Igor Volobouev Texas Tech University

Slides:



Advertisements
Similar presentations
From Quark to Jet: A Beautiful Journey Lecture 1 1 iCSC2014, Tyler Dorland, DESY From Quark to Jet: A Beautiful Journey Lecture 1 Beauty Physics, Tracking,
Advertisements

Charged Particle Jet measurements with the ALICE Experiment in pp collisions at the LHC Sidharth Kumar Prasad Wayne State University, USA for the ALICE.
Human-Computer Interaction Human-Computer Interaction Segmentation Hanyang University Jong-Il Park.
Jet and Jet Shapes in CMS
Michel Lefebvre, 2007/03/02Optimal Jet Finder1 Optimal Jet Finder work in progress Physics and Astronomy University of Victoria British Columbia, Canada.
Simulation of Z->jets in CMS Outline –Introduction –Technique –Results –Conclusion.
2007/03/20Optimal Jet Finder1 Physics and Astronomy University of Victoria British Columbia, Canada Damir Lelas Rolf Seuster Michel Lefebvre, Rob McPherson,
Introduction to Hadronic Final State Reconstruction in Collider Experiments Introduction to Hadronic Final State Reconstruction in Collider Experiments.
DØ Run II jet algorithms E. Busato (LPNHE, Paris) TeV4LHC Workshop 12/1/2004 Outline:  Introduction  The Ideal Jet Algorithm  DØ Run II Cone Jet Algorithm.
Jet Reconstruction in ATLAS Slide 1 Peter Loch University of Arizona Tucson, Arizona Peter Loch University of Arizona Peter Loch University of Arizona.
1 Hadronic In-Situ Calibration of the ATLAS Detector N. Davidson The University of Melbourne.
8.882 LHC Physics Experimental Methods and Measurements Jet Energy Scale [Lecture 23, May 04, 2009]
Peter Loch University of Arizona Tucson, Arizona USA
Jet finding Algorithms at Tevatron B.Andrieu (LPNHE, Paris) On behalf of the collaboration Outline: Introduction The Ideal Jet Algorithm Cone Jet Algorithms:
Introduction to Hadronic Final State Reconstruction in Collider Experiments Introduction to Hadronic Final State Reconstruction in Collider Experiments.
Frederik Rühr, KIP Heidelberg ATLAS Jets: Measurements, Calibration and Studies Frederik Rühr, Kirchhoff-Institut für Physik, Heidelberg Universität Heidelberg.
Introduction to Hadronic Final State Reconstruction in Collider Experiments Introduction to Hadronic Final State Reconstruction in Collider Experiments.
Introduction to Hadronic Final State Reconstruction in Collider Experiments Introduction to Hadronic Final State Reconstruction in Collider Experiments.
Alán Dávila for the STAR Collaboration WWND February, 8, 2011.
Tau Jet Identification in Charged Higgs Search Monoranjan Guchait TIFR, Mumbai India-CMS collaboration meeting th March,2009 University of Delhi.
Jet Studies at CMS and ATLAS 1 Konstantinos Kousouris Fermilab Moriond QCD and High Energy Interactions Wednesday, 18 March 2009 (on behalf of the CMS.
A Graph-based Friend Recommendation System Using Genetic Algorithm
EECS 274 Computer Vision Segmentation by Clustering II.
Studies of the jet fragmentation in p+p collisions in STAR Elena Bruna Yale University STAR Collaboration meeting, June
Jets at CMS Fedor Ratnikov, University of Maryland MIT, August 1, 2008.
August 30, 2006 CAT physics meeting Calibration of b-tagging at Tevatron 1. A Secondary Vertex Tagger 2. Primary and secondary vertex reconstruction 3.
1 P. Loch U of Arizona July 22, 2011 Experimental aspects of jet reconstruction and jet physics at the LHC (Part II) What Are Jets? The experimentalist’s.
CaloTopoCluster Based Energy Flow and the Local Hadron Calibration Mark Hodgkinson June 2009 Hadronic Calibration Workshop.
JETS 1 Konstantinos Kousouris Fermilab USCMS JTERM III.
Application of multi-resolution method on flow measurement in CBM Vojtech Petracek GSI – CBM week.
