Run II searches for dark matter at the LHC with the CMS experiment at - - PowerPoint PPT Presentation
Run II searches for dark matter at the LHC with the CMS experiment at s = 13 TeV C edric Prieels Instituto de F sica de Cantabria on behalf of the CMS Collaboration International Center for Theoretical Physics (Trieste) Interpreting
1 / 39
◮ Introduction ◮ Dark matter in particle accelerators ◮ Hunt strategies ◮ Mono-X/pmiss
T
+X searches
◮ Mono-jet/Mono-V ◮ Mono-γ ◮ Mono-Z ◮ t/¯ t + DM ◮ Mono-top
◮ Mediator searches
◮ Dijet bump hunting ◮ Dijet light resonances ◮ Dijet angular searches
◮ Higgs portal
◮ Higgs to invisible
◮ COmparation of the previous results ◮ Conclusions
2 / 39
Different astronomical observations lead to the birth of the dark matter hypothesis, such as: ◮ Apparent gravitational anomalies and difference between dynamic and luminous mass of the galaxies ◮ Anisotropies of the CMB (DM contributes to the gravitational collapse of matter, but is unaffected by the pressure from photons) ◮ Large scale structures of the Universe We now know/assume that: ◮ It accounts for ∼25% of the total content of the Universe ◮ Its nature cannot be explained by the Standard Model, extensions are needed ◮ Dark matter candidates are usually cold and only interact weakly and gravitationally ◮ The WIMPS (Weakly Interacting Massive Particles) are considered the best dark matter candidate in this talk
3 / 39
3 / 39
Dark matter can be produced within particle accelerators if: ◮ The dark matter mass is low enough ◮ Its production cross section is large enough ◮ Dark and ordinary matter interact at least weakly with each other The LHC is able to probe energies higher than ever with huge luminosities: ◮ Largest dataset to date to analyze at 13 TeV ◮ Perfect tool to try and detect DM particles ◮ Able to study a large range of particle masses and cross-sections ◮ The two multipurpose detectors (CMS, ATLAS) are mostly able to search for DM particles → However, if producing dark matter particles is theoretically possible, detecting them directly is impossible, as they are not expected to interact with our detector.
4 / 39
The Compact Muon Solenoid is one
LHC, designed to: ◮ Make precision measurements ◮ Search for the Higgs boson ◮ Search for new exotic processes
◮ Powerful tracker and muon detection system to measure the properties of the leptons in a large range of energies. ◮ Huge solenoid as central piece able to produce a 3.8T magnetic field, to curve the charged particles and study their properties. ◮ Made of different layers (such as the tracker, the calorimeters and the muon chambers), each having its own purpose, resulting in a great particle identification and precise momentum determination. ◮ A trigger system is used to select only interesting events out of the 600 million collisions per second produced (current bandwidth ∼ 1kHz).
5 / 39
6 / 39
The CMS detector is not able to directly detect eventual dark matter particles. How- ever, we can rely on visible particles to detect eventual invisible particles. The key variable to detect DM is the missing transverse energy (MET): ◮ Defined as the imbalance in transverse momentum in the plane perpendicular to the beam direction pmiss
T
= −| − → pT | = 0 ◮ This quantity is = 0 if something escapes the detector undetected (eg: neutrinos, DM) Most of the DM searches are therefore dependant on this variable, as they rely on high pmiss
T
values recoiling against visible objects (such as jets, leptons, photons,...). However, a pmiss
T
= 0 does not mean that we discovered new physics, as common processes can have the same effect: ◮ Neutrino production ◮ Limited detector resolution → A good understanding of the detector is therefore crucial to make a distinction, especially at high pmiss
T
values!
