - and can be applicable to searches --> the better modeling of this in a data-driven way will improve our understanding and then fold into all sorts of measurements (and searches)in the long run
- HF content (g-> bbar, ratios of HF/LF)
harder-to-reach fruit:
- interplay dsigma/dX (groomed) / dsigma/dX (ungroomed) --> access to the soft /hard separately
- photon fragmentation (hard quark -> photon + soft quark) - a collinear effect

3) New Ideas for BSM searches

I) Experimental:
- explore fancier triggers
II) Non-SM like topologies
e.g. Hidden Valley/Dark sector:
a) boosted + diffuse
b) boosted + boosted
Some brainstorming ideas:
- look for pair produced new particles decaying hadronically, can search for two highest-pT subjets in two different jets or subjets that have the same mass in two sides of the event
- look for two highest pT jets in event that have some substructure and plot jet mass in a 2-D plot to see if there is evidence for two highly boosted particles with same mass (could be Ws or some other new particle); could be used in complicated events with additional leptons, ETMiss, b-tags, etc.; could look at jet mass or subjet mass
- this sort of approach already used in CMS for VLQ top pair already -- i.e. extension of current techniques: RPV susy, Z'->Tt or ->TT (VLQ) etc
==> Led to idea of adopting a general application of a "catch-all approach" for massive subjets:
- pair of large diffuse jets containing a single hard subjet each with the same mass
- paired dijets with very high boost, see example by ATLAS hadronic search for RPV stops: http://arxiv.org/abs/1601.07453
- also could apply to more general situation to look for different signal regions depending on number of leptons, b-tags, ETmiss, etc.
III) Applications of Boosted leptons/taus and photons
For lepton isolation, 3 approaches exist:
a) "cone": transverse momentum sum relative to lepton in a fixed or shrinking jet cone
b) "lepton subjet fraction" (goes along the hidden valley applications): lepton vs. subjet inside large-R jet, could be used in displaced vertex searches in conjunction with use of pronginess ; could be useful in searches for dilepton resonances at mass < 100 GeV in events that are busy, could not afford standard isolation in such case; here have access to clustering sequence information to find out whether lepton is associated with a subjet; this method is good for any non-jet particle (lepton or photon) isolation
c) "specialized boosted" isolation: select in deltaR vs pT_rel(lep, jet) space