SUSY: new search channels and new search techniques Maurizio Pierini 1 Wednesday, November 9, 11
Disclaimer • I was asked to talk about new searches, so I will not cover classic approaches • I will focus on hadronic searches, which I know better • I will not show results. There are specific talks for that • The talk is CMS-centric, because I am biased and because results based on “new” approaches mainly come from CMS Wednesday, November 9, 11
Outline • The lesson from Tevatron: the “classic” approach • α T: rejecting QCD • M T2 : characterizing signal as two-missing-particles signature • The Razor: merging the two in a consistent framework • A few considerations thunking at 2012 3 Wednesday, November 9, 11
A “classic” SUSY search The typical signature: a lot of energy seen in the detector, recoiling against a lot of MET Several variables to quantify this behavior: 4 Wednesday, November 9, 11
A “classic” SUSY search -1 CMS Preliminary, L = 1.1 fb , s = 7 TeV Events / 100 GeV Events / 100 GeV Events / 100 GeV Events / 100 GeV Data Bkg. expectation from MC 3 3 3 3 10 10 10 10 W(l )+Jets ν Z( )+Jets ν ν t t +Jets A counting QCD 2 2 2 2 10 10 10 10 Susy LM4 experiment is 10 10 10 10 performed on the tail of the 1 1 1 1 distribution -1 -1 -1 -1 10 10 10 10 0 0 0 0 500 500 500 500 1000 1000 1000 1000 1500 1500 1500 1500 2000 2000 2000 2000 2500 2500 2500 2500 3000 3000 3000 3000 3500 3500 3500 3500 H H H H (GeV) (GeV) (GeV) (GeV) T T T T CMS Preliminary CMS Preliminary 700 (GeV) -1 ~ ~ L = 1.1 fb , s = 7 TeV CDF g , q , tan =5, <0 β µ int ~ ~ Observed D0 , , g q tan =3, <0 β µ ∼ P ± Expected 1 600 ± σ LEP2 χ S An exclusion limit is 1/2 1 L ∼ ~ τ ± m LEP2 l tan β =10, µ >0, A =0 -1 0 CMS 1.1 fb α T set on some NP 500 Observed 2010 ~ q (1000)GeV parameter space 400 ~ g (1000)GeV ~ q (750)GeV 300 ~ g (750)GeV ~ q (500)GeV 200 ~ g (500)GeV 0 200 400 600 800 1000 1200 1400 1600 1800 5 m (GeV) 0 Wednesday, November 9, 11
Backgrounds To Fight mismeasured jet QCD with fake MET related to pathological events require understanding of rare detector-related effects _ Fake MET ν mismeasured jet MET SM processes with real MET, e.g. Z( νν )+jets measurable from control samples defined on data ν 6 Wednesday, November 9, 11
The New Ways • The “classic” approach is still pursued by CMS and ATLAS, adapted to the new detectors • New approaches proposed to reduce the QCD to negligible level and deal with the residual SM background through data-driven control samples • Different layers of extra assumptions give different signal vs. background separation - α T: unbalanced events - M T2 : MET coming from two particles - RAZOR variables: pair production of heavy objects producing two missing particles 7 Wednesday, November 9, 11
α T : Rejecting QCD α ≡ p T 2 . α T = E Tjet 2 E Tjet 2 m jj ⌘ 2 , = r⇣ M T ⌘ 2 ⌘ 2 ⇣ i = 1 p jet i ⇣ i = 1 p jet i ∑ 2 i = 1 E Tjet i ∑ 2 ∑ 2 − − x y Randall & Tucker-Smith CMS Preliminary 2011 - α T = 0.5 for perfectly balanced dijet events Events / 0.025 - α T<0.5 for dijet + mismeasurements ∫ ∫ 4 -1 -1 10 L dt = 1.1 fb L dt = 1.1 fb , , s s = 7 TeV = 7 TeV Data - EW main bkg after α T cut Standard Model 3 10 QCD MultiJet t t , W, Z + Jets - QCD events could leak to α T>0.5 because of LM4 2 10 LM6 detector effects (rare) 10 - large fraction of signal events removed (efficiency vs purity) 1 -1 10 -1 CMS Preliminary 2011 1.1 fb s = 7 TeV 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 α T counts / bin 3 10 Data (hadronic sample) SM (QCD + EWK) - After α T cut the signal looks similar to EWK (t t + W + Z ) → ν ν Z → ν ν 2 LM6 (LO) 10 bkg in α T - another variable needs to be used to 10 characterize the signal 1 - Back to the “classic” paradigm”: HT used by CMS -1 10 300 400 500 600 700 800 900 8 H (GeV) T Wednesday, November 9, 11
α T : BKG Estimate • EW bkg is estimated using the R α T (*) ratio ratio R α T = N α T > θ / N α T < θ exhibits • This is computed scaling the p T of the jets with the HT threshold, to event of the ratio in all H bins topology • The ratio is found to be compatible with the flat hypothesis within the available data and SM MC statistics -3 -3 × 10 × 10 0.1 0.1 T T α α R ∫ -1 R ∫ -1 CMS preliminary 2011 L dt = 1.1 fb , s = 7 TeV CMS simulation 2011 L dt = 0.4 fb , s = 7 TeV 0.09 0.09 0.08 0.08 SM (nominal) SM+LM4 W (+15%) SM+LM6 0.07 0.07 W (-15%) SM Data Z → ν ν (+15%) 0.06 0.06 Z → ν ν (-15%) t t (+15%) 0.05 0.05 t t (-15%) t (+15%) 0.04 0.04 t (-15%) 0.03 0.03 0.02 0.02 0.01 0.01 0 0 400 600 800 400 600 800 H (GeV) H (GeV) T T • This is used to predict the bkg expected in each bin of HT. Then a fit to the HT shape is used (*) Number of EW events with α T> θ / number of QCD events with α T< θ 9 Wednesday, November 9, 11
