Search for the SUSY Electroweak Sector at ep Colliders

Georges Azuelos,1, 2, ∗ Monica D’Onofrio,3, † Sho Iwamoto,4, 5, 6, ‡ and Kechen Wang7, §

1Universite de Montreal, Montreal, Quebec H3C 3J7, Canada2TRIUMF, Vancouver V6T 2A3, Canada

3University of Liverpool, Oliver Lodge Laboratory, P.O. Box 147,Oxford Street, Liverpool L69 3BX, United Kingdom

4Dipartimento di Fisica e Astronomia, Universita di Padova, Via Marzolo 8, I-35131 Padua, Italy5INFN, Sezione di Padova, Via Marzolo 8, I-35131 Padua, Italy

6ELTE Eotvos Lorand University, Pazmany Peter setany 1/A, H-1117 Budapest, Hungary7Department of Physics, School of Science, Wuhan University of Technology, 430070 Wuhan, Hubei, China

The sensitivity of future electron-proton colliders, the LHeC and FCC-eh, to weakly-produced su-persymmetric particles is evaluated in this article. Supersymmetric scenarios where charginos (χ±1 )and neutralinos (χ0

1 and χ02) are nearly degenerate in mass are considered. Two sets of models, which

differ in the mass of sleptons (˜), are studied. Under the hypothesis that slepton masses are at themulti-TeV scale (“decoupled” scenario), the production processes for charginos and neutralinos atep colliders, p e− → j e− χχ with χ = χ0

1, χ±1 or χ02, are considered. For the models where slepton

masses are above but close to χ±1 , χ02 masses (“compressed” scenario), contributions from the pro-

cesses p e− → j χ e−L and j χ ν followed by the decays e−L → χ01,2+e− and ν → χ+

1 +e− are also takeninto account. These scenarios are analysed with realistic detector performance, using multivariatetechniques. Effects of systematic uncertainties and electron beam polarization dependence are alsodiscussed. The reach is found to be complementary to the one obtained at pp colliders, in particularfor the compressed-slepton scenario.

I. INTRODUCTION

Supersymmetry (SUSY) is one of the most promising newphysics scenarios beyond the Standard Model (SM). Ithas been explored widely in various production and de-cay channels at the Large Hadron Collider (LHC). Dueto the large production cross sections in strong interac-tions at pp colliders, the current LHC experiments haveproduced impressive constraints on the SUSY colouredsector (squarks and gluinos), excluding masses up to ap-proximately 2 TeV at a centre-of-mass energy of

√s =

13 TeV by the CMS collaboration with integrated lumi-nosity of 137 fb−1 [1, 2] and by the ATLAS collaborationwith 139 fb−1 integrated luminosity [3].

Because of their much lower direct production crosssections, less stringent constraints have been placed onweakly-produced SUSY particles, namely neutralinos χ0,

charginos χ±, and sleptons ˜±. Here, χ0 and χ± (col-lectively referred to as electroweakinos) are the SUSYpartners of the Higgs and gauge bosons described in themass eigenstates. If R-parity is conserved, the lightestneutralino χ0

1 can be the lightest SUSY particle (LSP)and a candidate for dark matter.

A variety of searches for χ±, χ0, and ˜± at√s =

13 TeV have been reported by the ATLAS [4–7] andCMS [8–15] collaborations with 36 fb−1 of integratedluminosity. Recently, the collaborations have updated

∗ georges.azuelos@cern.ch† Monica.D’Onofrio@cern.ch‡ sho.iwamoto@ttk.elte.hu§ kechen.wang@whut.edu.cn (Corresponding author)

the analyses with up to 139 fb−1 integrated luminos-ity [16–21]. Most of the studies target the processes

pp → χ±1 χ02 → W±Z χ0

1χ01 and pp → ˜+ ˜− → l+l−χ0

1χ01,

exploiting events characterized by multiple charged lep-tons and large missing transverse momentum,�ET . Dedi-cated searches for decays through the Higgs boson (pp→χ±1 χ

02 → W±h χ0

1χ01) are also performed. Searches

for electroweakinos decaying through sleptons excludemasses above 1 TeV (see for example [5]). Lower boundsof O(100) GeV on χ±1 /χ

02 masses are obtained, depend-

ing on the scenarios; for example, Ref. [6] excludes χ±1 /χ02

decaying via W± and Z bosons up to approximately 600GeV with 95% confidence level (CL), while Ref. [21] re-ports that the left-handed selectron eL masses can be ex-cluded up to approximately 550 GeV. These limits mostlyapply to scenarios characterised by a large mass differ-ence, ∆m, between χ0

1 and the next lightest supersym-

metric particle (NLSP), either χ±1 /χ02 or ˜±. In the case

of small mass differences (e.g., 1 GeV < ∆m < 50 GeV,with ∆m still sufficiently large to guarantee promptχ±1 /χ

02 decays), W± and Z bosons are off-shell and sce-

narios are referred to as “compressed”: the visible decayproducts from ˜and χ±1 /χ

02 have very soft transverse mo-

menta (pT ) and the SM backgrounds are kinematicallysimilar to the signal. The analyses therefore become chal-lenging and sensitivities decrease substantially.

Dedicated searches have been recently developed bythe LHC experiments to target low-∆m scenarios. Theyexploit the recoil of the sparticles system against an ini-tial state radiation (ISR) jet for efficient detection of softdecay products. Assuming ∆m ∼ 5 GeV, the ATLAScollaboration has excluded eL masses up to around 160GeV and χ±1 /χ

02 masses up to around 200 GeV [17]. A

similar reach has been reported by the CMS collabora-

arX

iv:1

912.

0382

3v2

[he

p-ph

] 2

5 M

ay 2

020

2

tion [12], as well as searches exploiting vector-boson fu-sion (VBF) production [15]. Phenomenological studieson weakly-produced SUSY particles in compressed sce-narios have been also reported in Refs. [22–33]. The stud-ies on the Bino-Wino coannihilation scenario at the LHCwith

√s = 13 TeV and the high-energy LHC (HE-LHC)

with√s = 27 TeV can also be found in [34, 35].

