Searches for C. Brust, AK, R. Sundrum, 126.2353 Z. Han, AK, M. Son, B. Tweedie, 121.XXXX Harvard University Frontiers Beyond the Standard Model III, Minneapolis, October 13, 212 (Harvard) October, 13 1 / 25
Outline (Harvard) October, 13 2 / 25
Why? SUSY was a good and motivated new physics candidate (hierarchy problem, unification). Now we know that there is no SUSY (in its simple version) beneath the TeV scale. How can we hide SUSY? Effective SUSY - light 3rd generation superpartners, heavy 1st and 2nd generation scalars. This scenario addresses the hierarchy problem of the SM and hides from the LHC. SM effective SUSY solves the little hierarchy up to 1 TeV 1 TeV EW scale solution of the large hierarchy problem (full SUSY, strong coupling...??) Adopt bottom-up approach to. We do not specify a particular mediation pattern and use a Natural SUSY as an effective theory with a cutoff 1 TeV (Harvard) October, 13 3 / 25
More on spectrum of Stops should not be too heavy to cancel the top-divergence in the Higgs mass. Perfect naturalness the average mass is 4 GeV. There is at least one sbottom (part of the LH multiplet) which is light m bl m tl Wino and Bino should not be too heavy m W 1 TeV, m B 3 TeV RH sbottom should be in the spectrum to cancel U(1) Y D-term divergence (1-3 TeV) Higgsinos should not be too heavy (tree-level fine tuning) Gluinos should cancel the leading divergence in the stop-sbottom mass. If gluinos are Majorana m g 2m t, but it can be heavier by factor of 2 or 3 if the gluinos are Dirac. All other particles can be of order 1 TeV, and we effectively integrate them out of the theory (Harvard) October, 13 4 / 25
1 1 1 theory Bounds on RPC [GeV] m χ 35 3 25 2 15 1 5 ~ ~ b 1 -b 1 ~, b 1 b+ χ CDF 2.65 fb D 5.2 fb 1 ~ b b χ 1 forbidden ATLAS s = 7 TeV 1 1 15 2 25 3 35 4 45 m~ [GeV] b CL s CL s CL s ± Observed Limit (95% C.L.) Expected Limit (95% C.L.) Expected Limit ±1 σ 1 σ NLO scale unc. L dt = 2.5 fb Reference point, Bounds on gluino are much stronger (multi-jets + /E T, SS dileptons, multi b-tag) [GeV] m χ 6 5 4 3 2 1 ~ g- ~ g, ~ g tt χ 1 CL S ATLAS 95% C.L. limits. σ SUSY -lepton, 6-9 jets ATLAS-CONF-2123 2-SS-leptons, 4 jets ATLAS-CONF-2125 3 b-jets arxiv:127.4686 g ~ ttχ forbidden 1 Preliminary [L int [L int [L int not included. = 5.8 fb, 8 TeV] = 5.8 fb, 8 TeV] = 4.7 fb, 7 TeV] Expected Observed Expected Observed Expected Observed 5 6 7 8 9 1 11 12 m~ [GeV] g Do these searches cover the entire parameter space of? (Harvard) October, 13 5 / 25
R parity: conserving or violating? We assumed that the cutoff of the theory is 1 TeV. Even if the R-parity is conserved, operators like QQQL will cause rapid proton decay. Therefore we should either consider UV completions which are supersymmetric with R-parity or assume that the proton is protected by some other symmetry (lepton or baryon number). In this talk we will consider RPV scenario. Gluinos As we will see gluinos are relatively easy to find, one can extend bounds on gluino up to O(TeV) even in RPV case. Focus first on heavy gluinos, beyond the reach of 8 TeV LHC. (Harvard) October, 13 6 / 25
Constraints on RPV SUSY RPV superpotential W RPV = µ H u L+λLLe c +λ LQd c +λ u c d c d c Proton decay problem either lepton or baryon number conservation. If lepton number is violated, SUSY typically leads to multi-lepton signatures and it is already constrained by the LHC searches. BNV W κ t R d c i dc j. Constraints n n, stringent constraints, but proportional to Majorana mass of gluino and very sensitive to tanβ and other parameters K K, forces κ 1 2...1 3 (Harvard) October, 13 7 / 25
