Photons in 900 Ge V collision data and

  • Slides: 17
Download presentation
Photons in 900 Ge. V collision data and plans for 7 Te. V Leonardo

Photons in 900 Ge. V collision data and plans for 7 Te. V Leonardo Carminati 02/02/2010 Atlas Italia 1

Photons in 900 Ge. V minimum bias data q 384186 events in Min. Bias

Photons in 900 Ge. V minimum bias data q 384186 events in Min. Bias COLLCAND stream, egamma GRL (solenoid ON, flags GREEN) q f r 141000+ and events 100000+ and dq LBSUMM#Det. Status. LBSUMMDecember 09 -00 pix, sct, lar, trtb, trte, til g and dq LBSUMM#Det. Status. LBSUMMDecember 09 -00 atlsol, atlgl, l 1 ctp g and lhc beamenergy 449 -451 and lhc stablebeams true q In the following a 2. 5 Ge. V PT cut is applied and barrel-endcap crack removed (1. 37<|h|<1. 52) 02/02/2010 Atlas Italia 2

Photon candidates in 900 Ge. V Min. Bias data q Prompt and fast analysis

Photon candidates in 900 Ge. V Min. Bias data q Prompt and fast analysis with an important contributions from Milano 02/02/2010 Atlas Italia 3

Photons in 900 Ge. V minimum bias q Mainly photons from p 0 decay

Photons in 900 Ge. V minimum bias q Mainly photons from p 0 decay : q 68. 5% true photons from pi 0 decay q 14. 1% truth-matched, but not photon q 13. 8% other true photons q 0. 66% true photons from q/g radiation q 3. 0% not matched q One should keep in mind the low energy regime of the candidates we are looking at 02/02/2010 Atlas Italia 4

(I) data/MC shower shapes comparison Photon identification is based on a set of discriminating

(I) data/MC shower shapes comparison Photon identification is based on a set of discriminating variables q Energy leakage in HAD calorimeter q Shower shapes in the second sampling of EM calorimeter q Shower shapes in the first sampling of EM calorimeter ATLAS preliminary 02/02/2010 ATLAS preliminary Atlas Italia 5

(I) data/MC shower shapes comparison Example of discriminating variables in the strips q Fair

(I) data/MC shower shapes comparison Example of discriminating variables in the strips q Fair agreement between data and montecarlo q Shape of the data is not well reproduced by MC. 2 hypothesis under test (work in progress): q Cross talk effect not well modeled in the montecarlo q Material in front not well modeled 02/02/2010 Atlas Italia 6

Experience from the analysis on 900 Ge. V real data : analysis q Loose

Experience from the analysis on 900 Ge. V real data : analysis q Loose photons : cuts on hadronic leakage and second sampling shapes q Tight photons : stricter cuts on the same loose variables + additional cuts on calo strips shower shapes. Efficiencies wrt to candidates in the photon container in fair agreement now looking into fine effects (cross-talk, material descriptions) 02/02/2010 Atlas Italia 7

(II) Search for p 0 and h 0 in 900 Ge. V data q

(II) Search for p 0 and h 0 in 900 Ge. V data q SW clusters are not suitable to search for low p. T photons (2. 5 Ge. V threshold) : q use Em. Topo. Clusters to have a first evidence Work in progress q Notice that the EMTopo. Clusters are calibrated at the EM scale only 02/02/2010 Atlas Italia 8

(II) Search for p 0 and h 0 in 900 Ge. V data Work

(II) Search for p 0 and h 0 in 900 Ge. V data Work in progress 02/02/2010 Atlas Italia 9

(II) Search for p 0 and h 0 in 900 Ge. V data q

(II) Search for p 0 and h 0 in 900 Ge. V data q Can we do something better ? q Calibrate EMTopo. CLusters with appropriate calibration coefficients q Use the EMTopo. Cluster as a seed to build a SW cluster and apply standard calibration Work in progress q The standard athena calibration seems to work much better than expected in such a low PT regime (it was optimized down to 10 Ge. V only) 02/02/2010 Atlas Italia 10

(III) Conversions reconstruction Conversion reconstruction : silicon – silicon 2 tracks conversions q Data