M T2 : two missing particles { χ + 1 → χ 0 1 π + . • pp → - We are looking for events with - χ + 1 → χ 0 1 π + . ~ two undetected neutral particles leaving the detector • We measure the sum of their pT ~ as MET • This is similar to the detection of the W, for which the edge of the mT distribution is used 3 CMS 10 × events / 2.5 GeV -1 36 pb at s = 7 TeV 2.5 • W + 1 jet → µ ν The presence of two missing 2 particles make the picture more data 1.5 W → µ ν complicated. But the physics non-top top intuition holds 1 0.5 0 50 100 150 10 M [GeV] T Wednesday, November 9, 11
M T2 : two missing particles • If we could see all the particles, we could compute χ 0 χ 0 � � m 2 1 = m 2 π + m 2 1 + 2 E π T E T cosh( ∆ η ) − p π 1 1 T · p χ + χ 0 T • If we could measure p T ( Χ 0 ), but not p z ( Χ 0 ), the best we could do would be χ 0 χ 0 χ 0 m 2 1 ) ≡ m 2 π + + m 2 T ( p π T , p T ; m χ 0 1 + 2( E π T E T ) ( 1 T − p π 1 1 T · p χ 0 • Since cosh>1, m T ≤ m, the equality holding for both pz( Χ 0 )=0. This means that max(m T ) has an “edge” at m • For each event we have two values of m T (two copies of the same decay). Both are such that m T <m. This means that max(m T (1), m T (2))<m • We only know p T ( Χ 01 )+ p T ( Χ 02 )=E Tmiss . A wrong assignment of the missing momenta brakes the m T <m condition. But the condition would hold for the correct assignment. This means that min(m T )<m T (true)<m. • This defined m T2 as T ( p π (1) q (1) T ( p π (2) q (2) � � �� m 2 m 2 T ; χ ) , m 2 T 2 ( χ ) min max , / , / T ; χ ) . ≡ T T / T + / T = / q (1) q (2) p T 11 Wednesday, November 9, 11
M T2 : two missing particles • The variable we have is a function m[ π ] m T4 ee of the mass of the LSP m T3 e π m T2 ππ • SUSY characterization: + ] - m[ χ 1 0 ] m[ χ 1 - Scan the LSP mass and look for the edge developing in your sample 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 of SUSY events (if you have one...) 0 ]) - m[ χ 1 0 ] / GeV m TX (m[ χ 1 Figure 3: Simulations of m TX ( m χ 0 1 ) − m χ 0 1 for X = 2 , 3 , 4 using a • SUSY search: simple phase-space Monte-Carlo generator program for a pair of q → χ + 1 q followed by χ + 1 π or χ + 1 → χ 0 1 → χ 0 decays ˜ 1 e ν e . As the number of invisible particles increases, the proportion of events - Assume a mass value (eg mLSP=0) near the upper limit decreases. Within the figure, subscripts are indicated by square brackets. - Assume that the visible system in has 0 mass - An analytical expression for M T2 is found ( M T 2 ) 2 = 2 A T = 2 p vis ( 1 ) p vis ( 2 ) ( 1 + cos φ 12 ) , T T - The edge is lost but we have an α T -like variable to kill the QCD 12 Wednesday, November 9, 11
M T2 : two missing particles -1 High M Analysis CMS Preliminary, s = 7 TeV, L = 1.1 fb T2 • QCD Events 5 10 M T2 is found to be useful for W+jets Z+jets 4 10 searches, since it allows to Top 3 LM6 10 data reduce QCD to negligible 2 10 level 10 1 • Signal is searched on the tail -1 10 of M T2 in a counting 0 200 400 600 M experiment T2 • -1 High M Analysis CMS Preliminary, s = 7 TeV, L = 1.1 fb T2 Other variables could be used QCD Events 15 W+jets to characterize the signal, in Z+jets Top case of a discovery. CMS Other 10 LM5 x 1 would use √ s min for that data 5 √ q q M 2 vis + P 2 M 2 / 2 s min ( M miss , min ) = T , vis + miss , min + E T 0 1000 2000 3000 4000 s min √ 13 Wednesday, November 9, 11
The Razor Frame • Two squarks decaying to quark and LSP . In their rest frames, they are two copies of the same monochromatic decay. In this y frame p(q) measures M Δ M 2 q − M 2 ˜ ˜ χ = 2 M ˜ M ∆ ≡ χ γ ∆ β ∆ , M ˜ q x • In the rest frame of the two incoming partons, the y two squarks recoil one against each other. • In the lab frame, the two squarks are boosted longitudinally. The LSPs x escape detection and the quarks are If we could see the LSPs, we could detected as two jets boost back by β L , β T , and β CM → β T In this frame, we would then get y |p j1 | = |p j2 | → β L Too many missing degrees of 14 freedom to do just this z Wednesday, November 9, 11
Recommend
More recommend