7 / 39
Different strategies are usually used to search for DM in the LHC: ◮ Mono-X/pmiss
T
+X searches: Search for DM in association with a SM particle, used to trigger the event (jet, lepton, photon) and recoiling against the invisible DM system
◮ pmiss
T
+X in the final state ◮ ∆φ(DM, X) ≃ π ◮ Mono-jet, mono-γ, mono-Z,... analyses ◮ Searching for global excesses in the MET spectrum
◮ Multi-object/Mediator searches: Initial and final state made out of SM particles, but a DM mediator appeared in the way
◮ Can probe the dark interaction even if DM is inaccessible ◮ Can look for both invisible and visible decays of the mediator ◮ Search for resonances and bumps in known spectrum, such as the dijet invariant mass
8 / 39
◮ Higgs portal: In this case the DM is produced as a result of the decay of a Higgs boson
◮ The SM decay of the Higgs to invisible (4ν) is possible but unlikely (BR ∼ 0.1%) ◮ Several Higgs production modes can be studied
◮ SUSY-like searches: These searches focus mostly on models in which the DM decays to SM particles, but SUSY also provides a DM candidate (such as the lightest supersymmetric particle)
◮ This subject will not be covered in this talk either
◮ Long-lived searches: Relatelively new searches when one of the particles is able to travel for a short distance before decaying
◮ Only makes sense if a SM particle that can be detected is produced in the decay as well ◮ The DM can either decay rapidly to SM particles after traveling for some distance, or a long lived partner can be produced with DM ◮ This subject will not be covered in this talk either
9 / 39
The ATLAS-CMS Dark Matter Forum for the Run2 searches was held in 2016 and resulted in the publication of a
paper in arXiv , defining the signal models that should
be studied by the analyzers. Main objectives of this report: ◮ Channel the efforts of the CMS and ATLAS collaborations ◮ Define the highest priority analyses that should be conducted ◮ Define the simplified models and EFTs to be used for the Run2 searches
◮ DM is supposed to be a Dirac fermion (choice Dirac/Majorana is only expected to produce minor changes in the kinematics) ◮ Simulate a set of prioritized set of operators and parameters with distinct kinematics for the interpretation of the results, based on the Run1 results
As a result, the mass of the mediator and DM particles along with their spins and couplings gq and gχ are defined and usually considered as the free parameters of all the models considered, and common for both ATLAS and CMS.
10 / 39
10 / 39
Results published in 2018 with 35.9 fb−1 of data in
tions considered (simplified DM, fermion portal, non-thermal dark matter models,...) Simple signature: at least one energetic jet (ISR - monojet - or from a W/Z boson decay -monoV-) recoiling against an invisible high pmiss
T
system: ◮ Main backgrounds:
◮ Z(νν)+jets (60%, irreducible) ◮ W(lν)+jets (30%) ◮ QCD multijet background with mismeasurements of the jet momenta
◮ Signal extraction performed using the distribution of the pT imbalance in each event category defined Main improvement since the previous publications: larger dataset, revised theoretical predictions and uncertainties for some processes (γ+jets, Z+jets, W+jets). processes
11 / 39
A binned likelihood fit to the data is performed on the pmiss
T
spectrum in 5 mutually exclusive control regions (dimuon, dielectron, single muon, single electron, γ + jets, and on the signal region. Transfer factors then link the yields from the CR to the SR
◮ Main selection applied depends on the model
◮ Mono-jet: jet pT > 100 GeV ◮ Mono-V: jet (pT > 250 GeV) from hadronic decays of Lorentz-boosted W or Z boson ◮ But both look for large pmiss
T
/jet separation and at high pmiss
T
values (> 250 GeV) ◮ A lepton/b-tag veto is applied to reduce the backgrounds
Mono-jet
Events / GeV
2 −
10
1 −
10 1 10
2
10
3
10
4
10
5
10
6
10 (13 TeV)
35.9 fb CMS monojet
Data inv. → H(125) = 2.0 TeV
medAxial-vector, m )+jets ν ν Z( )+jets ν W(l WW/WZ/ZZ Top quark +jets γ (ll), γ Z/ QCD Data / Pred. 0.8 0.9 1 1.1 1.2
[GeV]
miss T
p
400 600 800 1000 1200 1400
Unc. (Data-Pred.) 2 − 2
Mono-V
Events / GeV
2 −
10
1 −
10 1 10
2
10
3
10
4
10 (13 TeV)
35.9 fb CMS mono-V
Data inv. → H(125) = 2.0 TeV
medAxial-vector, m )+jets ν ν Z( )+jets ν W(l WW/WZ/ZZ Top quark +jets γ (ll), γ Z/ QCD Data / Pred. 0.8 1 1.2