Studies on the potential of the high-luminosity LHC(HL-LHC) [36], which foresees 3 ab−1 of data takenat a centre-of-mass energy of 14 TeV, have shown thatsearches for low-momentum leptons and ISR-jet boostwill be sensitive to chargino masses up to 400 (350) GeVfor ∆m(χ±1 /χ

02, χ

01) ∼ 5 GeV, and to mass splittings be-

tween 1 and 50 GeV, assuming Wino-like (Higgsino-like)cross sections for electroweakino productions. Similarsearch techniques can also be used to target pair pro-duced sleptons in compressed scenarios. It is worth not-ing that, to suppress SM backgrounds, analyses targetingvery small ∆m require soft-momentum leptons from thedecays of the sleptons, charginos or neutralinos and arethus complementary to searches targeting very large ∆mvia multiple high-pT leptons. Regions of intermediate∆m(χ±1 /χ

02, χ

01) (∼ 20–50 GeV) may still be elusive after

the HL-LHC.

This article focuses on compressed electroweakinos sce-narios, produced assuming Wino-like cross sections andmass differences between χ±1 , χ

02 and χ0

1 small, O(GeV),but still allowing their prompt decay. Two SUSY sce-narios with different hypotheses on the charged sleptonmasses are considered to evaluate the sensitivity of futureep colliders, the Large Hadron electron Collider (LHeC)and the electron-hadron mode of the Future Circular Col-lider (FCC-eh), to discover SUSY electroweakinos andsleptons. The search uses the multivariate analysis ap-proach (MVA) and Monte Carlo (MC) generated eventspassed through a realistic detector-level simulation. Theelectron and proton beam energies are assumed to be60 GeV×7 TeV (60 GeV×50 TeV) at the LHeC (FCC-eh),which correspond to

√s = 1.3 TeV (3.5 TeV). The maxi-

mal integrated luminosity at the LHeC is expected to be1 ab−1, while it could reach 2.5 ab−1 at the FCC-eh aftera 25-year running period. The centre-of-mass energy,

√s,

of ep colliders are considerably lower than the HL-LHC.However, since there are no gluon-exchange diagrams,the SM QCD backgrounds, which are dominant at ppcolliders, are much smaller. Furthermore, the number ofadditional interactions in the same event (pile-up) is neg-ligible at ep colliders, whilst it is expected to be very largeat the HL-LHC. Previous studies on Higgsino-like χ0/χ±

production at ep colliders can be found in Ref. [37, 38],where Ref. [37] focuses on decoupled-scenario with lightHiggsinos, and assume a proton beam of 50 (7) TeV en-ergy, colliding with an electron beam of 60 (60–140) GeVenergy at the FCC-eh (LHeC), while Ref. [38] exploresthe parameter space for long-lived Higgsinos.

The article is organised as follows. Sec. II presents theSUSY models considered. In Sec. III data simulation, sig-nal and background processes and search strategy are re-

ported. In Sec. IV, the results of the compressed-sleptonscenario are presented. The results of the decoupled-slepton scenario are reported in Sec. V. Conclusions anddiscussions on the effects of electron beam polarizationsare presented in Sec. VII.

II. SUSY SCENARIOS

FIG. 1. Some representative production diagrams for thesignal processes considered in this article. The decoupled-slepton scenario includes only the upper two diagrams, whilethe compressed-slepton scenario includes both the upper andlower diagrams.

Two SUSY compressed scenarios with different hypothe-ses on the charged slepton masses are considered. In thefirst scenario, referred to as the “decoupled-slepton” sce-nario, we assume that the only SUSY particles withinkinematic reach are the electroweakinos χ0

1,2 and χ±1 .

The LSP χ01 is assumed to be Bino-like, the χ0

2 and χ±1are Wino-like and degenerate in masses, and the massdifference between χ0

1 and χ±1 is small (∆m ∼ 1 GeV).All other SUSY particles are at the multi-TeV scale andtherefore decoupled. The upper two diagrams in Fig. 1represent the typical production processes in this sce-nario; charginos and neutralinos are produced via VBFprocesses. This scenario is motivated by dark mattercoannihilation arguments [39–42]. For a Bino-like darkmatter, the annihilation cross section is usually too low.The Wino-like χ±1 /χ

02 with masses of order 1-10 of GeV

larger than the LSP enhance the coannihilation pro-cesses, which may result in the dark matter relic den-sity consistent with observations. In the second sce-nario, referred to as “compressed-slepton” scenario, theleft-handed charged sleptons ˜±

L and sneutrinos ν are as-sumed to be kinematically accessible and slightly heavierthan χ±1 /χ

02. The signal production in this case includes

both the upper and lower diagrams in Fig. 1. This sce-nario is also motivated by coannihilation arguments [43].Furthermore, since the main SUSY contributions to the

3

muon anomalous magnetic moment, (g − 2)µ, are givenby the neutralino-smuon and chargino-sneutrino loop di-agrams, SUSY mass spectra containing neutralinos andsleptons of mass O(100) GeV may explain the discrep-ancy of the measured (g−2)µ with theoretical predictions(see Ref. [44] for a recent summary).

In the decoupled-slepton scenario, since the chargedsleptons and sneutrinos ν are at the multi-TeV scale, sig-nal events pe− → je−χχ (χ = χ0

1, χ02, χ±1 ) are produced

via VBF processes of charginos and neutralinos only (c.f.upper two diagrams in Fig. 1). In the compressed-slepton

scenario, however, since the left-handed slepton ˜L and ν

are slightly heavier than the χ±1 and χ02, the signal events

pe− → je−χχ include not only VBF processes of χ±1 and

χ02 but also the direct production of ˜

L and ν followedby the decays of e−L → χ0

2,1 + e− and ν → χ+1 + e− (i.e.

pe− → jχe−L , jχν → je−χχ). Thus both the upper andlower diagrams in Fig. 1 contribute to the signal rate inthe compressed-slepton scenario.

FIG. 2. Production cross sections σ(pe− → je−χχ) for boththe compressed- and decoupled-slepton scenarios at the FCC-eh and for the compressed-slepton scenario at the LHeC withunpolarized e− beam as varying the masses of χ±1 and χ0

2.Here χ0

1 is Bino while χ±1 and χ02 are Wino with almost degen-

erate masses. Cross sections are calculated at leading order byMadGraph5 aMC@NLO version 2.4.3. For the compressed-slepton scenario, the mass difference ∆m = m˜−mχ±

1 ,χ02

=

35 GeV, while the decoupled-slepton scenario assumes slep-tons are heavy and decoupled.

Figure 2 shows the signal cross sections σ(pe− →je−χχ) at FCC-eh for both the compressed- anddecoupled-slepton scenarios as a function of the massesof the χ±1 and χ0

2, and at the LHeC only for thecompressed-slepton scenario since the decoupled-sleptonscenario yields much smaller cross section. Cross sec-tions are calculated at leading order (LO) by Mad-Graph5 aMC@NLO version 2.4.3 [45] . Higher-order cor-rections are expected to be at the same level as those forelectroweakino pair-productions at the LHC [46], in gen-eral up to 20% with respect to LO calculations, but theyare not taken into account here. This leads to conserva-tive results in terms of signal sensitivity. For both SUSYscenarios, the signal final state events are characterisedby the presence of one jet, one electron, large �ET , and

undetected very soft-momenta particles arising from χ±1or χ0

2 decays.