The Restricted Matched Filter for Distributed Detection Charles Sestok and Alan Oppenheim MIT DARPA SensIT PI Meeting Jan. 16, 2002.
Jet Physics at CDF Sally Seidel University of New Mexico APS’99 24 March 1999.
July 7, 2008SLAC Annual Program ReviewPage 1 New Approaches to Hadronic Final State Reconstruction D. W. Miller SLAC ATLAS.
QCD Multijet Study at CMS Outline  Motivation  Definition of various multi-jet variables  Tevatron results  Detector effects  Energy and Position.
A Comparison Between Different Jet Algorithms for top mass Reconstruction Chris Tevlin University of Manchester (Supervisor - Mike Seymour) Atlas UK top.
CALOR April Algorithms for the DØ Calorimeter Sophie Trincaz-Duvoid LPNHE – PARIS VI for the DØ collaboration  Calorimeter short description.
From Quark to Jet: A Beautiful Journey Lecture 2 1 iCSC2014, Tyler Dorland, DESY From Quark to Jet: A Beautiful Journey Lecture 2 Jet Clustering, Classification,
April 5, 2003Gregory A. Davis1 Jet Cross Sections From DØ Run II American Physical Society Division of Particles and Fields Philadelphia, PA April 5, 2003.
24/08/2009 LOMONOSOV09, MSU, Moscow 1 Study of jet transverse structure with CMS experiment at 10 TeV Natalia Ilina (ITEP, Moscow) for the CMS collaboration.
David Berge – CAT Physics Meeting – 9 May Summary Hadronic Calibration Workshop 3 day workshop 14 to 16 March 2008 in Tucson, Arizona
Machine Vision Edge Detection Techniques ENT 273 Lecture 6 Hema C.R.
TeV muons: from data handling to new physics phenomena Vladimir Palichik JINR, Dubna NEC’2009 Varna, September 07-14, 2009.
Régis Lefèvre (LPC Clermont-Ferrand - France)ATLAS Physics Workshop - Lund - September 2001 In situ jet energy calibration General considerations The different.
July 27, 2002CMS Heavy Ions Bolek Wyslouch1 Heavy Ion Physics with the CMS Experiment at the Large Hadron Collider Bolek Wyslouch MIT for the CMS Collaboration.
First Measurement of Jets and Missing Transverse Energy with the ATLAS Calorimeter at and David W. Miller on behalf of the ATLAS Collaboration 13 May 2010.
V. Pozdnyakov Direct photon and photon-jet measurement capability of the ATLAS experiment at the LHC Valery Pozdnyakov (JINR, Dubna) on behalf of the HI.
Motion tracking TEAM D, Project 11: Laura Gui - Timisoara Calin Garboni - Timisoara Peter Horvath - Szeged Peter Kovacs - Debrecen.
Moriond 2001Jets at the TeVatron1 QCD: Approaching True Precision or, Latest Jet Results from the TeVatron Experimental Details SubJets and Event Quantities.
20/05/09EMCal meeting The Seedless Infrared Safe Cone algorithm (SISCone) Swensy Jangal, M. Estienne.
1 Jet Reconstruction and Energy Scale Determination in ATLAS Ariel Schwartzman 3 rd Top Workshop: from the Tevatron to ATLAS Grenoble, 23-Oct-2008.
KIT High Pt Jet Studies with CMS On behalf of the CMS Collaboration Andreas Oehler University of Karlsruhe (KIT) DIS 2009 XVII International Workshop on.
Jet Energy Scale and Calibration Framework
Performance of jets algorithms in ATLAS
T2 Jet Reconstruction Studies
Mean Shift Segmentation
Igor Volobouev Texas Tech University
Data Analysis in Particle Physics
Jeremy Bolton, PhD Assistant Teaching Professor
Individual Particle Reconstruction
Inclusive Jet Cross Section Measurement at CDF
Simulation study for Forward Calorimeter in LHC-ALICE experiment
SM-like Higgs searches
Jet/Photon/Hadron Correlations at RHIC-PHENIX
Plans for checking hadronic energy
Lecture 2: Edge detection
Peter Loch University of Arizona Tucson, Arizona USA
Argonne National Laboratory
Lecture 2: Edge detection
Peter Loch University of Arizona Tucson, Arizona USA
Measurement of b-jet Shapes at CDF
Presentation transcript:

Igor Volobouev Texas Tech University i.volobouev@ttu.edu Multiresolution Jet Reconstruction with FFTJet Calor2010, IHEP, Beijing, China May 11 2010 Igor Volobouev Texas Tech University i.volobouev@ttu.edu

Iterative Cone and Mean Shift The iterative cone algorithm we all know is called ”Mean Shift” clustering in the pattern recognition literature. The procedure was first described in a 1975 paper by K. Fukunaga and L.D. Hostetler. In 1977, Sterman and Weinberg proposed jet definition based on cones Important generalization of Mean Shift: Y. Cheng, “Mean Shift, Mode Seeking, and Clustering”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol 17, Aug 1995, p. 790. Independently, S.D. Ellis invented the concept of “Snowmass potential” in 2002 (?) Fast seedless cone algorithms: FSLC (2006), SISCone (2007) Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Connecting Mean Shift and KDE Cheng (and Ellis) have shown that, mathematically, the problem of searching for stable cones is equivalent to locating the peaks of the energy density built in the η-φ space using kernel density estimation (KDE) with the “Epanechnikov” kernel function. That is, one convolutes with and finds the peaks K Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Problems with the Cone Algorithms and the Split-Merge Procedure Inherent configuration ambiguity The area of the jet is not fixed (this is a problem, e.g., for SISCone) The energy of two nearby jets is miscalibrated because standard jet energy calibration procedures employ well-isolated jets, and the combined energy fraction leaked out-of-cone is different for nearby jets Energy correlations are not handled The goal of assigning each tower to just one particular jet is unphysical because of an irreducible spatial energy smearing during shower development in the calorimeter Seeded cone algorithms (CDF, D0) have additional problems Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Jet Reconstruction as Spatial Filtering Convolute the energy deposition picture (as a function of η and φ) with a low-pass filter. This can be done quickly by DFFT. Computational complexity is O(N log N). Find the intensity peaks. These are the precluster locations. Apply a threshold to the peak magnitudes Assign a cluster membership function (MF) to each surviving precluster and to the pileup/noise Distribute calorimeter towers among jets and pileup/noise with weights generated by the MFs (fuzzy clustering) or assign each tower to the jet with the highest MF value (crisp clustering) Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Seedless Cone Algorithm as a Spatial Filter Compare with Gaussian filtering Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Multiresolution Jet Reconstruction with FFTJet FFTJet is a recently developed framework for building global, efficient, collinear and infrared safe jet algorithms using the spatial filtering approach The method is described in arXiv:0907.0270v1.The code and the manual are at http://projects.hepforge.org/fftjet/ FFTJet allows its users to take into account the detector properties (magnetic field, noise, etc) inside jet clustering and energy recombination procedures. Note that this goes against the “conventional wisdom” of keeping jet algorithms identical between theory and experiment. Magnetic field in the detectors and shower development inside the calorimeters are important effects! Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Jets in the 3.8 T Magnetic Field η φ pT = 10 GeV/c pT = 20 GeV/c Average angular energy profiles from the Pythia jet gun (light quark) Distributions are built 2 m away from the jet origin A CMS dijet event Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

pT Dependence of the Jet Shape CMS HCAL Barrel (simulated) Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Reconstructing Jets: Better Jet Shape Model Wins Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Multiresolution Jet Reconstruction Jets can be reconstructed using many angular resolution scales. Technically, using the cone algorithm with many values of R is not the best way to do this (the split-merge stage is a problem). Gaussian filtering works much better. FFTJet builds the “clustering tree”: preclusters formed at different resolution scales are related to each other by parent-daughter relationships. Using the parent and the “closest daughter”, precluster characteristics can be determined as a function of the resolution scale. Studies of precluster behavior in the scale space result in Advanced pattern recognition capabilities Optimal determination of jet properties Control over bifurcation points Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Example Clustering Tree Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Pattern Recognition in Scale Space Multiresolution analysis opens new ways of thinking about jet reconstruction: Choose an optimal resolution scale, best according to some criterion. Similar to “normal” analysis but with essentially continuous resolution choices. Choose optimal resolution scale for each jet (perhaps, using scale space “blob detection” techniques) Choose a configuration with a certain number of jets. Assess the stability of the configuration in the scale space. Look for jet substructure expected in the signal (e.g., for boosted heavy particles) Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Multiresolution Jet Reconstruction with FFTJet Bifurcation Points There are no “completely IR-safe” algorithms The field of science which studies significant topological structure changes in response to small changes in parameters or initial conditions is known as “Bifurcation Theory” We want to be as far away as possible from a bifurcation point when the jet 4-momentum is determined: jets near bifurcations are noisy! The clustering tree provides a good handle for controlling bifurcations Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Jet Energy Determination The most precise jet energy reconstruction will take into account a variety of jet characteristics in addition to the “one-shot” 4-momentum: Electromagnetic energy fraction Charged energy fraction Jet shape Jet flavor Detector noise and pileup Separation from other jets FFTJet “jet membership functions” is the mechanism by which such information can be included directly into the jet clustering and energy recombination process A powerful jet shape model: detector-level fragmentation function Igor Volobouev Multiresolution Jet Reconstruction with FFTJet

Multiresolution Jet Reconstruction with FFTJet Conclusions Better pattern recognition improves jet reconstruction on many levels Spatial filtering and multiresolution image analysis are powerful tools, already studied by mathematicians and computer vision scientists for many years. These tools should be incorporated into the next-generation jet reconstruction algorithms for Tevatron and LHC. Multiresolution approaches shine when pattern recognition becomes nontrivial: for signal identification in multijet processes, or when jet-like objects are not really jets (boosted tops or Ws) and the jet substructure analysis becomes necessary With these techniques, jet energy determination can be optimized on process-by-process and jet-by-jet basis I am far from knowing where this road will lead us, just making a few steps out the door… Igor Volobouev Multiresolution Jet Reconstruction with FFTJet