Spectrum and decay modes The LSP always decays either into jets or into a top and jets. Neutralino is not necessarily an LSP, stop can easily be an LSP. Assume: A stop is LSP (mostly RH, w/ mixing), RPV is so small, that the resonant is tiny, decays proceed through RPV only if it is the only available decay. What decays should we look for? mass gluino > 1 TeV the heavy stop Z emission the lightest sbottom W emission the lightest stop Lacking part - Higgsinos. Decays of sbottom into Higgsino and stop are suppressed either by Y b or by phase space, therefore b W ( ) t 1 is not affected. BR( t 2 Z(h) t 1 ) can be reduced if Higgsino is lighter that t 2. j j (Harvard) October, 13 8 / 25
Light stop: search for dijet resonances Signature: two equal mass resonances decaying into two jets. Should look for 4-jets events reconstructing two equal-mass resonances (similar to existing coloron search). The existing bound on a BNV stop decaying into 2 jets is 82 GeV (LEP) Very light stops very hard for the LHC. For m t = 1 GeV we will never trigger on vast majority of signal events. σ( t t ) m=15 GeV 6 pb, σ( t t ) m=3 GeV 1 pb No exclusions yet after full 7 TeV run, becomes even harder since all the trigger thresholds went up at 8 TeV run. (Harvard) October, 13 9 / 25
b tw ( ) transitions Assume: the LH doublet soft mass is larger than m tr. The splitting inside the LH doublet mostly comes from F tr (SUSY top contribution). Spectrum t 2 b t 1. If m t1 < m H < m b, the transition Γ( b Hb) y b 2. b Ht transition is likely phase space suppressed, unless the mass splittings are not too large. We neglect all b decays through Higgsinos. Needed search: 2 relatively soft leptons with 4 or more jets which reconstruct 2 resonance with the same mass. Naturalness prefers modest mass splittings between the stop and the sbottom. (Harvard) October, 13 1 / 25
Search strategies and cuts Dominant background: dileptonic t t with two additional hard jets Cluster the event with anti-k T, clustering R =.7. Demand two isolated leptons, at least 4 jets with p T > 3 GeV. Try all possible pairings between the 4 leading jets and pick up the combination which minimizes the invariant mass difference between the pairs. Discard the event if the minimal mass difference exceeds 1 GeV. /E T > 35 GeV, S T > 4 GeV. What else distinguishes signal from background? P T of the leading lepton in signal and background samples miss E T distribution in signal and background Normalized rate.9.8.7.6.5 Normalized rate.1.8.6.4.3.4.2.2.1 t 1 2 4 6 8 1 12 14 16 18 2 p (l ) 5 1 15 2 25 miss E T (Harvard) October, 13 11 / 25
Search strategy - 2 p T (l) and /E T distributions look so different for reason. The W in the signal events is either off shell or just marginally on shell. We can use a cut on dimensionless variables r /ET = / E T S T, r l = p T(l 1 ) S T Ratio of the leading lepton to S T Ratio of missing E T to S T Normalized rate.8.7.6.5.4.3.2 Normalized rate.9.8.7.6.5.4.3.2.1.1 T 1 T.5.1.15.2.25.3 p (l )/S.5.1.15.2.25.3.35.4 miss E /S T T We use the cuts r /ET <.15 and r l <.15. (Harvard) October, 13 12 / 25
Results How does stop decay? Stop decays into two jets, coming from two different down-type quarks. Possible combinations are ds, db, sb. We do not know the flavor structure of the RPV operator. It might have the dominant coupling to the b-quark b-jets both in our signal and in our background. If it couples to ds, we can perform b-veto to reduce the background. Reference point: m b = 3 GeV, m t = 217 GeV Without b-veto: With b-veto: Dijet invariant mass Dijet invariant mass Events/1 GeV 8 7 6 Events/1 GeV 45 4 35 5 3 4 25 3 2 15 2 1 1 5 jj 5 1 15 2 25 3 35 4 45 M jj 5 1 15 2 25 3 35 4 45 M (Harvard) October, 13 13 / 25
