360 likes | 373 Views
Explore methods to boost the discovery potential of the Higgs boson and uncover physics beyond the Standard Model in hadron collider data from 2005. Featuring strategies to enhance Higgs decay mode identification and data mining approaches.
E N D
Finding the Higgs or something else ideas to improve the discovery potential at hadron colliders Sascha Caron Freiburg Seminar, Sept. 2005
The situation in 2005 • We still don’t know the origin of EW symmetry breaking • The Higgs boson is not discovered yet • Even with the SM Higgs: • ‘fine tuning’ is required in the model to remain valid to high energies?, • Gravity is not included?, Fermion masses? • typical solutions by increasing the number of • symmetries, dimensions, forces, … Higgs ? Something else? Sascha Caron page 1
The situation in 2005 Investigate if EW symmetry breaking is caused by the Higgs. Investigate if there is other physics beyond the Standard Model Part 1 Increase Higgs finding capabilities in ‘most likely‘ SM decay modes/channels (I chose H -> bb) Part 2 Data mining strategies How to find something potentially interesting and previously unexpected in the data? Sascha Caron page 2
Outline • Part 1 Some ideas to improve H->bb • Part 2 Is there something else? Sascha Caron page 3
Part 1 Some ideas to improve H->bb
The quest for H->bb_bar • Background for Higgs->bb is in some channels so high that • even triggering becomes difficult: • B-triggering at DZero • b-jet identification important for early Higgs discovery • How can we further improve the b-identification? • Study b-jets using top events at ATLAS Sascha Caron page 5
B trigger at DØ Events per second QCD ET>30 GeV dijet production >10 Find b-events early to keep high efficiency at an acceptable rate 0.1 b-jets ET>30 GeV 0.01 Z-> b bbar Goals Z->bb, HZ->bbvv, H->bb, etc. maybe B physics Higgs->b bbar ZH-> bbvv, bH->bbb etc. Sascha Caron page 6
The Silicon Track Trigger at D0 DØ in Run II The Silicon Track Trigger is based on information of the : Silicon Microstrip Tracker Central Fiber Tracker Sascha Caron page 7
The Silicon Track Trigger at D0 Trigger System p p bunch crossing frequency ¯ L1 Trigger decision time about 4 μs 2000 Hz L2 Trigger decision time about 200 μs 1000Hz L3 Trigger decision time about 50 ms 2.5MHz 50 Hz • Hardware based • tracks made with • central fiber tracker, calorimeter towers, muons • Hardware/Software • simple jets, electrons, muons, taus • Silicon Microvertex improved tracks (STT) • L2 global processor combines information • (e.g. STT tracks for very fast B-id) • Software based • partial event • reconstruction • (also simple B-id) Sascha Caron page 8
The Silicon Track Trigger at D0 Principal Idea B decay products B decay length is mm Interaction point is mean beam spot Impact parameter (2d in x-y plane) • Silicon Improved Tracks with • 2d impact parameter • Select events with large impact parameter tracks Sascha Caron page 9
The Silicon Track Trigger at D0 How can the tracking be improved? Silicon detector Fiber Tracker • Tracks found at L1 with the Central Fiber Tracker are used to define roads into the Silicon • Silicon hits are clustered • Track is re-fit within the road • (IP, χ2) within about 50 µs IP resolution ≈ 50 μm Old idea : select event by a cut on IP Sascha Caron page 10
A fast B-id algorithm for Level 2 New Idea: Combine tracks in a fast, multivariate algorithm • Derive probability density functions • of tracks in B-events : PB • and non-B events : Pnon-B • Store their ratio into a lookup table on the L2 global processor Probability ratio PB/Pnon-B ONLINE ALGORITHM Loop over the 5 ‘good’ tracks with largest IP and derive the product : P B,i/ P non-B,i Sascha Caron page 11
A fast B-id algorithm for Level 2 Derive performance of the STT+B-id algorithm with D0 data B-id algorithm Data with offline b-tags Events Signal efficiency Cut method Data without offline b-tag Discriminator of the B-id algorithm Background efficiency Sascha Caron page 12
The quest for H->bb_bar • Silicon Track Trigger at DZero works • Further improvement by up to a factor 2 • with the B-id algorithm • Impact in next Higgs trigger strategy for difficult channels Next step: Have we learned something for ATLAS/CMS? Sascha Caron page 13
Improving B-id at ATLAS/CMS Can we further improve the b-jet identification? Yes, by using b-jets from data and not from MC to make b-id algorithms Correctly assigned jets Idea: Select clean sample of b-jets from data We know which jet is the b-jet from top kinematics in the background free and large tt sample at ATLAS Combinatorical background Mqqb (GeV) Sascha Caron page 14
Improving B-id at ATLAS/CMS W=Two jets with highest momentum in reconstructed jjj C.M. frame. Correctly assigned jets -> We know the b-jet ! |Mqq-MW|<10 GeV Use this side to get b-jet (3-jets with highest vector summed pt) M qqb (GeV) Expected purity >70% without doing kinematic fit or anything sophisticated Use this side to get a completely clean sample … many ideas how to improve this … Sascha Caron page 15
Improving B-id at ATLAS/CMS … get all b-jet info from data … Old Idea: - Derive b-efficiency using this b-jet New Idea: - Important to derive PB and Pnon-B distributions using b-jets in different samples and to use data information for tagging Can we reproduce this? PB (MC b-jets) ALTAS b-tagging: P B,i/ P non-B,i Pnon-B(MC u-jets) Tracks i in the jet Sascha Caron page 16
Part 2 Is there something else ?
