1 Iterative dynamically stabilized (IDS) method of data unfolding (*) (*arXiv:0907.3791) Bogdan MALAESCU CERN PHYSTAT 2011 Workshop on unfolding.

Slides:



Advertisements
Similar presentations
HARP Anselmo Cervera Villanueva University of Geneva (Switzerland) K2K Neutrino CH Meeting Neuchâtel, June 21-22, 2004.
Advertisements

Low x workshop Helsinki 2007 Joël Feltesse 1 Inclusive F 2 at low x and F L measurement at HERA Joël Feltesse Desy/Hamburg/Saclay On behalf of the H1 and.
1 Data Analysis II Beate Heinemann UC Berkeley and Lawrence Berkeley National Laboratory Hadron Collider Physics Summer School, Fermilab, August 2008.
Use of Kalman filters in time and frequency analysis John Davis 1st May 2011.
Sampling: Final and Initial Sample Size Determination
1 Approximated tracking of multiple non-rigid objects using adaptive quantization and resampling techniques. J. M. Sotoca 1, F.J. Ferri 1, J. Gutierrez.
N.D.GagunashviliUniversity of Akureyri, Iceland Pearson´s χ 2 Test Modifications for Comparison of Unweighted and Weighted Histograms and Two Weighted.
RooUnfold unfolding framework and algorithms
Introduction to Hadronic Final State Reconstruction in Collider Experiments Introduction to Hadronic Final State Reconstruction in Collider Experiments.
MINOS Feb Antineutrino running Pedro Ochoa Caltech.
A Search for Point Sources of High Energy Neutrinos with AMANDA-B10 Scott Young, for the AMANDA collaboration UC-Irvine PhD Thesis:
Statistical Image Modelling and Particle Physics Comments on talk by D.M. Titterington Glen Cowan RHUL Physics PHYSTAT05 Glen Cowan Royal Holloway, University.
2015/6/23 1 How to Extrapolate a Neutrino Spectrum to a Far Detector Alfons Weber (Oxford/RAL) NF International Scoping Study, RAL 27 th April 2006.
1 Tau Workshop, Nara, Sept 14-17, 2004 M. Davier – ALEPH  Results ALEPH Results on  Branching Ratios and Spectral Functions Michel Davier Laboratoire.
Statistical Treatment of Data Significant Figures : number of digits know with certainty + the first in doubt. Rounding off: use the same number of significant.
Definitions In statistics, a hypothesis is a claim or statement about a property of a population. A hypothesis test is a standard procedure for testing.
TOPLHCWG. Introduction The ATLAS+CMS combination of single-top production cross-section measurements in the t channel was performed using the BLUE (Best.
Resonances in decay for 400fb -1 DC Meeting April 10th, 2006 J.Brodzicka, H.Palka INP Kraków J.Brodzicka, H.Palka INP Kraków B +  D 0 D 0 K + B +  D.
The paired sample experiment The paired t test. Frequently one is interested in comparing the effects of two treatments (drugs, etc…) on a response variable.
Distinguishability of Hypotheses S.Bityukov (IHEP,Protvino; INR RAS, Moscow) N.Krasnikov (INR RAS, Moscow ) December 1, 2003 ACAT’2003 KEK, Japan S.Bityukov.
A. Blondel, M.Campanelli, M.Fechner Energy measurement in quasi-elastics Unfolding detector and physics effects Alain Blondel Mario Campanelli Maximilien.
Today’s lesson Confidence intervals for the expected value of a random variable. Determining the sample size needed to have a specified probability of.
 Candidate events are selected by reconstructing a D, called a tag, in several hadronic modes  Then we reconstruct the semileptonic decay in the system.
Unfolding jet multiplicity and leading jet p T spectra in jet production in association with W and Z Bosons Christos Lazaridis University of Wisconsin-Madison.
Christof Roland / MITSQM 2004September 2004 Christof Roland / MIT For the NA49 Collaboration Strange Quark Matter 2004 Capetown, South Africa Event-by-Event.
RooUnfold unfolding framework and algorithms Tim Adye Rutherford Appleton Laboratory ATLAS RAL Physics Meeting 20 th May 2008.
Point Source Search with 2007 & 2008 data Claudio Bogazzi AWG videconference 03 / 09 / 2010.
Walid DRIDI, CEA/Saclay n_TOF Collaboration Meeting, Paris December 4-5, 2006 DAPNIA Neutron capture cross section of 234 U Walid DRIDI CEA/Saclay for.
RooUnfold unfolding framework and algorithms Tim Adye Rutherford Appleton Laboratory Oxford ATLAS Group Meeting 13 th May 2008.
Relative Values. Statistical Terms n Mean:  the average of the data  sensitive to outlying data n Median:  the middle of the data  not sensitive to.