III. SEARCH STRATEGY

Simulated data are generated at parton-level by Mad-Graph5 aMC@NLO 2.4.3 by importing the generalMSSM model with the model parameter file preparedby the SUSY-HIT package [47]. Pythia performs partonshowering and hadronization, and Delphes [48] is usedfor detector simulation. The detectors at ep collidersare assumed to have a cylindrical geometry comprisinga central tracker followed by an electromagnetic and ahadronic calorimeter. The forward and backward regionsare also covered by a tracker, an electromagnetic and ahadronic calorimeter. The angular acceptance of chargedtracks and the detector performance in terms of momen-tum and energy resolution of electrons, muons, and jetsare based on the LHeC (FCC-eh) detector design [49, 50].In particular, electrons are selected in the pseudorapidityrange of −4.3 < η < 4.9 (−5.0 < η < 5.2) at the LHeC(FCC-eh), while jets are selected within an η range be-tween ± 5 for the LHeC and ± 6 for the FCC-eh. Forthe simulation, a modified Pythia version tuned for theep colliders 1 and the Delphes card files for the LHeC andFCC-eh detector configurations [52] are used.

Due to the presence of large missing transverse mo-mentum in the final state, processes with production ofneutrinos will contribute to the SM background. Weseparate the background into two categories: the 2-neutrino process p e− → j e− νν and the 1-neutrino pro-cess p e− → j e− `ν, where ν stands for the neutrino andanti-neutrino with all flavors of e, µ, τ . The 2-neutrinoprocess has the same final state as the signal and is an ir-reducible background. The 1-neutrino process has a largecross section and will therefore have a non-negligible con-tribution to the background if one of the two leptons isundetected.

To make our simulation more realistic, we set follow-ing cuts at the parton level in the MadGraph: pT (j) >5 GeV, pT (`) > 1 GeV, |η(j)| < 10, |η(`)| < 10, etc.,which are looser than the default cuts. The following pre-selections in the final state after detector simulation areapplied for LHeC (FCC-eh, reported if differing) studies:

1. At least one jet with pT > 20 GeV, |η| ≤ 5.0(6.0);

2. Exactly one electron with pT > 10 GeV, −4.3 <η < 4.9 (−5.0 < η < 5.2);

1 The modification concerns the treatment of beam remnants,that in some cases might cause generation failure. The param-eter setting preserves the correct flow of the generation in theseinstances. As a cross check, the hadronic showering is not ex-pected to change the kinematics of the DIS scattered lepton. Thishas been shown, see page 11 of Ref. [51]: a very good level ofagreement of the NC DIS electron kinematics with and withoutthe ep-customized Pythia showering is found.

4

3. No b-jet with pT > 20 GeV;

4. No muon or tau with pT > 10 GeV;

5. Missing transverse momentum �ET > 50 GeV.

After the pre-selections, 17 observables listed in the fol-lowing are used as inputs to the TMVA package [53] toperform a multi-variate analysis through a boost decisiontree (BDT) approach.

1. Global observables:

1.1. the missing transverse momentum �ET ;

1.2. the scalar sum of pT of all jets, HT .

2. Observables for the visible objects:

2.1. pT and the pseudorapidity η of the first lead-ing jet j1 and the first leading electron e1:pT (j1), η(j1), pT (e1), η(e1);

2.2. the pseudorapidity difference ∆η and the az-imuthal angle difference ∆φ between j1 ande1: ∆η(j1, e1), ∆φ(j1, e1);

2.3. the invariant mass M , pT and η reconstructedfrom the 4-vector of the system of j1 and e1:M(j1 + e1), pT (j1 + e1), η(j1 + e1).

3. Angular correlations between the missing trans-verse momentum and visible objects:

3.1. ∆φ between �ET and j1, e1, or j1 + e1:∆φ(�ET , j1), ∆φ(�ET , e1), ∆φ(�ET , j1 + e1);

3.2. the transverse mass MT of the system of

�ET and j1, e1, or j1 + e1: MT (�ET , j1),MT (�ET , e1), MT (�ET , j1 + e1).

The BDT score for each event is used to separate thesignal events from the SM background. The BDT thresh-old value is optimized for each SUSY scenario as well asfor each collider, since kinematical distributions mightvary as a function of mχ±

1 ,χ02

and the beam energies. The

number of signal and background events passing the BDTselection, Ns and Nb, are used to calculate the signifi-cance. Assuming no systematic uncertainty, the statisti-cal significance, σstat, of the potential signal is evaluatedas

σstat =

√2[(Ns +Nb)ln(1 +

NsNb

)−Ns]. (1)

Including a systematic uncertainty σb in the evaluationof the number of background events, the significance isgiven by

σstat+syst =

[2

((Ns +Nb) ln

(Ns +Nb)(Nb + σ2b )

N2b + (Ns +Nb)σ2

b

− N2b

σ2b

ln

[1 +

σ2bNs

Nb(Nb + σ2b )

])]1/2.

(2)

IV. COMPRESSED-SLEPTON SCENARIO

The compressed-slepton scenario is characterised bythree sets of masses, which are

m˜,ν , mχ±1 ,χ

02, mχ0

1(3)

in the descending order, where m˜,ν represents the massof charged sleptons, here assumed dominated by the left-handed component, or sneutrinos. In this article, we as-sume nearly degenerate electroweakinos, with mχ±

1 ,χ02

=

mχ01+ 1 GeV, and prompt χ±1 , χ

02 decays. The mass split-

ting between ˜L and ν is generally below 10 GeV for tanβ

between 3 and 40. In this study, the benchmark tanβ =30 is assumed, which results in the mass difference be-tween ˜

L and ν of around 9 GeV.In this section, the ∆m symbol is used to denote the

mass difference between the ˜ and χ±1 , χ02, i.e.