The situation in 2005 What do we expect to find at the LHC? One physicist's schematic view of particle physics in the 21st century (Courtesy of Hitoshi Murayama) Sascha Caron page 17
The situation in 2005 MSSM CMSSM MN2SSM (2 Q’s with Mirror particles In addition) SUSY with extra Dim Or SUSY with extra forces Or …. SUSY VERSIONSOF THE SM NMSSM (+ an additional Higgs singlet) Choose this point, look at the LHC data, exclude or not! Sascha Caron page 18
We found no deviation • We have excluded this point/area which is epsilon of the parameter space We found a deviation Does this mean that the ‘real’ modelis this parameter point? Is it efficient to work like this?
Finding the unexpected – explaining the origin • New Strategy: START FROM THE DATA • Search for deviations in all final states(they are all interesting either as signal or to understand background) • Determine the regions of ‘greatest deviation’ • Determine the origin of these deviations Is this possible? YES, IT HAS BEEN DONE ! General Search for new Phenomena at H1 Sascha Caron page 19
H1 General Search • Event yields for HERA 1 • data • (all channels with SM exp. • > 0.01 event) • Good agreement for • (almost) all channels Channels which have not been syst. studied before Sascha Caron page 19
General Search I spend some time at the New Phenomena web pages at LHC experiments A count of final states planned to be studied leads to 100-500 However consider permutations of j,b,e,µ,τ,v,γ, + consider e.g. charge? Up to 8 particle final states lead to about 40000 Did you have events with 2 photons , a jet and a muon at your LEP exp.? Sascha Caron page 19
H1 General Search Search for deviations • Need to explore automated data analysis strategies • Idea to completely automate a search (DØ Sleuth analysis) • H1 General Search : Search for deviations between data and SM prediction in 1 dim. distributions most sensitive to new physics Very simple and remarkably powerful Sascha Caron page 20
H1 General Search • Check all connected regions with a size ≥ resolution in a histogram, i.e. calculate the probability p that data agrees with the SM • Region of greatest interest is the one with the smallest p Investigate Mall and ΣPT distributions for each channel Sascha Caron page 21
H1 General Search • Check all connected regions with a size ≥ resolution in a histogram, i.e. calculate the probability p that data agrees with the SM • Region of greatest interest is the one with the smallest p Investigate Mall and ΣPT distributions for each channel Sascha Caron page 21
Investigate all Mall and ΣPT distributions Sascha Caron page 22
Wait - What is the SM? “SM” = State of the art MCs + δ theory (pdf, scale, model) + δ data (jet energy scale, etc.) At the beginning of data taking
Wait - What is the SM? “SM” = State of the art MCs + δ theory (pdf, scale, model) + δ data (jet energy scale, etc.) … a factor of 10 in luminosity later Derive uncertainties and MC tuning from data by looking at various final states
A significant danger is finding correlations and signals that do not really exist. Many examples in particle physics history We are looking for deviations … How surprised should we be to find some? How likely is a 4-5 sigma deviation at LHC even if there is nothing in the data? Unsolvable problem if you use 2000 PhD students Sascha Caron page 24
Quantify the deviations Step 1: Repeat the whole analysis with a pseudo data experiment (dice your own MC data) many times. Step 2: Count how many times you find deviations bigger than in those in your real data. 3% Number of channels 3% of the “Pseudo H1 experiments” have found a bigger deviation 1 10-1 10-2 Probability to find deviation in this channel I know that this is not a new idea, but we do not use it Sascha Caron page 25
Going the way into the other direction… An General analysis of LHC data
Summary I’ve tried to illustrate some ideas to improve the discovery potential at LHC and the Tevatron. Improving the Higgs discovery potential by an improved Trigger and B-id A General Search for new phenomena strategy for the LHC
H1 General Search Search for deviations • Check all connected regions with a size ≥ resolution in a histogram, i.e. calculate the probability p that data agrees with the SM Search for deviations between data and SM prediction in 1 dim. Distributions (Mall and ΣPT)