aa r X i v : . [ h e p - e x ] J un A General Search for New Phenomena at HERA
E. SauvanOn behalf of the H1 CollaborationCPPM, IN2P3-CNRS et Universit´e de la M´editerran´ee163 Avenue de Luminy F-13288 Marseille, FranceA model-independent search for deviations from the Standard Model prediction is per-formed in e + p and e − p collisions at HERA II using all high energy data recorded bythe H1 experiment. This corresponds to a total integrated luminosity of 337 pb − . Allevent topologies involving isolated electrons, photons, muons, neutrinos and jets withhigh transverse momenta are investigated in a single analysis. Events are assigned toexclusive classes according to their final state. A statistical algorithm is used to searchfor deviations from the Standard Model in distributions of the scalar sum of transversemomenta or invariant mass of final state particles and to quantify their significance. Agood agreement with the Standard Model prediction is observed in most of the eventclasses. The most siginificant deviation is found in the µ - j - ν channel in e + p collisions. At HERA electrons a and protons collide at a centre-of-mass energy of up to 319 GeV. Thesehigh-energy electron-proton interactions provide a testing ground for the Standard Model(SM) complementary to e + e − and pp scattering. The approach presented here consists of acomprehensive and generic search for deviations from the SM prediction at large transversemomenta. The present analysis follows closely the strategy of the previous publication fromthe H1 experiment [2]. All high P T final state configurations involving electrons ( e ), muons( µ ), jets ( j ), photons ( γ ) or neutrinos ( ν ) are systematically investigated. The completeHERA II data sample (2003–2007) is used, corresponding to a total integrated luminosityof 337 pb − shared between e + p (178 pb − ) and e − p (159 pb − ) collisions. All final states containing at least two objects ( e , µ , j , γ , ν ) with P T >
20 GeV in the polarangle range 10 ◦ < θ < ◦ are investigated. All selected events are classified into exclusiveevent classes according to the number and types of objects detected in the final state (e.g. e - j , µ - j - ν , j - j - j - j - j ). The criteria used in the identification of each type of particle arechosen to ensure an unambiguous identification, while retaining high efficiencies [2]. Allexperimentally accessible combinations of objects have been studied and data events arefound in 23 event classes.A precise and reliable estimate of all relevant processes present at high transverse mo-mentum in ep interactions is needed to ensure an unbiased comparison to the SM. Henceseveral Monte Carlo generators are used to generate a large number of events in all eventclasses, carefully avoiding double-counting of processes. The simulation contains the order α S matrix elements for QCD processes, while second order α matrix elements are used tocalculate QED processes. Additional jets are modelled using leading logarithmic partonshowers as a representation of higher order QCD radiation. a In this paper “electrons” refers to both electrons and positrons, unless otherwise stated.
DIS 2007 he event yields observed in each event class are presented and compared to the SMexpectation in figures 1(a) and (b) for e + p and e − p collisions, respectively. In each class, agood description of the number of observed data events by the SM prediction is seen. Thisdemonstrates the good understanding of the detector response and of the SM processes inthe considered phase space. Distributions of the scalar sum of transverse momenta P P T ofall objects are presented in figure 2 for e + p data. j-je-j-j m n j- n e-e-e m e- m - m g j- g e-j-j-je-j-j n j-j- n e-j- n -j- m g j-j- g e-j-e-j-j-j n j-j-j-j-j-j-j Events - -
10 1 10 j-je-j-j m n j- n e-e-e m e- m - m g j- g e-j-j-je-j-j n j-j- n e-j- n -j- m g j-j- g e-j-e-j-j-j n j-j-j-j-j-j-j Events - -
10 1 10 j-je-j-j m n j- n e-e-e m e- m - m g j- g e-j-j-je-j-j n j-j- n e-j- n -j- m g j-j- g e-j-e-j-j-j n j-j-j-j-j-j-j Events - -
10 1 10 SMH1 Data (prelim.) -2 -1
10 1 10 -1 p (178 pb + H1 General Search, HERA II e (a) j-je-j-j m n j- n e-e-e m e- m - m g j- g e-j-j-je-j-j n j-j-e-e-je-e-e n e-j- n -j- m g e-j- g - n j-e-j-j-j n j-j-j-j-j-j-j n j-j-j-j- Events - -
10 1 10 j-je-j-j m n j- n e-e-e m e- m - m g j- g e-j-j-je-j-j n j-j-e-e-je-e-e n e-j- n -j- m g e-j- g - n j-e-j-j-j n j-j-j-j-j-j-j n j-j-j-j- Events - -
10 1 10 j-je-j-j m n j- n e-e-e m e- m - m g j- g e-j-j-je-j-j n j-j-e-e-je-e-e n e-j- n -j- m g e-j- g - n j-e-j-j-j n j-j-j-j-j-j-j n j-j-j-j- Events - -
10 1 10 SMH1 Data (prelim.) -2 -1
10 1 10 -1 p (159 pb - H1 General Search, HERA II e (b)
Figure 1: The data and the SM expectation for all event classes with observed data eventsor a SM expectation greater than one event. The results are presented separately for e + p (a) and e − p (b) collisions. In order to quantify the level of agreement between the data and the SM expectation andto identify regions of possible deviations, the same search algorithm as developed in [2] isused. All possible regions in the histograms of P P T and M all distributions are considered.The number of data events ( N obs ), the SM expectation ( N SM ) and its total systematicuncertainty ( δN SM ) are calculated for each region. A statistical estimator p is defined tojudge which region is of most interest. This estimator is derived from the convolution of thePoisson probability density function (pdf) to account for statistical errors with a Gaussianpdf, G ( b ; N SM , δN SM ), with mean N SM and width δN SM , to include the effect of nonnegligible systematic uncertainties [2]. The value of p gives an estimate of the probability ofa fluctuation of the SM expectation upwards (downwards) to at least (at most) the observednumber of data events in the region considered. The region of greatest interest (of greatestdeviation) is the region having the smallest p -value, p min .The possibility that a fluctuation with a value p min occurs anywhere in the distribution DIS 2007 -1 -2 -1 j - j -1 -1 e - j -1 -1 - j m n j - -1 -1 n e - -1 -1 e - e -1 -1 m e - -1 -1 m - m -1 -1 g j - -1 -1 g e - -1 -1 j - j - j -1 -1 e - j - j n j - j - -2 -1 -2 -1 n e - j - -2 -1 -2 -1 n - j - m -2 -1 -2 -1 g j - j - -1 -1 g e - j - -1 -1 e - j - j - j -1 -1 n j - j - j - -2 -1 -2 -1 j - j - j - j Distributions T P S ) - -1 p (178 pb + H1 General Search, HERA II e (GeV) T P S E ve n t s Region of greatest deviationSMH1 Data (prelim.)