(III) Conversions reconstruction Conversion reconstruction : silicon – silicon 2 tracks conversions q Data – MC comparison (normalized) : shapes in nice agreement q Use the beam pipe to normalize the MC q Compare the number of reconstructed conversions in the data q Good to within a factor of 2 (barrel) due to limited stat 02/02/2010 Atlas Italia 11

Experience from the analysis on 900 Ge. V real data and readiness for 7

Experience from the analysis on 900 Ge. V real data and readiness for 7 Te. V q Distributed analysis worked nicely as well as DDM tools q Not as nicely in all sites: efficiency of italian sites will become critical in the 1 fb -1 scenario q Use of the group space tokens on italian T 2 and subscriptions successfully exercised q Data quality selection worked easily: database query exercised and GRLs used for the analysis at different levels q Good. Run. List. Selector tool tested in the stream preparation, before the athena based analysis and in ROOT. Detailed cross checks oing q Egamma performance groups GRL used up to now q Data selection tools and strategies more and more important : q Did not check yet the various d. ESD : probably d. ESD out of the COLLCAND stream would help q Need to exercise deeply the selection on TAGs (not done yet!) q Is it up to the physics group to prepare specific streams and eventually prepare group ntuples? 02/02/2010 Atlas Italia 12

Plans for 7 Te. V data: q We expect a sizeable number of prompt

Plans for 7 Te. V data: q We expect a sizeable number of prompt photons in 7 Te. V data : ds/dp. T for inclusive photon and diphoton production Di-photon production Inclusive photon production 02/02/2010 Atlas Italia Inclusive photon production 13

What on the photon side with the first data? q A measurement of the

What on the photon side with the first data? q A measurement of the inclusive photon cross section with the first 100 pb -1 or less is realistic, provided that we could achieve a good control of all the ingredients: q Required ingredients for the analysis: q Understand/optimize trigger efficiency q Understand selection efficiency / optimization of ID cuts q Understand kinematic cuts acceptances (pythia, jetphox, resbos/diphox ? ): q Understand background contamination : several methods to estimate photon purity/fake rate estimation in close connection with performance groups. 25/08/2009 Egamma workshop 2009 (Leysin) 14

Preparation for first analysis : efficiencies q Trigger efficiency : reasonable to assume trigger

Preparation for first analysis : efficiencies q Trigger efficiency : reasonable to assume trigger efficiency for g 20_loose wrt offline is 100% for early data. Systematics arising from this are likely at the % level and are expected to have little impact on early measurements q Need probably to keep the discussion alive on what should be done if the rate is too high: raise threshold vs tighten the cuts q In situ estimation of photon efficiency is probably a > 200 pb -1 task: not an easy and clean source of photons q Need to rely on MC prediction for quite some time. The assumption is that the MC tuned on electrons in situ measurement will be good for photons also (believe in the e -> MC extrapolation) q Infer photon efficiency from Z ee events applying photon cuts on electrons: photon efficiency estimated believing in the electron to photon MC extrapolation q Select a pure sample of photon in Z ee and Z events. Need a quite ‘large’ statistics, ~ two hundred of useful photons in ~200 pb -1 25/08/2009 Egamma workshop 2009 (Leysin) 15

Preparation for first analysis : fake rate estimation Fake rate understanding will be the

Preparation for first analysis : fake rate estimation Fake rate understanding will be the most challenging activity at the beginning: several techniques are being developed in close connection with egamma performance group. q Simple technique at the very beginning, 2 dimensional side-bands: use 2 uncorrelated discriminating variables and estimate the background contamination in the signal region q Choose a separation variable having pdfs sensibly different for signal and background (conversion probability, energy in the presampler…) q Need to validate the shapes of the sensible variables with very first data : isolation, strips variables, conversions related variables, presampler energy. q Strategies to extract pdf for signal and background from data exercised on MC : create signal and background enriched samples playing with the isolation 25/08/2009 Egamma workshop 2009 (Leysin) 16

Conclusions q Photons analyses in the first data are interesting : comparison with MC

Conclusions q Photons analyses in the first data are interesting : comparison with MC is a key point for any serious physics study. 2 CONF notes in preparation q Shapes in good agreement : now going in the details q Studies on p 0 mass are ongoing q Milano group is involved in both CONF notes in preparation q Plans for 7 Te. V analysis are defined : need to address the question of how to handle such a big amount of data we expect in ~2 years (1 fb -1? ) especially on the computing and storage resources side 02/02/2010 Atlas Italia 17