[GeV]
miss T
p
300 400 500 600 700 800 900 1000
Unc. (Data-Pred.) 2 − 2
12 / 39
Vector mediator Axial-vector mediator ◮ Upper limits are computed at 95% CL
measured signal cross section to the predicted one ◮ Vector (axial-vector) mediators excluded up to ∼ 1.9 (1.7) TeV ◮ DM excluded up to ∼ 750 (500) GeV Scalar mediator
[GeV]
med
m
100 200 300 400 500 600 theory
σ / σ 95% CL upper limit on
1 2 3 4 5 6 (13 TeV)
35.9 fb CMS Preliminary
Observed 95% CL Median expected 95% CL 68% expected 95% expected = 1 µ = 1
DM
= 1, g
q
= 1 GeV g
DM
Scalar med, Dirac DM, m
Pseudoscalar mediator ◮ No exclusion obtained for the scalar case ◮ Pseudoscalar mediators excluded up to ∼ 400 GeV, DM up to ∼ 150 GeV This search is currently one of the most sensitive DM search of CMS, mainly because
13 / 39
Results published in 2018 with 35.9 fb−1 of data in
Journal of High Energy Physics
q ¯ q γ χ ¯ χ Analysis similar to the mono-jet analysis, with a photon instead of a jet emitted against the DM invisible system: ◮ Emission less probable, makes the analysis a bit less sensitive ◮ However, high efficiency and purity can be
◮ Simultaneous fit on the E γ
T variable on the
signal and control regions (instead of a ”simple” cut and count) ◮ Typical selection:
◮ One photon (pT > 165 GeV) ◮ Two DM particles (pmiss
T
> 170 GeV) ◮ Charged leptons are vetoed
◮ Main backgrounds: Z(νν)+γ (50%), W(lν)(+γ) (30%) ◮ Two signal regions to constrain the beam halo normalization
Data Pre-fit Background-only fit γ + ν ν → Z γ + ν l → W Electron fakes Hadron fakes Other SM Non-collision Signal
Preliminary CMS (13 TeV)
35.9 fb
(GeV)
γ T
E 200 400 600 800 1000 Events / GeV
3 −
10
2 −
10
1 −
10 1 10
2
10
Data / Pred. 0.5 1 1.5 2 2.5 Preliminary CMS (13 TeV)
35.9 fb
Events / GeV
3 −
10
2 −
10
1 −
10 1 10
2
10
Data Pre-fit Background-only fit γ + ν ν → Z γ + ν l → W Electron fakes Hadron fakes Other SM Non-collision Signal
Preliminary CMS (13 TeV)
35.9 fb
(GeV)
γ T
E 200 400 600 800 1000 Events / GeV
3 −
10
2 −
10
1 −
10 1 10
2
10
Data / Pred. 0.5 1 1.5 2 2.5 Preliminary CMS (13 TeV)
35.9 fb
Events / GeV
3 −
10
2 −
10
1 −
10 1 10
2
10
14 / 39
Transfer factors between the four control and signal region are also used to reduce the systematics uncertainty and take advantage of the higher statistics in the single lepton control samples. The upper limits on the signal production at 95% confidence level have been obtained considering vector and axial-vector mediators, in the 2D phase space (Mmed − mDM). Vector mediator
95
µ Observed
2 −
10
1 −
10 1 10
2
10
95
µ Observed
2 −
10
1 −
10 1 10
2
10 (13 TeV)
35.9 fb
Preliminary CMS
= 1
DM
g = 0.25,
q
g Vector, Dirac, = 1
95
µ Observed Theoretical uncertainty = 1
95
µ Median expected 68% expected Relic
[GeV]
med
M
500 1000
[GeV]
DM
m
100 200 300 400 500 600 700 800
Axial-vector mediator
95
µ Observed
2 −
10
1 −
10 1 10
2
10
95
µ Observed
2 −
10
1 −
10 1 10
2
10 (13 TeV)
35.9 fb
Preliminary CMS
= 1
DM
g = 0.25,
q
g Axial Vector, Dirac, = 1
95
µ Observed Theoretical uncertainty = 1
95
µ Median expected 68% expected Relic
[GeV]
med
M
500 1000
[GeV]
DM
m
100 200 300 400 500 600 700 800
In both cases, mediator masses up to 950 GeV are excluded for mχ < 1 GeV.
15 / 39
Results published in 2018 with 35.9 fb−1 of data in
Journal of High Energy Physics
The results obtained are interpreted in terms of simplified models of DM production with spin 0 or 1 scalar, pseudoscalar, vector or axial-vector mediators: Spin 0 mediator φ Spin 1 mediator Z’ Global selection applied: ◮ Exactly two isolated leptons (pT > 25, 20 GeV) ◮ Third lepton veto ◮ Mass close to the Z (15 GeV window) ◮ large pmiss
T
(> 100 GeV) ◮ Since little hadrnoic activity is expected, no jet or bjet are expected
16 / 39
This channel is competitive thanks to its small and well-known backgrounds: ◮ ZZ → 2l + 2ν (same final state, 60%) ◮ WZ → 3l + ν (25%) ◮ WW → 2l + 2ν (5%) Three variables are used for discrimination and to reduce DY/top backgrounds: ◮ The missing transverse momentum pmiss
T
◮ Azimuthal angle formed between the dilepton system and the pmiss
T
◮ The pmiss
T
− pll
T balance ratio
Several improvements since the 2015 results: larger dataset, new techniques to es- timate the backgrounds, improvements in the event selection and new BSM models probed.