∆m = m˜−mχ±1 ,χ

02. (4)

A. Results with fixed ∆m

We first discuss the case with ∆m = 35 GeV, where35 GeV is used as a proxy for intermediate-low-∆m sce-narios. The branching ratios for sleptons are BR(e−L →χ02,1 + e−) ≈ 40% and BR(ν → χ±1 + e−) ≈ 60%. The

other decay channels, e−L → χ−1 + ν and ν → χ02,1 + ν,

have no electrons in the final state (except for very softleptons), and therefore will not contribute to the signal.

Figure 9 of the Appendix shows kinematic distri-butions of the input observables, both for signal andbackground, for the FCC-eh case assuming an unpolar-ized electron beam (P(e−)=0). A SUSY model withmχ±

1 ,χ02

= 400 GeV and production e−p → j e− χχ is

assumed, while the two sources of SM backgrounds,e−p→ j e− νν and j e− `ν, are shown separately. Distri-butions of the corresponding BDT response are presentedin the upper panel of Fig. 3.

FCC-eh [1 ab−1] Signal Backgroundmχ±1 ,χ

02

[GeV] 400j e− νν j e− `ν

m˜ [GeV] 435initial 4564 1.08× 106 7.96× 106

Pre-selection 3000 3.87× 105 5.71× 105

BDT > 0.262 149 600 86σstat+syst 3.3

TABLE I. Number of events after selections applied sequen-tially on SUSY signal j e− χχ with m

χ±1 ,χ

02

= 400 GeV in

the compressed-slepton scenario, and for the SM backgroundprocesses of j e− νν and j e− `ν. The numbers of events cor-respond to an integrated luminosity of 1 ab−1 at the FCC-ehwith unpolarized electron beam. The significances including5% systematic uncertainties on the background are presentedin the last row.

5

FIG. 3. Distributions of BDT response at the FCC-eh (upper)and the LHeC (lower) with the unpolarized electron beamfor signal j e− χχ (black with filled area) in the compressed-slepton scenario, and the SM background of the j e− νν (red)and j e− `ν (blue) processes after applying the pre-selectioncuts. For signals, m

χ±1 ,χ

02

is 400 GeV and 250 GeV at the

FCC-eh and LHeC, respectively.

LHeC [1 ab−1] Signal Backgroundmχ±1 ,χ

02

[GeV] 250j e− νν j e− `ν

m˜ [GeV] 285initial 1231 2.80× 105 2.01× 106

Pre-selection 453 6.60× 104 1.66× 105

BDT > 0.172 49 486 278σstat+syst 1.0

TABLE II. Number of events after selections applied sequen-tially on SUSY signal j e− χχ with m

χ±1 ,χ

02

= 250 GeV in

the compressed-slepton scenario, and for the SM backgroundprocesses of j e− νν and j e− `ν. The numbers of events corre-spond to an integrated luminosity of 1 ab−1 at the LHeC withunpolarized electron beam. At each stage, the significance in-cluding 5% systematic uncertainties on the background arepresented in the last row.

The number of events at each selection stage is shownin Table I, which corresponds to an integrated luminosityof 1 ab−1. The optimized value for the BDT is found tobe 0.262 for this SUSY model. Assuming a systematicuncertainty of 5%, i.e., σb = 0.05Nb, the number of eventsafter the optimized BDT cut, Ns = 149 and Nb = 686,

results in a significance of σstat+syst = 3.3 2. The 5% sys-tematic uncertainties on SM backgrounds are assumed toinclude experimental sources (e.g. jet energy scale andresolution, lepton identification) and modelling uncer-tainties (renormalisation and factorisation scale choices,PDF). They are expected to be constrained by semi data-driven techniques and improved Monte Carlo predictionsand the 5% value is assumed on the basis of a conserva-tive extrapolation of uncertainties estimated in pp mono-jet prospect studies at HL-LHC [54].

The same analysis is performed for the LHeC case,where we assume mχ±

1 ,χ02

= 250 GeV as the benchmark

masses. The distributions of input observables are shownin Fig. 10 and the BDT-score distribution is reported inthe lower panel of Fig. 3. The cut flow is given in Table II,where the optimized BDT cut of 0.172 is used. Consider-ing an integrated luminosity of 1 ab−1 and 5% systematicuncertainty on the background, the significance of about1.0 is obtained for this benchmark point.

FIG. 4. Significances as varying the masses of χ±1 and χ02 for

the compressed-slepton scenario. Upper plot: at the FCC-eh with unpolarized beams and integrated luminosities of 1ab−1 and 2.5 ab−1; Lower plot: at the LHeC with unpolar-ized beams and 1 ab−1 luminosity. For dashed (solid) curve,a systematic uncertainty of 0% (5%) on the background isconsidered.

In the upper panel of Fig. 4, the significance curves

2 In case the η range of the jet is restricted to ± 5.5, equivalent toa less-than-ideal calorimeter coverage at the FCC-eh, both thesignal and background rates are reduced by less than 5%, andthe significance is only slightly reduced.

6

as a function of mχ±1 ,χ

02

for the FCC-eh with unpolar-

ized electron beam and integrated luminosities of 1 ab−1

and 2.5 ab−1 are presented. With 0% (5%) systematicuncertainty on the background and 2.5 ab−1 integratedluminosity, the 2-σ limits on the χ±1 , χ0

2 mass are 616(466) GeV, while the 5-σ discovery limits are 517 (367)GeV.

The lower panel of Fig. 4 shows the significance curvesat the LHeC with unpolarized electron beam and an in-tegrated luminosity of 1 ab−1. With 0% (5%) systematicuncertainty on the background, the limits on the massare 266 (224) GeV and 227 (187) GeV corresponding tothe 2 and 5-σ significances, respectively.

B. Effects of Varying ∆m

FIG. 5. Distributions of the BDT response when fixingmχ±1 ,χ

02

= 400 GeV and varying the mass difference ∆m =

m˜−mχ±1 ,χ

02

in the compressed-slepton scenario at the FCC-

eh.

The effect of varying the assumption on ∆m is evalu-ated. A range between 5 to 65 GeV is considered, withmχ±

1 ,χ02

being fixed at 400 GeV 3. The distributions of the

BDT response at the FCC-eh for different ∆m assump-tions are presented in Fig. 5. Although the same back-ground is analyzed, the BDT distributions still change for

3 When ∆m < 9 GeV, tanβ is changed from 30 to smaller valuesuch that the LSP can still be χ0

1, rather than ν.

different ∆m assumptions, because the signal kinematicsvaries with ∆m values. As ∆m increases, the kinematicsof the 2-neutrino background j e− νν becomes similar tothat of the signal process j e− χχ, which renders its BDTdistribution largely overlapping with that of the signal.When ∆m > 35 GeV, part of the 1-neutrino backgroundj e− `ν events also begins to mimic the signal. Becauseof clearer distinctions between the BDT distributions ofthe signal and background, the BDT is most effectivein rejecting the SM background if ∆m between 10 and40 GeV.