1 Bunch length measurement with the luminous region : status B. VIAUD, C. O’Grady B. VIAUD, C. O’Grady One problem in some data collections One problem.
Steve Geer IDS Meeting CERN March Neutral Currents and Tests of 3-neutrino Unitarity in Long-Baseline Exeriments Steve Geer Barger, Geer, Whisnant,
Study of neutrino oscillations with ANTARES J. Brunner.
Study of neutrino oscillations with ANTARES J. Brunner.
P. Vahle, Oxford Jan F/N Ratio and the Effect of Systematics on the 1e20 POT CC Analysis J. Thomas, P. Vahle University College London Feburary.
On quasi-two-body components of (for 250fb -1 ) (for 250fb -1 ) J.Brodzicka, H.Palka INP Krakow DC Meeting May 16, 2005 B +  D 0 D 0 K + B +  D 0 D 0.
Unfolding in ALICE Jan Fiete Grosse-Oetringhaus, CERN for the ALICE collaboration PHYSTAT 2011 CERN, January 2011.
JET CHARGE AT LHC Scott Lieberman New Jersey Institute of Technology TUNL REU 2013 Duke High Energy Physics Group Working with: Professor Ayana Arce Dave.
Study of pair-produced doubly charged Higgs bosons with a four muon final state at the CMS detector (CMS NOTE 2006/081, Authors : T.Rommerskirchen and.
A bin-free Extended Maximum Likelihood Fit + Feldman-Cousins error analysis Peter Litchfield  A bin free Extended Maximum Likelihood method of fitting.
ES 07 These slides can be found at optimized for Windows)
8th December 2004Tim Adye1 Proposal for a general-purpose unfolding framework in ROOT Tim Adye Rutherford Appleton Laboratory BaBar Statistics Working.
Spectrum Reconstruction of Atmospheric Neutrinos with Unfolding Techniques Juande Zornoza UW Madison.
Christof Roland / MITQuark Matter 2004January 2004 Christof Roland / MIT For the NA49 Collaboration Quark Matter 2004 Oakland,CA Event-by-Event Fluctuations.
G. Cowan CERN Academic Training 2012 / Statistics for HEP / Lecture 41 Statistics for HEP Lecture 4: Unfolding Academic Training Lectures CERN, 2–5 April,
14/06/11 Jet physics meetingV.Kostyukhin 1 Flavour fractions in di-jet system V.Kostyukhin C.Lapoire M.Lehmacher Bonn.
A High Statistics Study of the Decay M. Fujikawa for the Belle Collaboration Outline 1.Introduction 2.Experiment Belle detector 3.Analysis Event selection.
06/2006I.Larin PrimEx Collaboration meeting  0 analysis.
September 10, 2002M. Fechner1 Energy reconstruction in quasi elastic events unfolding physics and detector effects M. Fechner, Ecole Normale Supérieure.
Tau31 Tracking Efficiency at BaBar Ian Nugent UNIVERSITY OF VICTORIA Sept 2005 Outline Introduction  Decays Efficiency Charge Asymmetry Pt Dependence.
1 Bunch length measurement with the luminous region Z distribution : evolution since 03/04 B. VIAUD, C. O’Grady B. VIAUD, C. O’Grady Origin of the discrepancies.
Charm Mixing and D Dalitz analysis at BESIII SUN Shengsen Institute of High Energy Physics, Beijing (for BESIII Collaboration) 37 th International Conference.
1 D *+ production Alexandr Kozlinskiy Thomas Bauer Vanya Belyaev
Extrapolation Techniques  Four different techniques have been used to extrapolate near detector data to the far detector to predict the neutrino energy.
Muon Energy reconstruction in IceCube and neutrino flux measurement Dmitry Chirkin, University of Wisconsin at Madison, U.S.A., MANTS meeting, fall 2009.
LNF 12/12/06 1 F.Ambrosino-T. Capussela-F.Perfetto Update on        Dalitz plot slope Where we started from A big surprise Systematic checks.
DESY BT analysis - updates - S. Uozumi Dec-12 th 2011 ScECAL meeting.
23 Jan 2012 Background shape estimates using sidebands Paul Dauncey G. Davies, D. Futyan, J. Hays, M. Jarvis, M. Kenzie, C. Seez, J. Virdee, N. Wardle.
Update on A FB Myfanwy Liles μ-μ- μ+μ+ q q θ*θ*. A FB Reminder  Forward-Backward Asymmetry  Due to parity violation of the weak interaction  Interference.
ESTIMATION.
Reduction of Variables in Parameter Inference
Results of dN/dt Elastic
Unfolding Problem: A Machine Learning Approach
p0 life time analysis: general method, updates and preliminary result
° status report analysis details: overview; “where we are”; plans: before finalizing result.. I.Larin 02/13/2009.
Introduction to Unfolding
Juande Zornoza UW Madison
Unfolding with system identification
Partial Wave Analysis results from JETSET the Jetset experiment
° status report analysis details: overview; “where we are”; plans: before finalizing result.. I.Larin 02/13/2009.
Presentation transcript:

1 Iterative dynamically stabilized (IDS) method of data unfolding (*) (*arXiv: ) Bogdan MALAESCU CERN PHYSTAT 2011 Workshop on unfolding

2 Outlook Introduction: main effects to deal with Additional problems in practice An iterative unfolding method A complex example Discussion and conclusions

3 Introduction: detector effects, folding and unfolding Example of transfer matrix (MC) A ij i j Folding: Unfolding of detector effects (acceptance corrected afterwards) Unfolding is not a simple numerical problem Must use a regularization method. Resolution + Distortion

4 Problems in practice: fluctuations due to background subtraction A “standard” unfolding could propagate large fluctuations into precise regions of the spectrum The uncertainties of the data points must be taken into account in the unfolding! (used to compute the significance of data-MC differences in each bin) Folding Unfolding Background subtraction

5 Problems in practice: transfer matrix simulation  perfect Key: use the significance of data-MC differences in each bin New structure (not simulated) MC  improved normalization MC  standard normalization Detector simulation (folding): systematic uncertainty New structures in data: must also be corrected for detector effects could bias MC normalization (needed in the unfolding, for data-MC comparison)

6 Ingredient for the unfolding procedure: a regularization function Used to “measure” significance in the (bin by bin) comparison of experimental data and MC simulation Allows one to perform a different treatment of fluctuations and significant new structures in data Important for the dynamical regularization of fluctuations Depends (monotonously) on the absolute data – MC difference, their uncertainties and a parameter (scale factor) Behavior at small/large parameter values is important, but the exact choice of the function is not critical Used at all the steps of the unfolding procedure, with different values for

7 Model for the test of the method Transfer matrix model: For the folding Fluctuated matrix used for the unfolding Reconstructed MC Generated MC Resolution effect Systematic transfer of events

8 Generated MC Data New Structures Data  Reconstructed MC Data  Generated MC Model for the test of the method Reconstructed MC Generated MC + New Structures  Truth Data

9 First estimation of the number of events in data, corresponding to structures simulated by MC: A better estimation: The same method at the level of (corrected spectrum/ generated MC) # data ev., in the bin k # background subtraction fluctuation ev., in the bin k ITERATIONS Ingredients for the unfolding procedure: the MC normalization procedure

10 Relative improvement of the normalization: (N D – N D MC )/N D The number of iterations is important only in the unstable region The size of the unstable region depends on the amplitude of fluctuations in background subtraction Study performed directly on data! 50 iterations ( at most ) λ N Choice λNλN StableUnstable Ingredients for the unfolding procedure: the MC normalization procedure

11 Ingredients for the unfolding procedure: one step of the unfolding method Folding: Unfolding matrix (like d’Agostini method): By construction: Unfolding: compare data and reconstructed MC spectra General equation Only approximate for spectra other than MC Fluctuation in background subtraction True MCSignificant difference (unfolded) Not significant difference (fixed) A ij i j