17 / 39
Vector mediator Axial-vector mediator ◮ Vector (axial-vector) mediators excluded up to ∼ 680 (700) GeV ◮ DM excluded up to ∼ 240 (160) GeV Scalar mediator Pseudoscalar mediator ◮ No exclusion obtained for the scalar or pseudoscalar mediators with this analysis
18 / 39
Results currently targeting publication with 35.9 fb−1 of data in the Journal of High Energy Physics. A search for DM with a Higgs boson and high pmiss
T
is performed: ◮ Five orthogonal decay channels considered: b¯ b, ττ, γγ, ZZ and, for the first time WW → a combination is then performed to maximize the sensitivity of the search (first combination based on 5 Higgs decay channels!) ◮ The ISR of a Higgs boson is strongly suppressed → possible to directly inspect the interaction between DM mediator and Higgs boson ◮ Two simplified benchmarks models considered: the decay of a Z’ to a pseudoscalar A (Z’-2HDM) and the radiation of a h when the Z’ is the mediator
Z’-2HDM Baryonic Z’
19 / 39
Z’-2HDM
500 600 1000 2000 3000 4000 [GeV]
Z'
m
2 −
10
1 −
10 1 10
2
10
3
10
4
10 th
σ / σ
combination ) b h(b σ 1 ± combination ) τ τ h( σ 2 ± combination ) γ γ h( h(ZZ) h(WW)
CMS Preliminary
Z'-2HDM, Dirac DM =100 GeV
χ
=300 GeV, m
A
m =1 β =1, tan
χ
=0.8, g
Z'
g
H
=m
± H=m
A
m (13 TeV)
35.9 fb Solid (dashed) lines:
Baryonic Z’
500 1000 1500 2000 [GeV]
Z'
m
2 −
10
1 −
10 1 10
2
10
3
10 th
σ / σ
combination ) b h(b σ 1 ± combination ) τ τ h( σ 2 ± combination ) γ γ h( h(ZZ) h(WW)
CMS Preliminary
Baryonic Z', Dirac DM = 1 GeV
χ
= 1, m
χ
= 0.25, g
q
g (13 TeV)
35.9 fb Solid (dashed) lines: observed (expected) 95% CL limits
◮ Limits obtained for the 5 individual decay channels ◮ The b¯ b decay channel gives the best limits because of its high BR Z’-2HDM
500 1000 1500 2000 2500 3000 3500 4000 [GeV]
Z'
m 400 500 600 700 800 900 1000 [GeV]
A
m
2 −
10
1 −
10 1 10
2
10
+ WW +ZZ) τ τ + γ γ + b DM + h(b Observed 95% CL Expected 95% CL
theory
σ 1 ± Obs.
experiment
σ 1 ± Exp.
experiment
σ 2 ± Exp. (13 TeV)
35.9 fb Z'-2HDM, Dirac DM = 1
χ
= 0.8, g
Z'
g =100 GeV
χ
m =1 β tan
H
=m
±H
=m
A
m theory
σ / σ
CMS Preliminary
Baryonic Z’
200 400 600 800 1000 1200 1400 1600 1800 2000 [GeV]
Z'
m 100 200 300 400 500 600 700 [GeV]
χ
m
2 −
10
1 −
10 1 10
2
10
+ WW +ZZ) τ τ + γ γ + b DM + h(b Observed 95% CL Expected 95% CL
theory
σ 1 ± Obs.
experiment
σ 1 ± Exp.
experiment
σ 2 ± Exp. (13 TeV)
35.9 fb Baryonic Z' = 1
χ
= 0.25, g
q
Dirac DM, g theory
σ / σ
CMS Preliminary
→ Most stringent limits on the parameters of these two models to date.