14 12 1 8 6 4 2 12 1 8 6 4 2 Could this be discovered by cut-and-count? Here we used kinematics of the signal event to see it on top of t t background. Can we give up on this and simply use all our cuts for cut-and count search? Apply the same cuts, do not reconstruct two same-mass resonances and do not discard events with large minimal mass difference. Events/1 GeV distribution S T 4 5 6 7 8 9 1 11 12 S T Invariant mass of two leading jets in the event Events/1 GeV Maybe the invariant mass of two leading jets can help? 1 2 3 4 5 6 7 M jj (Harvard) October, 13 14 / 25
t 2 t 1 : strength and weakness of multileptons t 2 t 1 Z(h) ideal for multilepton searches, do not need to reconstruct the full resonance: Why it is not generic? ~ t This mode Y ~ + t and it is not phase space H suppressed. Unlike b W ( ) t 1 this mode is very sensitive to Higgsinos and we can b be left without multileptons. Analysis of Higgsino mode is challenging. (Harvard) October, 13 15 / 25
How does gluino event look like? Decay into the lightest stop: Naively looks like a t t event with four (or more) extra jets. Can we see these events in simple cut-and-count experiments? Can we we use more elaborate kinematic handles? The most straightforward cut-and-count technique one can suggest: SS tops in the same event SS dileptons in the same event. (Harvard) October, 13 16 / 25
Same-sign dileptons Gripaios and Allanach, 212 Big advantage of this approach: very distinctive, low-background signature. Since the backgrounds are small we can safely decrease the cut on /E T, the backgrounds are still under control. Why would we like to have other tools: This is a really powerful tool in RPC case, where we have 4 tops in the event, but in the RPV case there are only 2 tops, BR 2.5% This rate is not model independent, it can we further diminished if gluino masses are mostly Dirac Can we do even better? Can we use more abundant channels (e.g. semi-leptonic) and base our cut-and-count techniques on the number of jets and the kinematic of the events? (Harvard) October, 13 17 / 25
Cut-and-count in semileptonic channel Originally suggested by Lisanti, Shuster, Strassler, Toro, 211 What are the differences between a signal event and t t event: Expect more jets in average (naively 4 extra-jets) Expect much higher H T m g = 6 GeV, m t = 1/4 GeV m g = 8 GeV, m t = 1/6 GeV m g = 6/8 GeV, m t = 14/2 GeV m gluino = 6 GeV > 6 n jets normalized rate.45.4.35 normalized rate.45.4.35 normalized rate.2.18.16.3.3.14.25.25.12.2.15.2.15.1.8.6.1.1.4.5.5.2 2 4 6 8 1 12 2 4 6 8 1 12 5 1 15 2 25 N j N j H T (Harvard) October, 13 18 / 25
Cut and count experiment Distribution of H T in events with 7 or more jets: Stop masses 1 GeV and 4 GeV: Events, 2 fb m gluino = 6 GeV, n jets 9 8 7 > 6 Events, 2 fb Stop masses 1 GeV and 6 GeV: m gluino = 8 GeV, n jets 6 5 >6 6 4 5 4 3 3 2 2 1 1 5 1 15 2 25 H T 5 1 15 2 25 H T Gluinos up to 8 GeV are accessible with simple cut-and-count techniques, but we should try different cuts windows of H T. (Harvard) October, 13 19 / 25
Reconstruction of resonances: 2 regimes If m g m t we can easily reconstruct the entire event, because stops and tops in the event are boosted, significantly reducing combinatorial uncertainties. Realistic choice of parameters for the boosted case: m g = 6 GeV, m t = 1 GeV; m g = 8 GeV, m t = 2 GeV These mass splittings are big enough to remove combinatorial uncertainties, but still too small to merge jets and leptons still can apply the cut on more than 6 jets. If the mass difference is not so big, we cannot use boosted techniques, but there are still two equal-mass jet resonances that we can try to reconstruct. (Harvard) October, 13 2 / 25
