Longitudinal Vector Boson Scattering with Deep Machine Learning - PowerPoint PPT Presentation

Longitudinal Vector Boson Scattering with Deep Machine Learning Jake Searcy, Lillian Huang, Marc-Andre Pleier, Junjie Zhu 1

VBS and the Higgs Without a Higgs the matrix element for Longitudinal VBS (VLVL) grows with energy until it becomes strongly coupled A Higgs fixes this The Questions: Does our Higgs fix this? How can we find out? 2

Extra useful info Longitudinal Scattering http://www.sciencedirect. com/science/article/pii/0550321385 900380 Longitudinal bosons grow with M(WW) if there is no Higgs Fraction of Longitudinal W Bosons Higgs-126 GeV Higgs-less No Higgs With Higgs 3

Event Signature ● Look in Same Sign W ∓ W ∓ ● Two same sign leptons ● Two jets ○ high M(j,j) ○ high Δn(j,j) Phys. Rev. Lett. 113, 141803 4

Same Sign W ∓ W ∓ Di-leptonic ● Aim: Longitudinal scattering ● Pros ○ Easy to find ○ Signal is clean ● Cons Phys. Rev. Lett. 113, 141803 ○ Two neutrinos ● Other options ○ OS WWjj-All top ○ Semileptonic WWjj-not yet sensitive to SM ■ ZZjj,WZjj-Very few events ● What can we do with 2 neutrinos Phys. Rev. Lett. 114, 051801 ● Study parton level truth for 5 W ∓ W ∓

How do we do it? Some Proposals K. Doroba, J. Kalinowski, J. Kuczmarski, S. Pokorski, J. Rosiek, M. Szleper, S. Tkaczyk Use specially built R pT variable R pT = p T (Lep1)p T (Lep2) / p T (Jet1)p T (Jet2) http://xxx.lanl.gov/pdf/1201.2768v2 A. Freitas, J. S. Gainer Use Matrix Element Analysis to differentiate different Higgs models http://xxx.lanl.gov/pdf/1212.3598v2.pdf We want to try and measure the longitudinal fraction directly 6

Measuring VLVL ● We’ve seen the first signs of VBS in W + W + ○ Next step is to see V L V L ■ Then can we measure V L V L at high M(W,W)? ● Effect of polarization is on the θ* distribution e + θ* Boost to W rest frame W + Direction ν e 7

Cos(θ*) distributions - 1D http://arxiv.org/pdf/1203.2165v2.pdf ● Fits give polarization fractions ● Of course can’t do this in real events because of the two missing neutrinos ● Do we have any sensitivity with measurable quantities? 8

Machine Learning Neural Networks Really common in HEP to use multivariate techniques classification (discret estimation) Signalness . . . . . . ... . . . Squash output between (0,1) Event Hidden Layers Outputs Inputs Just a simple f(x i ) ➝ Output Train weights so this mapping gives you the best discriminate between signal and background 9

Machine Learning Neural Networks You can also train NN to approximate continuous functions (Regression) Signalness . . . . . . ... . . . Squash outputbetween (0,1) Event Hidden Layers Outputs Inputs Don’t squash output 10

Machine Learning Neural Networks You can also train NN to approximate continuous functions (Regression) My favorite truth value . . . . . . My other favorite truth ... . . . value Event Hidden Layers Outputs Inputs Just a simple f(x i ) ➝ Output Train weights so this mapping gives you the best approximation of the function you want (minimum error 2 ) 11

Goal 1. Take measurable quantities in Same Sign W ∓ W ∓ , (pt,eta,phi, leptons and jets + met) 2. Train Neural Network to output the two true values of Cos(θ*) (one of each W) 3. Fit Neural Networks Cos(θ*) approximation to measure Longitudinal Fractions Run Neural Select Signal Subtract Fit Data Network Events Background 12

Training the Neural Network: Deep Learning ● Deep learning is simply extending a simple neural network with many layers ○ Conceptually simple, computationally a little difficult ● Has had a lot of success in recent days ○ In HEP and elsewhere ○ P.Baldi, P. Sadowski, D. Whiteson http://arxiv.org/pdf/1410.3469.pdf “The deep networks trained on the low-level variables performed better than shallow networks trained on the high-level variables engineered by physicists, and almost as well as the deep network trained high-level variables,” ● Since we have only one “High-level variable” this is exactly what we want. 13

Train Neural Network ● Use Deep Learning ○ Network with 20 Layers 200 Nodes ○ Instead of shallow one-layer ■ Though Results look pretty good with 1 layer ■ Gain ~20% with Deep learning ● Validate on independent data ○ Far from perfect, but certainly usable 14

1 ab -1 Fit Fit Neural Network ● 6 templates ○ ++,--,+-.LL,+L,-L ● Combine into 3 ○ Transverse-Transverse ○ Transverse-Longitudinal ○ Longitudinal-Longitudinal 15

Other Fit option Also possible to do all six at once 16

Real Life ● Have to separate Signal and Background ● Get a little help from the NN ● Still need to make event level cuts ● Finite detector resolution ● ATLAS CUTS ○ MET > 30 GeV ○ M(j,j) > 500 ○ Lepton pT > 25 GeV GeV ○ dY(j,j) > 2.4 ○ Jet pT > 30 GeV 17

Sensitivity ATLAS Cuts + Delphes Delphes Simulation ATLAS Cuts Parton Level 18

Some Comparisons Compare against variable from Doroba et al. R pT = p T (Lep1)p T (Lep2)/p T (jet1)p T (jet2) Linear Scale Log Scale 19

Fit Comparison Precision at 3ab -1 68% Parton with Cuts NN: 7.5 +2.6 -2.4 % -7.5 % RpT: 7.5 +11.3 NN has > 4x the sensitivity to the LL fraction. (Equivalent to 16x the data or hundreds of years of running the LHC!) 20

Future Studies ● A few nice properties of this regression that could be explored more ● What differential measurements possible? ● Train on Delphes samples ● Can be combined with cuts to enhance LL fraction (i.e. Doroba et. al.) 21

Conclusions ● Regression is a great tool for pulling out “hidden” information ○ Many applications beyond this one ● Measuring the properties of VBS possible in the same sign WW state ○ Neural Networks can more than double the sensitivity of the current state of the art methods. ● Limits can be made as early as the first measurements. ● Strong bounds will require the High Lumi LHC ● Differential distributions could be possible ● Better understanding (generation) of the polarization distributions will help ● Extra Thanks to ○ Olivier Mattelaer ○ Sally Dawson 22