20 / 39
The following model is being considered: ◮ Spin 1/2 DM χ (1 GeV, Dirac fermion) ◮ Spin 0 scalar (S)/pseudoscalar (PS) mediator φ ◮ Coupling gχ between the mediator and the DM set to 1 (same for gq coupling for all the quarks) ◮ A signature with top quarks takes advantage from the large Yukawa coupling of the mediator Typical signature includes pmiss
T
, 2 b-jets from the top decays and a different number
Hadronic channel
200 300 400 Events / bin 10
2
10
3
10
4
10
5
10
Data t t +V t t )+jets ν W(l *+jets γ Z/ Single t Diboson Multijets
Prefit =1 GeV
χ=100 GeV, m
am
(13 TeV)
35.9 fb CMS
Preliminary
miss T+p t 2RTT all-hadronic t
[GeV]
miss T
p 200 250 300 350 400 450 Obs / Fitted
0.5 1.0 1.5
Semi-leptonic channel
200 300 400 Events / bin 10
2
10
3
10
4
10
Data t t +V t t )+jets ν W(l *+jets γ Z/ Single t Diboson
Prefit =1 GeV
χ=100 GeV, m
am
miss T
+p t l+jets t
(13 TeV)
35.9 fb CMS
Preliminary
[GeV]
miss T
p 200 250 300 350 400 450 Obs / Fitted
0.5 1.0 1.5
Dileptonic channel
100 200 300 400 500 Events / bin 1 10
2
10
3
10
4
10 Data (2l) t t +V t t *+jets γ Z/ Single t (2l) Diboson (1l) / tW(1l) t )+jets / t ν W(l
Prefit =1 GeV
χ
=100 GeV, m
a
m
(13 TeV)
35.9 fb CMS
Preliminary > 110 GeV
ll T2
M channel µ e
[GeV]
miss T
p 50 100 150 200 250 300 350 400 450 500 Obs / Fitted
0.5 1.0 1.5
21 / 39
Scalar mediator [GeV]
φ
m
10
2
10
q
Upper limit on g
0.5 1.0 1.5 2.0 2.5
Median expected 95% CL 68% Expected 95% Expected Observed
(13 TeV)
35.9 fb =1 GeV
χ
=1, m
χ
Scalar, Dirac, g
CMS
Preliminary
Pseudoscalar mediator [GeV]
a
m
10
2
10
q
Upper limit on g
0.5 1.0 1.5 2.0 2.5
Median expected 95% CL 68% Expected 95% Expected Observed
(13 TeV)
35.9 fb =1 GeV
χ
=1, m
χ
Pseudoscalar, Dirac, g
CMS
Preliminary
This analysis has set the following observed (expected) upper limits on signal produc- tion for the model considered, at 95% confidence level: ◮ Up to 160 (240) GeV for scalar mediators ◮ Up to 220 (320) GeV for pseudoscalar mediators Results published in 2016 with 35.9 fb−1 in
22 / 39
Analysis published in the
Journal of High Energy Physics
with the 2016 data (35.9 fb−1). Results interpretation: simplified model in which a scalar or pseudoscalar mediator particle couples to a single top quark and decays into DM particles. Main characteristics of this model: ◮ First search for single top + DM mediated by a neutral spin-0 particle ◮ Lower production cross-section than the t¯ t + DM ◮ Non-flavour violating single top quark processes kinematically favored ◮ Takes advantage of the large Yukawa coupling with massive particles ◮ Additional production of DM in association with a single top quark being studied for the first time ◮ Search for an excess of data over the SM expectations in the pmiss
T
spectrum ◮ Two different signal regions studied (with, without lepton)
23 / 39
The results obtained from the single top and t¯ t analyses have been combined, im- proving the limits by a factor 2. Scalar mediator
(GeV)
φ
m
50 100 150 200 250 300 350 400 450 500
th
σ / σ
1 2 3 4 5 6 7 8 9 10
(13 TeV)
35.9 fb CMS Preliminary
=1 GeV
χ
=1, m
q
=g
χ
Scalar, Dirac, g Observed Expected 95% CL (t+DM, tt+DM) 1 s. d. ± 2 s. d. ± Expected 95% CL (t+DM) Expected 95% CL (tt+DM)
Pseudoscalar mediator
(GeV)
a
m
50 100 150 200 250 300 350 400 450 500
th
σ / σ
1 2 3 4 5 6 7 8 9 10
(13 TeV)
35.9 fb CMS Preliminary
=1 GeV
χ
=1, m
q
=g
χ
Pseudoscalar, Dirac, g Observed Expected 95% CL (t+DM, tt+DM) 1 s. d. ± 2 s. d. ± Expected 95% CL (t+DM) Expected 95% CL (tt+DM)
The combination of the analyses leads to the exclusion of scalar (pseudoscalar) medi- ators up to 290 (300) GeV at 95% confidence level. This analysis provides the most stringent limits derived at the LHC for these new spin-0 mediator particles.
24 / 39
Analysis published in the
Journal of High Energy Physics
with the 2016 data (35.9 fb−1). ◮ Search for a similar final state as the previous analysis with a hadronically decaying Lorentz-boosted top quark and DM ◮ However, no additional jet or W boson is produced in this case Flavor changing neutral current V Additional properties of this model: ◮ Consider events with a top quark decaying to a bottom quark and a W, where the W boson decays to two light quarks (large BR ∼ 67%, reconstructable) and a high pmiss
T
(> 250 GeV) ◮ Production heavily suppressed in the standard model (SM) → signature used to probe the production of DM particles via a flavor-violating mechanism ◮ New techniques (BDT) for the reconstruction and identification of the decay products of the highly Lorentz-boosted top quarks used
25 / 39
◮ Major backgrounds: Z+jets, W+jets, t¯ t, constrained from CR and transfer factors ◮ BDT defined as discriminator based on the jet substructure, to distinguish between top quarks and gluons/lighter quarks Vector mediator Axial mediator → Vector and axial mediators have been excluded up to ∼ 1.8 TeV.