FIG. 6. Significance shown as a function of the mass differ-ence ∆m = m˜−mχ±

1 ,χ02

when fixing electroweakino masses,

mχ±1 ,χ

02

= 400 GeV, in the compressed-slepton scenario at the

FCC-eh.

The significances assuming either 5% or 0% system-atic uncertainties on the background as a function of themass differences ∆m are reported in Fig. 6 for luminosi-ties of 1 and 2.5 ab−1. It is found that the ∆m = 20GeV can be tested with maximum efficiency. This isbecause, as shown in Fig. 5, for small ∆m, the BDTdistributions present a better separation between signaland background. However, when ∆m << 20 GeV, theelectron from the slepton and sneutrino decays becometoo soft to pass the pre-selections, pT (e−) > 10 GeV,suppressing the signal number of events.

It is worth noting that, although the significances for∆m = 12 and 5 GeV are slightly lower than ∆m = 20GeV, when 5% systematic uncertainty on backgroundis included, they are still larger than the benchmarkcase with ∆m = 35 GeV. However, when no system-atic uncertainty on background is considered, ∆m = 35GeV gives larger significance compared to ∆m = 12 and5 GeV. Overall, significances are quite sensitive to theassumption of systematic uncertainty when ∆m > 20GeV. Therefore, to enhance the discovery potential to theSUSY electroweak sector, it will be important to controlthe level of systematic uncertainties on the background.

V. DECOUPLED-SLEPTON SCENARIO

In the decoupled-slepton scenario, where the sleptons areassumed to have multi-TeV mass, the signal events are

7

mainly arising from the VBF processes of charginos andneutralinos, pe− → je−χχ. As this is a four-body pro-duction process, the signal cross sections are consider-ably lower than in the case of compressed-slepton sce-nario (cf. Fig. 2). At the LHeC they are of the order of 1fb or lower, hence the decoupled-slepton scenario is onlyconsidered for the FCC-eh case.

FIG. 7. Distributions of BDT response at the FCC-eh for sig-nal j e− χχ with m

χ±1 ,χ

02

= 250 GeV (black with filled area)

in the decoupled-slepton scenario, and SM background pro-cesses of j e− νν (red) and j e− `ν (blue) after applying thepre-selections.

As in the previous section, the kinematic distributionsof input observables for events passing the pre-selectionsare shown in Fig. 11 in the Appendix A, where thesignal events are from the process pe− → je−χχ withmχ±

1 ,χ02

= 250 GeV and the SM backgrounds are from of

j e− νν and j e− `ν. The corresponding BDT-inputs dis-tributions are presented in Fig. 7. Table III shows thenumber of events after each selection is applied sequen-tially corresponding to 1 ab−1. In the last row, signifi-cances calculated assuming 5% systematic uncertaintiesfor the backgrounds are also included.

FCC-eh [1 ab−1] Signal Backgroundmχ±1 ,χ

02

[GeV] 250j e− νν j e− `ν

m˜ [GeV] -initial 909 1.08× 106 7.96× 106

Pre-selection 399 3.87× 105 5.71× 105

BDT > 0.251 14 326 357σstat+syst 0.3

TABLE III. Number of events after selections applied sequen-tially on the signal j e− χχ with m

χ±1 ,χ

02

= 250 GeV in the

decoupled-slepton scenario, and for the SM background pro-cesses j e− νν and j e− `ν. The numbers correspond to anintegrated luminosity of 1 ab−1 at the FCC-eh with unpo-larized electron beam. The significances, calculated including5% systematic uncertainties on the background, are presentedin the last row.

Figure 8 shows the significance as a function of mχ±1 ,χ

02.

When considering 0% (5%) systematic uncertainty onthe background, the 2-σ limits on the χ±1 and χ0

2 masses

FIG. 8. Significances as a function of the masses of χ±1 andχ02 for the decoupled-slepton scenario at the FCC-eh with

unpolarized beams and integrated luminosities of 1 ab−1 and2.5 ab−1. For dashed (solid) curve, a systematic uncertaintyof 0% (5%) on the background is considered.

are 230 (125) GeV for 2.5 ab−1 luminosity. Because ofthe small signal production, the discovery power for thedecoupled-slepton scenario is limited at the future ep col-liders.

VI. EFFECTS OF BEAM POLARIZATION

The possibility of having a polarized electron beam at theLHeC and FCC-eh colliders could potentially enhance thesensitivity to electroweakinos. A high degree of longitu-dinal polarization could be envisaged (|P(e−)| = 80%).When applying an electron beam polarization of -80%(+80%) at the FCC-eh, the production cross sections ofSM background processes increase (decrease) by about55% (20%), compared to the rate in case of unpolar-ized electron beam. On the other hand, in the case ofthe compressed-slepton scenario with benchmark mχ±

1 ,χ02

= 400 GeV and m˜ = 435 GeV, the signal productioncross section increases (decreases) by about 80%. In thedecoupled-slepton scenario with mχ±

1 ,χ02

= 250 GeV, the

signal production cross section increases (decreases) byabout 40%. The changes in the cross sections are dueto the electroweak nature of the interactions involvedand the assumptions made: the sleptons considered aresuperpartners of left-handed electrons, and the gaugebosons and Wino-like electroweakinos couple more toleft-handed electrons or sleptons. A negative polariza-tion can be therefore advantageous for such searches ofelectroweak SUSY particles, whilst a positive polariza-tion might reduce the sensitivity, although a new opti-mization of the analysis would be needed in this case.

Compared with the unpolarized electron beam with1 ab−1 luminosity, the -80% electron beam polarizationincreases the significances for the benchmark mass hy-pothesis in compressed-slepton scenarios by about 40%(15%), from 9 to 13 (3.3 to 3.8) when including 0% (5%)systematic uncertainty on the background. In case ofdecoupled-slepton scenarios, the -80% electron beam po-

8

larization increases the significances by about 30% (10%),from 1.0 to 1.3 (0.32 to 0.35) when including 0% (5%) sys-tematic uncertainty on the background. Hence, thanks tothe much larger signal production, our preliminary anal-ysis indicates that the -80% electron beam polarizationcould lead to a stronger discovery potential for the SUSYelectroweak sector especially for the compressed-sleptonscenario with small systematic uncertainty. A more ac-curate simulation and estimate of acceptances and yieldsin case of polarized electron beams is needed consideringseveral mass hierarchy hypotheses, which is beyond thescope of this study and we leave it for future studies.