12 1 st step of the unfolding method Choice: (all differences between data and reconstructed MC spectra treated as not significant) Reconstructed MC Generated MC + New Structures  Truth Data Data New Structures Data  Reconstructed MC Data  Generated MC Corrected spectrum Corrected spectrum  generated MC If one would choose L =0 …

13 Ingredients for the unfolding procedure : Comparison of the corrected spectrum and generated MC: Estimation of large fluctuations in background subtraction: not significant deviations, with large uncertainties Transfer matrix improvement: use significant structures The folding matrix (P), describing detector effects, stays unchanged. Only the generated MC spectrum is improved. Normalization procedure

14 The Iterative Unfolding Method 1 st unfolding, where the large fluctuations due to background subtraction are kept unchanged 1)Estimation of large fluctuations due to background subtraction 2)Transfer matrix improvement (hence of the unfolding probability matrix) 3)Improved unfolding  Dynamical regularization: from the treatment of fluctuations in each bin, at each step of the procedure When should the iterations stop? Comparison of data and reconstructed MC Study the number of needed iterations, with toys  Choice of parameters used at different steps, with a model for data. One can (in general) give up some of the parameters (by performing a maximal unfolding & transfer matrix modification).

15 Results after iterations Data – improved reconstructed MC Estimation of background fluctuations Data  Reconstructed MC New structures

16 Unfolding Result New Structures Initial reconstructed MC Initial generated MC + New Structures  Truth Data Data Data  Initial reconstructed MC Data  Initial generated MC Corrected spectrum Corrected spectrum  Initial generated MC Statistical uncertainties propagated using pseudo-experiments (“toys”).

17 Discussion Studied but not discussed: N bins data  N bins result (rebinning in the unfolding or afterwards) Effect of rebinning on correlations Effect of regularization on uncertainties and correlations (see Kerstin’s talk) Treatment of bins with negative number of events (data) Empty bins in MC Preventing the existence of negative bins in the improved generated MC

18 Conclusion New general method for the unfolding of binned data Can treat problems that were not considered previously Dynamic regularization procedure, bin by bin at each step This method allows one to keep some control of bin to bin correlations in the unfolded spectrum Root code is available

19 Backup

20 Zoom on the narrow resonance region

21 Simplified example: Reduced effects of the transfer matrix Smoother « bias », without structures No « deeps » in the spectrum No important fluctuations from background subtraction Statistics reduced by a factor 20 A simple example for the use of the unfolding method Data uncertainties Data  Final reconstructed MC (after one iteration) Data  Initial reconstructed MC

22 Simplified unfolding method : Standard normalization for the MC No estimation of left fluctuations (from background subtraction) 1 st unfolding with λ = λ L ( = 1.5, justified by a study (see next)) One iteration with λ U = λ M =0 Effect of the 2 nd unfolding Effect of the 1 st unfolding Data uncertainties A simple example for the use of the unfolding method

23 Use (data – reconstructed MC) as bias with respect to the generated MC, in order to build « generated data » (toys) Folding with the matrix A ij (Do not) Fluctuate the folded data Unfolding with the matrix A’ ij (A ij fluctuated) Compare the result with the « generated data » A test with known « generated data » (before folding) No extra data fluctuations: test systematic effectsWith statistical data fluctuations: stability test Data uncertainties Data  Initial reconstructed MC Data  Final reconstructed MC (after one iteration)

24 Bias measurement after unfolding (without statistical fluctuations of folded data) Result – generated data (1 st step) Result – generated data (2 nd step) Bias measurement after unfolding (without statistical fluctuations of folded data) in large bins The 1 st unfolding provides a good result λ L = 1.5 : very small bias and reduced correlations with respect to the case λ L = 0 Data uncertainties A test with known « generated data » (before folding)

25 Diagonal uncertainties after the 1 st unfolding: larger in the non trivial case (less correlations between the bins) Uncertainties after 1 st unfolding λ L = 0 Uncertainties after 1 st unfolding λ L = 1.5 Data uncertainties A simple example for the use of the unfolding method