26 / 39
26 / 39
As explained, many of the BSM models require new particles that couple to quarks/gluons and decay to dijets: ◮ The dijet invariant mass spectrum is usually studied ◮ An eventual signal is expected to appear as a bump instead of a global excess in the spectrum ◮ The width of the resonances increase with the coupling, and can vary from narrow to broad Usual issue with these analyses: the trigger limitation, due to the high dijet production cross-section: ◮ We either have to go at high pT to select events ◮ Find ways to speed up the reconstruction at the trigger level ◮ Or try to recover inefficiencies due to this issue → This will be the main the focus of the following analyses.
27 / 39
This search for narrow resonances decaying to two jets has been published last year in the
Journal of High Energy Physics
with the 2016 data. The global strategy is quite simple: ◮ Select events with two reconstructed jets ◮ Fit a smooth function to the mjj spectrum ◮ Search for eventual excesses/bumps due to new resonances
Dijets from calorimeter information
600 800 1000 1200 1400 1600 1800 2000
[pb/TeV]
jj
/dm σ d
(13 TeV)
27 fb
CMSPreliminary
Data Fit gg (0.75 TeV) qg (1.20 TeV) qq (1.60 TeV)
/ ndf = 20.3 / 20 = 1.0
2
χ Wide Calo-jets < 2.04 TeV
jj
0.49 < m | < 1.3 η ∆ | < 2.5, | η |
6
10
5
10
4
10
3
10
2
10 10 1
1 −
10
Dijet mass [TeV] Uncertainty (Data-Fit)
3 − 2 − 1 − 1 2 3 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Dijets reconstructed from PF algorithm
[pb/TeV]
jj
/dm σ d
(13 TeV)
36 fb
CMSPreliminary
Data Fit gg (2.0 TeV) qg (4.0 TeV) qq (6.0 TeV)
/ ndf = 38.9 / 39 = 1.0
2
χ Wide PF-jets > 1.25 TeV
jj
m | < 1.3 η ∆ | < 2.5, | η |
4
10
3
10
2
10 10 1
1 −
10
2 −
10
3 −
10
4 −
10
Dijet mass [TeV] Uncertainty (Data-Fit)
3 − 2 − 1 − 1 2 3 2 3 4 5 6 7 8
28 / 39
Limits on the production of new particles decaying to parton pairs are set using the dijet mass spectrum: ◮ Approach sensitive to high mass mediators (≥ 1 TeV), since the sensitivity to low mass mediators is limited by the trigger bandwidth ◮ Both narrow and wide resonances are considered, as resonances containing gluons (emiting more QCD radiation than the quarks) are wider ◮ Vector and axial-vector mediators have been considered, in simplified model of DM interactions ◮ The dijet searches allow to put limits to the couplings gq between quarks and mediators Vector-axial mediator Vector mediator → Dark matter mediators between 600 GeV and 2.6 TeV are excluded with this anal- ysis.
29 / 39
The dijet searches also allow to put limits to the couplings gq between quarks and mediators since at a fixed resonance mass, we can exclude models with smaller cou- plings when the sensitivity increases. The 95% CL upper limits on the universal quark coupling g′
q as a function of reso-
nance mass for a Z ′ resonance have been calculated. Vector-axial mediator
Z' mass [GeV]
500 1000 1500 2000 2500 3000 3500
'
q
g Coupling
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
CMS Preliminary
(13 TeV)
& 36 fb
27 fb 95% CL upper limits
/ 2
Med
> M
DM
m = 0
DM
m Observed Expected 1 std. deviation ± 2 std. deviation ± ←→ Low mass High mass
30 / 39
A dedicated low mass search has been performed in order to recover the ineficiencies due to the trigger bandwith of the previous analysis. Narrow vector resonances decay- ing to a quark-antiquark pair are searched for: ◮ Looking at events with an energetic ISR jet and highly boosted jets to reduce the backgrounds ◮ In this case, the decay products of the resonance are merged in a single massive jet ◮ The soft drop jet mass is studied:
◮ Soft and wide-angle radiation inside the jet removed (from parton shower, PU interactions
Number of events/5 GeV
5000 10000 15000 20000 25000 Data Multijet pred. Total SM pred. qq+jets → W qq+jets → Z =135 GeV
Z'=1/6, m
qZ'(qq), g (13 TeV)
35.9 fb
CMS Preliminary
: 500-600 GeV
T
p (GeV)
PUPPI SD
AK8 m 40 60 80 100 120 140 160 180 Data/Prediction
0.9 1 1.1
Mediator mass [GeV]
2
10
3
10
[GeV]
DM
m
200 400 600 800 1000 1200 1400
Dijet, resolved Dijet, boosted DM = 2 x m Med M = 1 DM = 0.25, g q g Dijet, resolved Dijet, boosted Dirac DM Vector mediator
Preliminary CMS
(13 TeV)
35.9 fb 0.12 ≥
2
h
c
Ω
◮ The 95% CL limits have been obtained for the vector mediator case ◮ The 50-(240)300 GeV mediator masses range has been excluded