VII. CONCLUSIONS AND DISCUSSIONS

The LHC experiments have produced impressive con-straints on the masses of the SUSY coloured sector. Be-cause of the low direct production cross sections for neu-tralinos, charginos and sleptons, the LHC constraints ontheir masses are limited, particularly in the compressedscenarios where the experimental analysis is challeng-ing, due to the presence of soft momenta decay prod-ucts. In this article, we have reported a search strategyfor the SUSY electroweak sector and its discovery po-tential at the future electron-proton colliders, LHeC andFCC-eh. Our study shows that the search for the SUSYelectroweak sector in the compressed scenarios at the epcolliders could be complementary to the analyses at thepp colliders, mainly due to the cleaner environment andsmaller SM background at ep colliders.

We focus on the compressed scenarios where the LSPχ01 is Bino-like, χ±1 and χ0

2 are Wino-like with almostdegenerate masses, and the mass difference between χ0

1

and χ±1 is small. The signal is produced via the process“p e− → j e− χχ”, where χ = χ0

1, χ±1 or χ02. LO cross

sections are considered for the SUSY signal models, henceresults are conservative. The kinematic observables areinput to the TMVA package to perform a multivariateanalysis at the detector level.

In the compressed-slepton scenario, the case where theleft-handed slepton ˜

L and sneutrino ν are slightly heav-ier than χ±1 or χ0

2 is considered. When fixing the massdifference ∆m = m˜−mχ±

1 ,χ02

= 35 GeV and assuming no

systematic uncertainty on the background, our analysisindicates that the 2 (5)-σ limits on the χ±1 , χ0

2 mass are

616 (517) GeV for 2.5 ab−1 luminosity at the FCC-eh,and 266 (227) GeV for 1 ab−1 luminosity at the LHeC,respectively. The effects of varying ∆m are investigated:fixing mχ±

1 ,χ02

to be 400 GeV, we find that at the FCC-eh

the significance is maximal when ∆m is around 20 GeV.In the decoupled-slepton scenarios where only χ0

1, χ±1and χ0

2 are light and other SUSY particles are heavy anddecoupled, the 2-σ limits obtained on the χ±1 , χ0

2 mass

are 230 GeV for 2.5 ab−1 luminosity at the FCC-eh whenassuming no systematic uncertainty on the background.Large systematic uncertainties on the SM backgroundprocesses can substantially affect the sensitivity, hence

good control of experimental and theoretical sources ofuncertainties is very important. Finally, it is found thatthe possibility of having a negatively polarized electronbeam (P(e−) = 80%) could potentially extend the sensi-tivity to electroweakinos by up to 40%.

Searches at pp colliders usually target either hardor very soft leptons from the decays of the sleptons,charginos or neutralinos, which corresponds to eitherlarge or very small ∆m(˜, χ0

1) and ∆m(χ±1 /χ02, χ

01).

Given the difficulty to probe intermediate regions with∆m(˜, χ0

1) ∼ 35 GeV and ∆m(χ±1 /χ02, χ

01) ∼ 20 − 50

GeV, these scenarios may still be elusive even after theHL-LHC. Therefore, measurements at ep colliders mayprove to offer complementary or additional reaches, inparticular for the compressed scenarios.

ACKNOWLEDGMENTS

We thank Oliver Fisher, Satoshi Kawaguchi, Max Klein,Uta Klein, and Peter Kostka for helpful communication.We appreciate comments from members in the LHeC /FCC-eh Higgs and BSM physics study groups. K.W.thanks Christophe Grojean and Cai-dian Lu for theirsupports. G.A. is supported by National Science andEngineering Research Council (Canada). The work ofS.I. is partially supported by the MIUR-PRIN project2015P5SBHT 003 “Search for the Fundamental Lawsand Constituents”. K.W. is supported by the Excel-lent Young Talents Program of the Wuhan University ofTechnology, the National Natural Science Foundation ofChina under grant no. 11905162, and the CEPC theorygrant (2019–2020) of IHEP, CAS.

Appendix A: Distributions of Input Observables

9

FIG. 9. Kinematical distributions of some input observablesfor signal j e− χχ with m

χ±1 ,χ

02

= 400 GeV (black with filled

area) in the compressed-slepton scenario, and for the SMbackground of the j e− νν (red) and j e− `ν (blue) processesafter applying the pre-selection cuts at the FCC-eh with anunpolarized electron beam.

FIG. 10. Kinematical distributions of some input observ-ables for signal j e− χχ with m

χ±1 ,χ

02

= 250 GeV (black with

filled area) in the compressed-slepton scenario, and for theSM background of the j e− νν (red) and j e− `ν (blue) pro-cesses after applying the pre-selection cuts at the LHeC withan unpolarized electron beam.

10

FIG. 11. Kinematical distributions of some input observablesfor signal j e− χχ with m

χ±1 ,χ

02

= 250 GeV (black with filled

area) in the decoupled-slepton scenario, and for the SM back-ground of the j e− νν (red) and j e− `ν (blue) processes afterapplying the pre-selection cuts at the FCC-eh with an unpo-larized electron beam.

[1] CMS, A. M. Sirunyan et al., Searches for physics be-yond the standard model with the MT2 variable inhadronic final states with and without disappearingtracks in proton-proton collisions at

√s = 13 TeV,

(2019), arXiv:1909.03460, CMS-SUS-19-005, CERN-EP-2019-180.

[2] CMS, A. M. Sirunyan et al., Search for supersymmetryin proton-proton collisions at 13 TeV in final states withjets and missing transverse momentum, JHEP 10, 244(2019), arXiv:1908.04722, CMS-SUS-19-006, CERN-EP-2019-152.

[3] ATLAS, T. A. collaboration, Search for squarks andgluinos in final states with jets and missing transversemomentum using 139 fb−1 of

√s =13 TeV pp collision

data with the ATLAS detector, (2019), ATLAS-CONF-2019-040.

[4] ATLAS, M. Aaboud et al., Search for electroweakproduction of supersymmetric states in scenarios withcompressed mass spectra at

√s = 13 TeV with the

ATLAS detector, Phys. Rev. D97, 052010 (2018),arXiv:1712.08119, CERN-EP-2017-297.