Reconstruction of boosted events Use events which passed cut on H T, N 7 narrow jets and an isolated lepton; recluster with R = 1.5, C/A algorithm Run top-tagger on the fat jets, if a jet is identified as a top candidate (very loose internal cuts) do not consider it for the next step. If more than one candidate have been identified, choose the candidate with the closest mass to 173 GeV. Pick up the highest p T fat jet (which is not top candidate) and decluster it using BDRS procedure. Plot the invariant mass of the subjets inside this jet. m BDRS jj (ht > 11 && nj > 6) m BDRS jj (ht > 11 && nj > 6) ) # events (2 fb 1 8 ) # events (2 fb 1 8 6 6 4 4 2 2 5 1 15 2 25 3 35 4 BDRS mjj 5 1 15 2 25 3 35 4 BDRS mjj (Harvard) October, 13 21 / 25
Not boosted regime Basic observation. Assume that the stop is much heavier than the top. In this case it is likely that all four leading jets in the event are coming from the stops. We can start from the 4 jets with the highest p T and reconstruct two same-mass resonances. ) # events (2 fb mjj (ht > 11 && nj > 5) 1 8 6 ) # events (2 fb mjj (ht > 11 && nj > 5) 1 8 6 4 4 2 2 1 1 2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 mjj mjj This technique works well for m t = 6 GeV, but for 4 GeV the peak is less sharp. For m t = 25 GeV, the peak is completely erased. Should include lower p T jet if we are looking for lighter stops. (Harvard) October, 13 22 / 25
) ) ) ) Not boosted regime - modification Include more jets in the search for the same mass resonances. Try all possible pairing between 5 or 6 leading jets in the even, choose the pair with the minimal invariant mass difference: Leading 4j, m t = 4 GeV # events (2 fb mjj (ht > 11 && nj > 5) 1 8 6 4 Leading 5j, m t = 4 GeV # events (2 fb mjj (ht > 11 && nj > 5) 14 12 1 8 6 4 2 2 1 2 3 4 5 6 7 8 mjj 1 2 3 4 5 6 7 8 mjj mjj (ht > 13 && nj > 6) mjj (ht > 13 && nj > 6) Leading 4j, m t = 25 GeV # events (2 fb 4 35 3 25 2 15 Leading 6j, m t = 25 GeV # events (2 fb 9 8 7 6 5 4 3 1 2 5 1 1 2 3 4 5 6 7 8 mjj 1 2 3 4 5 6 7 8 mjj (Harvard) October, 13 23 / 25
Estimated reach and comparison with other techniques Current bounds Current bounds from SS dileptons are m g 5 GeV for sufficiently light stops (estimate of Gripaios and Allanach). In fact bounds from OS dileptons are slightly lower m g 5 GeV. The strongest bound comes from ATLAS BH searches, which almost excludes 6 GeV gluino. SS dileptons after 8 TeV run should have sensitivity up to 8 GeV (if the gluino mass is Majorana) OS dileptons should have slightly lower sensitivity BH - the best reach among the existing searches Our techniques - estimated reach up to m g 1.1TeV (Harvard) October, 13 24 / 25
1. RPV with natural SUSY is a well motivated scenario 2. Propose new search which should be sensitive to certain RPV spectra, including those with very heavy gluinos. The cut-and-count search is unlikely to be conclusive in this case, but bump reconstruction should be unambiguous. 3. If gluinos are below TeV, even cut-and-count experiment in monoleptonic channel should have a good sensitivity for gluinos in RPV scenario. This search should not necessarily have a cut on /E T. 4. Cut and count combined with peak reconstruction should tell us whether there are gluinos below 1.1 TeV scale even in RPV case. (Harvard) October, 13 25 / 25