31 / 39
An additional search for DM in the dijet final state using the angular distribution of the jets has been published in the
European Physical Journal C
The angular distribution: ◮ Is particularly effective in the case of wide resonances or non-resonant searches ◮ Is expressed as χdijet = e|y1−y2| where y1 and y2 are the rapidities of the two highest pT jets of the event ◮ Is divided into several Mjj categories to increase the sensitivity The angular searches are interesting mainly to recover sensitivity when the mediator has a high decay width or when the BSM production mechanism is non-resonant → classical bump searches inefficient in this case
2 4 6 8 10 12 14 16
dijet
χ /d
dijet
σ d
dijet
σ 1/
0.05 0.1 0.15 0.2 0.25 0.3
> 6.0 TeV
jj
M Data NLO QCD+EW prediction (CI) = 13 TeV
+ LL
Λ (GRW) = 10 TeV
T
Λ = 6 ADD) = 8 TeV
ED
(n
QBH
M = 1.0) = 4.5 TeV
q
(DM g
Med
M
Preliminary CMS (13 TeV)
35.9 fb
2 4 6 8 10 12 14 160.05 0.1
< 6.0 TeV
jj
M 5.4 <
2 4 6 8 10 12 14 160.05 0.1
< 5.4 TeV
jj
M 4.8 <
2 4 6 8 10 12 14 160.05 0.1
< 4.8 TeV
jj
M 4.2 <
2 4 6 8 10 12 14 160.05 0.1
< 4.2 TeV
jj
M 3.6 <
2 4 6 8 10 12 14 160.05 0.1
< 3.6 TeV
jj
M 3.0 < dijet
χ
2 4 6 8 10 12 14 16 0.05 0.1
< 3.0 TeV
jj
M 2.4 <
32 / 39
The eventual presence of new physics is expected to show up as an excess of events at low χ values. Since no significative excess has been found, the limits have been obtained consider- ing both spin-1 vector and axial-vector mediators in the plane of dark matter vs medi- ator mass. Vector mediator
[GeV]
Med
M
1000 2000 3000 4000 5000 6000 7000 8000
[GeV]
DM
M
500 1000 1500 2000 2500 3000
D M
= 2 x m
M e d
M . 1 2 ≥
2
h
c
Ω = 1.0
DM
g = 1.0
q
g & Dirac DM Vector mediator CMS 95% CL Dijet Chi Observed Dijet Chi Expected Preliminary CMS (13 TeV)
35.9 fb
Axial-vector mediator
[GeV]
Med
M
1000 2000 3000 4000 5000 6000 7000 8000
[GeV]
DM
M
500 1000 1500 2000 2500 3000
D M
= 2 x m
M e d
M 0.12 ≥
2
h
c
Ω = 1.0
DM
g = 1.0
q
g & Dirac DM Axial-vector mediator CMS 95% CL Dijet Chi Observed Dijet Chi Expected Preliminary CMS (13 TeV)
35.9 fb
→ This search is able to exclude dark matter mediator with masses between 2.5 and 5 TeV for both the vector and axial-vector mediators, for the couplings considered.
33 / 39
33 / 39
To be published in Physics Letter B (35.9 fb−1), currently available in
arXiv
Several Higgs production modes studied: ◮ qqH (VBF): two jets, with a large mjj and a large η/small |∆φjj| separation → most sensitive channel ◮ Z(→ ll), Z/W(→ qq) as main backgrounds estimated in CR from data ◮ gg → H + high pT jet ◮ Search for a high pmiss
T
(> 250 GeV)
Events / GeV
3 −
10
2 −
10
1 −
10 1 10
2
10
3
10
4
10
5
10
Data )+jets (QCD) ν ν Z( )+jets (QCD) ν W(l )+jets (EW) ν ν Z( )+jets (EW) ν W(l Top quark Dibosons Other backgrounds inv. → qqH(125) inv. → ggH(125)(13 TeV)
35.9 fb
CMS
Preliminary B-only fit
Data / Pred. 0.5 1 1.5
Pre-fit Post-fit Uncertainty
[GeV]
jj
m
1000 2000 3000 4000 5000 Unc. (Data-Pred.) 2 − 2
◮ The Higgs to invisible process exists in the SM, but is rare (H → ZZ → νν, BR ∼ 0.1%) ◮ A shape analysis of the mjj distribution is performed instead of a counting experiment previously ◮ Dominant background is V+jets (∼ 95%), estimated from four mutually exclusive control regions:
◮ V+jets (EW): jets from W/Z, kinematically close to VBF signal events and more important at high mjj ◮ V+jets (QCD), with QCD jets
34 / 39
The signal is expected to show up as an excess of events at large values of mjj. Since the data is in agreement with the SM expectations, upper limits on the signal production cross-section have been set at the 95% confidence level, considering differ- ent Higgs production mechanisms and their combination:
Combined VBF-tag Z(ll)H-tag V(qq')H-tag ggH-tag
SM
σ inv.)/ → x B(H σ 95% CL upper limit on 0.2 0.4 0.6 0.8 1 1.2 1.4
Observed Median expected 68% expected 95% expected
(13 TeV)
35.9 fb CMS Preliminary
The combination gives an observed (expected) upper limit of 0.19 (0.15) on the branching ratio of the Higgs to invisible process. The results have also been interpreted in terms of DM candidate through Higgs portal models, providing the strongest constraints on the fermion (scalar) DM particles up to 18 (7) GeV.