[5] ATLAS, M. Aaboud et al., Search for electroweak pro-duction of supersymmetric particles in final states withtwo or three leptons at

√s = 13 TeV with the ATLAS de-

tector, Eur. Phys. J. C78, 995 (2018), arXiv:1803.02762,CERN-EP-2017-303.

[6] ATLAS, M. Aaboud et al., Search for chargino-neutralinoproduction using recursive jigsaw reconstruction in fi-nal states with two or three charged leptons in proton-proton collisions at

√s = 13 TeV with the ATLAS detec-

tor, Phys. Rev. D98, 092012 (2018), arXiv:1806.02293,CERN-EP-2018-113.

[7] ATLAS, M. Aaboud et al., Search for chargino and neu-tralino production in final states with a Higgs boson andmissing transverse momentum at

√s = 13 TeV with

the ATLAS detector, Phys. Rev. D100, 012006 (2019),arXiv:1812.09432, CERN-EP-2018-306.

[8] CMS, A. M. Sirunyan et al., Search for electroweak pro-duction of charginos and neutralinos in WH events inproton-proton collisions at

√s = 13 TeV, JHEP 11, 029

(2017), arXiv:1706.09933, CMS-SUS-16-043, CERN-EP-2017-113.

[9] CMS, A. M. Sirunyan et al., Search for electroweak pro-duction of charginos and neutralinos in multilepton fi-nal states in proton-proton collisions at

√s = 13 TeV,

JHEP 03, 166 (2018), arXiv:1709.05406, CMS-SUS-16-039, CERN-EP-2017-121.

[10] CMS, A. M. Sirunyan et al., Search for supersymme-try with Higgs boson to diphoton decays using the ra-zor variables at

√s = 13 TeV, Phys. Lett. B779, 166

(2018), arXiv:1709.00384, CMS-SUS-16-045, CERN-EP-2017-158.

[11] CMS, A. M. Sirunyan et al., Search for new phe-nomena in final states with two opposite-charge, same-flavor leptons, jets, and missing transverse momentum inpp collisions at

√s = 13 TeV, JHEP 03, 076 (2018),

arXiv:1709.08908, CMS-SUS-16-034, CERN-EP-2017-170.

[12] CMS, A. M. Sirunyan et al., Search for new physicsin events with two soft oppositely charged leptons andmissing transverse momentum in proton-proton colli-sions at

√s = 13 TeV, Phys. Lett. B782, 440 (2018),

arXiv:1801.01846, CERN-EP-2017-336, CMS-SUS-16-048.

[13] CMS, A. M. Sirunyan et al., Combined search forelectroweak production of charginos and neutralinos inproton-proton collisions at

√s = 13 TeV, JHEP 03, 160

(2018), arXiv:1801.03957, CMS-SUS-17-004, CERN-EP-2017-283.

[14] CMS, A. M. Sirunyan et al., Search for supersymmetric

11

partners of electrons and muons in proton-proton colli-sions at

√s = 13 TeV, Phys. Lett. B790, 140 (2019),

arXiv:1806.05264, CMS-SUS-17-009, CERN-EP-2018-132.

[15] CMS, A. M. Sirunyan et al., Search for supersymmetrywith a compressed mass spectrum in the vector bosonfusion topology with 1-lepton and 0-lepton final states inproton-proton collisions at

√s = 13 TeV, JHEP 08, 150

(2019), arXiv:1905.13059, CMS-SUS-17-007, CERN-EP-2019-093.

[16] ATLAS, T. A. collaboration, Search for electroweak pro-duction of charginos and sleptons decaying in final stateswith two leptons and missing transverse momentum in√s = 13 TeV pp collisions using the ATLAS detector,

(2019), ATLAS-CONF-2019-008.[17] ATLAS, T. A. collaboration, Searches for electroweak

production of supersymmetric particles with compressedmass spectra in

√s = 13 TeV pp collisions with the AT-

LAS detector, (2019), ATLAS-CONF-2019-014.[18] ATLAS, T. A. collaboration, Search for direct production

of electroweakinos in final states with missing transverseenergy and a Higgs boson decaying into photons in ppcollisions at

√s = 13 TeV with the ATLAS detector,

(2019), ATLAS-CONF-2019-019.[19] ATLAS, T. A. collaboration, Search for chargino-

neutralino production with mass splittings near the elec-troweak scale in three-lepton final states in

√s = 13 TeV

pp collisions with the ATLAS detector, (2019), ATLAS-CONF-2019-020.

[20] ATLAS, G. Aad et al., Search for direct production ofelectroweakinos in final states with one lepton, missingtransverse momentum and a Higgs boson decaying intotwo b-jets in (pp) collisions at

√s = 13 TeV with the

ATLAS detector, (2019), arXiv:1909.09226, CERN-EP-2019-188.

[21] ATLAS, G. Aad et al., Search for electroweak productionof charginos and sleptons decaying into final states withtwo leptons and missing transverse momentum in

√s =

13 TeV pp collisions using the ATLAS detector, (2019),arXiv:1908.08215, CERN-EP-2019-106.

[22] A. Barr and J. Scoville, A boost for the EW SUSY hunt:monojet-like search for compressed sleptons at LHC14with 100 fb−1, JHEP 04, 147 (2015), arXiv:1501.02511.

[23] Z. Han, G. D. Kribs, A. Martin, and A. Menon, Hunt-ing quasidegenerate Higgsinos, Phys. Rev. D89, 075007(2014), arXiv:1401.1235.

[24] H. Baer, A. Mustafayev, and X. Tata, Monojetplus soft dilepton signal from light higgsino pair pro-duction at LHC14, Phys. Rev. D90, 115007 (2014),arXiv:1409.7058, FTPI-MINN-14-28, UH-511-1237-14.

[25] B. Dutta et al., Probing Compressed Sleptons at theLHC using Vector Boson Fusion Processes, Phys. Rev.D91, 055025 (2015), arXiv:1411.6043, MIFPA-14-33.

[26] Z. Han and Y. Liu, MT2 to the Rescue – Searching forSleptons in Compressed Spectra at the LHC, Phys. Rev.D92, 015010 (2015), arXiv:1412.0618.

[27] C. Han, L. Wu, J. M. Yang, M. Zhang, andY. Zhang, New approach for detecting a compressedbino/wino at the LHC, Phys. Rev. D91, 055030 (2015),arXiv:1409.4533.