Figure 2: Distributions of P P T for classes with at least one data event, for e + p data. Theshaded areas show the regions of greatest deviation chosen by the search algorithm. DIS 2007 s estimated. This is achieved by creating hypothetical data histograms following the pdfsof the SM expectation. The algorithm is then run on those hypothetical histograms to findthe region of greatest deviation and the corresponding p SM min is calculated. The probabilityˆ P is then defined as the fraction of hypothetical data histograms with a p SM min equal to orsmaller than the p min value obtained from the real data. ˆ P is a measure of the statisticalsignificance of the deviation observed in the data. The event class of most interest for asearch is thus the one with the smallest ˆ P value. Depending on the final state, a p min -valueof 5 . · − (“5 σ ”) corresponds to a value of − log ˆ P , the negative decade logarithm of ˆ P ,between 5 and 6. The overall degree of agreement with the SM can further be quantifiedby taking into account the large number of event classes studied in this analysis. Amongall studied classes there is some chance that small ˆ P values occur. This probability can becalculated with MC experiments. A MC experiment is defined as a set of hypothetical datahistograms following the SM expectation with an integrated luminosity equal to the amountof data recorded. The complete search algorithm and statistical analysis are applied to theMC experiments analgously as to the data. This procedure is repeated many times. Theexpectation for the ˆ P values observed in the data is then given by the distribution of ˆ P SM values obtained from all MC experiments. The probability to find in the MC experimentsa ˆ P value smaller than in the data can be calculated and gives us the global significance ofthe observed deviation. P -log N u m b e r o f E ve n t C l asses -1 P -log N u m b e r o f E ve n t C l asses -1 H1 Data (prelim.)MC Experiments ) -1 p (178 pb + H1 General Search, HERA II e Scan T P S (a) P -log N u m b e r o f E ve n t C l asses -1 P -log N u m b e r o f E ve n t C l asses -1 H1 Data (prelim.)MC Experiments ) -1 p (159 pb - H1 General Search, HERA II e Scan T P S (b) Figure 3: The − log ˆ P values for the data event classes and the expected distribution fromMC experiments as derived by investigating the P P T distributions in e + p (a) and e − p (b)data.The ˆ P values observed in the real data in all event classes b are compared in figure 3 tothe distribution of ˆ P SM obtained from the large set of MC experiments, normalised to oneexperiment. The comparison is presented for the scans of the P P T distributions. All ˆ P val-ues range from 0 .
01 to 0 .
99, corresponding to event classes where no significant discrepancybetween data and the SM expectation is observed. These results are in agreement with theexpectation from MC experiments. The most significant deviation from SM predictions isobserved in the µ - j - ν event class and in e + p collisions with a value of − log ˆ P equal to 1 . e + p collisions, the b Due to the uncertainties of the SM prediction in the j - j - j - j and j - j - j - j - ν event classes at highest M all and P P T (see [2]), where data events are observed, no reliable ˆ P values can be calculated for these classes.These event classes are not considered to search for deviations from the SM in this extreme kinematicdomain. DIS 2007 argest deviation was also found in this event class, with − log ˆ P = 3. All the data collected with the H1 experiment during HERA II running period (2003–2007)have been investigated for deviations from the SM prediction at high transverse momentum.All event topologies involving isolated electrons, photons, muons, neutrinos and jets areinvestigated in a single analysis. A good agreement between data and SM expectation isfound in most event classes. In each event, class the invariant mass and sum of transversemomenta distributions of particles have been systematically searched for deviations usinga statistical algorithm. No significant deviation is observed in the phase-space and in theevent topologies covered by this analysis. The largest deviation from SM expectation isobserved in the µ - j - ν event class in e + p collisions. References [1] Slides: http://indico.cern.ch/contributionDisplay.py?contribId=127&sessionId=9&confId=9499 [2] A. Aktas et al. [H1 Collaboration], Phys. Lett. B (2004) 14 [hep-ex/0408044].(2004) 14 [hep-ex/0408044].