35 / 39
35 / 39
The same combination can be performed considered spin-0 mediators as well (pseu- doscalar, scalar mediators): Pseudoscalar mediator Scalar mediator In this case, considering a Dirac DM of 1 GeV with couplings gDM = 1.0 and gq = 1.0, the exclusion is mainly driven by: ◮ The t/¯ t+DM analysis at low mediator masses ◮ The Mono-jet/Mono-V analysis at higher masses Pseudoscalar mediators have been excluded up to ∼ 400 GeV, and scalar mediators up to ∼ 150 GeV.
36 / 39
All the previous results can be combined together, first for the spin-1 mediators (axial- vector, vector mediators): Axial-vector mediator Vector mediator In this case, considering a Dirac DM with couplings gDM = 1.0, gq = 0.25 and gl = 0, the exclusion is mainly driven by the dijet analysis, exluding mediator masses up to ∼ 2.5 TeV for both axial-vector and vector mediators.
37 / 39
The previous results can be further reinterpreted in terms of DM-nucleon scattering, to compare the results with the ones obtained by the direct detection experiments, as explained in this
publication .
This is done both for the spin-dependant (SD) and spin-independant (SI) interactions. SD interaction SI interaction For this reinterpretation, Dirac DM candidates are considered, with the usual cou- plings gDM = 1.0, gq = 0.25 and gl = 0.
38 / 39
The main searches for DM production with the CMS detector have been presented. Different signatures have been studied, such as the: ◮ Production of dark matter in association with SM particles ◮ Production of SM particles from the decay of dark matter mediators ◮ Production of dark matter through the Higgs portal No significant discrepancies with the SM expectaction have been observed. The re- sults have then been interpreted by setting limits on the mass and couplings of the DM interaction mediators considering simplified DM models. These results are then reinterpreted in terms of DM-nuclei interaction to compare them with direct detection experiments. Most of the analyses presented here are using the 2016 dataset: ◮ Most of them will be updated and improved by taking advantage of the complete Run2 dataset (2016, 2017, 2018) ◮ Four times more data waiting to be analyzed (∼ 150 fb−1 of data instead of ∼ 36 fb−1) ◮ Stay tuned for more exciting news in the next few months!
39 / 39
39 / 39
Additional (smaller) instrumental effects can also lead to anomalous pmiss
T
values, such as: ◮ ECAL mismeasurements from beam halo or eventual dead cells ◮ HCAL mismeasurements from electronic noise and direct particle interactions with the light guides and photomultiplier tubes of the forward calorimeter These effects can be estimated and substracted, leading to a nice agreement even in the pmiss
T
distribution tail. More details in
CMS-PAS-JME-16-004
[GeV]
miss T
E
500 1000 1500 2000 2500 3000
Events / 30 GeV
1 −
10 1 10
2
10
3
10
4
10
5
10
Top quark EWK QCD Data after cleaning Data before cleaning
(13 TeV, 2016)
12.9 fb
CMS
Preliminary
40 / 39
Each of the decay channel has its own strategy and advantages: ◮ h → b¯ b has the highest BR and uses a simultaneous CR/SR fit for the baryonic Z’ and parametric fits for the Z’-2HDM. ◮ h → γγ exhibits a good reconstruction of the Higgs invariant mass and uses a fit to the diphoton invariant mass in two pmiss
T
categories. ◮ h → ττ has lower backgrounds and considers the 3 leptonic decays of the τ with the highest BR (simultaneous fit on the Higgs reconstructed mass) ◮ h → WW considers the fully leptonic decay of the Ws and the eµ channel to reduce the backgrounds ◮ h → ZZ also consideres the fully leptonic decay of the Z. The h → ττ, h → WW and h → ZZ benefits from lower backgrounds and can be competitive for signals with soft pmiss
T
spectrum.
41 / 39