[28] A. G. Delannoy et al., Probing Dark Matter at the LHCusing Vector Boson Fusion Processes, Phys. Rev. Lett.111, 061801 (2013), arXiv:1304.7779.

[29] A. G. Delannoy et al., Probing Supersymmetric Dark

Matter and the Electroweak Sector using Vector Bo-son Fusion Processes: A Snowmass Whitepaper, (2013),arXiv:1308.0355.

[30] P. Schwaller and J. Zurita, Compressed elec-troweakino spectra at the LHC, JHEP 03, 060 (2014),arXiv:1312.7350, CERN-PH-TH-2013-319, MITP-13-084.

[31] S. Gori, S. Jung, and L.-T. Wang, Corneringelectroweakinos at the LHC, JHEP 10, 191 (2013),arXiv:1307.5952, EFI-13-15, ANL-HEP-PR-13-35,KIAS-P13040.

[32] C. Han et al., Probing Light Higgsinos in Natural SUSYfrom Monojet Signals at the LHC, JHEP 02, 049 (2014),arXiv:1310.4274.

[33] G. F. Giudice, T. Han, K. Wang, and L.-T. Wang,Nearly Degenerate Gauginos and Dark Matter at theLHC, Phys. Rev. D81, 115011 (2010), arXiv:1004.4902,CERN-PH-TH-2010-056, MADPH-10-1556, IPMU09-0102.

[34] G. H. Duan, K.-I. Hikasa, J. Ren, L. Wu, and J. M.Yang, Probing bino-wino coannihilation dark matter be-low the neutrino floor at the LHC, Phys. Rev. D98,015010 (2018), arXiv:1804.05238.

[35] M. Abdughani and L. Wu, On the coverage of neu-tralino dark matter in coannihilations at the upgradedLHC, (2019), arXiv:1908.11350.

[36] Working Group 3, X. Cid Vidal et al., Beyond theStandard Model Physics at the HL-LHC and HE-LHC,(2018), arXiv:1812.07831, CERN-LPCC-2018-05.

[37] C. Han, R. Li, R.-Q. Pan, and K. Wang, Searching forthe light Higgsinos in the MSSM at future e-p colliders,Phys. Rev. D98, 115003 (2018), arXiv:1802.03679.

[38] D. Curtin, K. Deshpande, O. Fischer, and J. Zu-rita, New Physics Opportunities for Long-Lived Parti-cles at Electron-Proton Colliders, JHEP 07, 024 (2018),arXiv:1712.07135, TTP17-053.

[39] A. Birkedal-Hansen and B. D. Nelson, Relic neutralinodensities and detection rates with nonuniversal gaug-ino masses, Phys. Rev. D67, 095006 (2003), arXiv:hep-ph/0211071, LBNL-51698, UCB-PTH-02-50, MCTP-02-53.

[40] H. Baer et al., Exploring the BWCA (bino-wino co-annihilation) scenario for neutralino dark matter, JHEP12, 011 (2005), arXiv:hep-ph/0511034, FSU-HEP-050817, BNL-HET-05-24, UH-511-1079-05.

[41] N. Arkani-Hamed, A. Delgado, and G. F. Giudice, TheWell-tempered neutralino, Nucl. Phys. B741, 108 (2006),arXiv:hep-ph/0601041, CERN-PH-TH-2005-260.

[42] K. Harigaya, K. Kaneta, and S. Matsumoto, Gaug-ino coannihilations, Phys. Rev. D89, 115021 (2014),arXiv:1403.0715, IPMU-14-0047.

[43] J. R. Ellis, T. Falk, and K. A. Olive, Neutralino - Staucoannihilation and the cosmological upper limit on themass of the lightest supersymmetric particle, Phys. Lett.B444, 367 (1998), arXiv:hep-ph/9810360, CERN-TH-98-326, UMN-TH-1725-98, MADPH-98-1087.

[44] M. Endo, K. Hamaguchi, S. Iwamoto, and K. Yanagi,Probing minimal SUSY scenarios in the light of muong − 2 and dark matter, JHEP 06, 031 (2017),arXiv:1704.05287, UT-17-14, IPMU17-0061, KEK-TH-1973.

[45] J. Alwall et al., The automated computation of tree-leveland next-to-leading order differential cross sections, andtheir matching to parton shower simulations, JHEP 07,

12

079 (2014), arXiv:1405.0301, CERN-PH-TH-2014-064,CP3-14-18, LPN14-066, MCNET-14-09, ZU-TH-14-14.

[46] J. Debove, B. Fuks, and M. Klasen, Threshold resum-mation for gaugino pair production at hadron colliders,Nucl. Phys. B842, 51 (2011), arXiv:1005.2909, IPHC-PHENO-10-02, LPSC-10-050.

[47] A. Djouadi, M. M. Muhlleitner, and M. Spira, Decaysof supersymmetric particles: The Program SUSY-HIT(SUspect-SdecaY-Hdecay-InTerface), Acta Phys. Polon.B38, 635 (2007), arXiv:hep-ph/0609292, CERN-PH-TH-2006-200.

[48] DELPHES 3, J. de Favereau et al., DELPHES 3, A mod-ular framework for fast simulation of a generic colliderexperiment, JHEP 02, 057 (2014), arXiv:1307.6346.

[49] LHeC Study Group, J. L. Abelleira Fernandez et al., ALarge Hadron Electron Collider at CERN: Report on thePhysics and Design Concepts for Machine and Detector,J. Phys. G39, 075001 (2012), arXiv:1206.2913, SLAC-R-999, CERN-OPEN-2012-015, LHEC-NOTE-2012-001-

GEN.[50] M. Klein, LHeC Detector Design, 25th In-

ternational Workshop on DIS, Birmingham (2017),https://indico.cern.ch/event/568360/contributions/2523637/.

[51] U. Klein, Higgs into heavy quarks, Work-shop on the LHeC, Chavannes (2014),https://indico.cern.ch/event/278903/contributions/631181/attachments/510303/704309/Chavannes UKLein 20.01.2014.pdf.

[52] Private communications with Max Klein, Uta Klein, Pe-ter Kostka, and Satoshi Kawaguchi.

[53] A. Hocker et al., TMVA - Toolkit for Multivariate DataAnalysis, (2007), arXiv:physics/0703039, CERN-OPEN-2007-007.

[54] ATLAS, Extrapolation of EmissT + jet search re-sults to an integrated luminosity of 300 fb−1 and3000 fb−1, (2018), ATL-PHYS-PUB-2018-043,https://cds.cern.ch/record/2650050.