DirectFit reconstruction of the Aya’s two HE cascade events Dmitry Chirkin, UW Madison Method of the fit: exhaustive search simulate cascade events with.

Slides:



Advertisements
Similar presentations
Sean Grullon w/ Gary Hill Maximum likelihood reconstruction of events using waveforms.
Advertisements

Dialogue Policy Optimisation
14.5m 100 m ~480 m Junction Box ~70 m String-based detector; Downward-looking (45°) PMTs; 2475 m deep; 12 detection lines 25 storeys / line 3 PMTs / storey.
Bayesian Estimation in MARK
Use of Kalman filters in time and frequency analysis John Davis 1st May 2011.
CHAPTER 8 A NNEALING- T YPE A LGORITHMS Organization of chapter in ISSO –Introduction to simulated annealing –Simulated annealing algorithm Basic algorithm.
Sampling Distributions (§ )
SPICE Mie [mi:] Dmitry Chirkin, UW Madison. Updates to ppc and spice PPC: Randomized the simulation based on system time (with us resolution) Added the.
June 6 th, 2011 N. Cartiglia 1 “Measurement of the pp inelastic cross section using pile-up events with the CMS detector” How to use pile-up.
Hidden Markov Models Theory By Johan Walters (SR 2003)
ISSPIT Ajman University of Science & Technology, UAE
HMM-BASED PATTERN DETECTION. Outline  Markov Process  Hidden Markov Models Elements Basic Problems Evaluation Optimization Training Implementation 2-D.
Bayesian Analysis of X-ray Luminosity Functions A. Ptak (JHU) Abstract Often only a relatively small number of sources of a given class are detected in.
Motion Analysis (contd.) Slides are from RPI Registration Class.
Using Rational Approximations for Evaluating the Reliability of Highly Reliable Systems Z. Koren, J. Rajagopal, C. M. Krishna, and I. Koren Dept. of Elect.
M. Kowalski Search for Neutrino-Induced Cascades in AMANDA II Marek Kowalski DESY-Zeuthen Workshop on Ultra High Energy Neutrino Telescopes Chiba,
End of Chapter 8 Neil Weisenfeld March 28, 2005.
Sept , 2005M. Block, Phystat 05, Oxford PHYSTAT 05 - Oxford 12th - 15th September 2005 Statistical problems in Particle Physics, Astrophysics and.
Using ranking and DCE data to value health states on the QALY scale using conventional and Bayesian methods Theresa Cain.
Physics 114: Lecture 15 Probability Tests & Linear Fitting Dale E. Gary NJIT Physics Department.
Overview G. Jogesh Babu. Probability theory Probability is all about flip of a coin Conditional probability & Bayes theorem (Bayesian analysis) Expectation,
Sean Grullon with Gary Hill Maximum likelihood reconstruction of events using waveforms.
Bayesian parameter estimation in cosmology with Population Monte Carlo By Darell Moodley (UKZN) Supervisor: Prof. K Moodley (UKZN) SKA Postgraduate conference,
880.P20 Winter 2006 Richard Kass 1 Maximum Likelihood Method (MLM) Does this procedure make sense? The MLM answers this question and provides a method.
Introduction to MCMC and BUGS. Computational problems More parameters -> even more parameter combinations Exact computation and grid approximation become.
W  eν The W->eν analysis is a phi uniformity calibration, and only yields relative calibration constants. This means that all of the α’s in a given eta.
Exam I review Understanding the meaning of the terminology we use. Quick calculations that indicate understanding of the basis of methods. Many of the.
R. Kass/W03P416/Lecture 7 1 Lecture 7 Some Advanced Topics using Propagation of Errors and Least Squares Fitting Error on the mean (review from Lecture.
IceCube: String 21 reconstruction Dmitry Chirkin, LBNL Presented by Spencer Klein LLH reconstruction algorithm Reconstruction of digital waveforms Muon.
1 Robot Environment Interaction Environment perception provides information about the environment’s state, and it tends to increase the robot’s knowledge.
1 Methods of Experimental Particle Physics Alexei Safonov Lecture #23.
A taste of statistics Normal error (Gaussian) distribution  most important in statistical analysis of data, describes the distribution of random observations.
Sparse Signals Reconstruction Via Adaptive Iterative Greedy Algorithm Ahmed Aziz, Ahmed Salim, Walid Osamy Presenter : 張庭豪 International Journal of Computer.
ME Mechanical and Thermal Systems Lab Fall 2011 Chapter 3: Assessing and Presenting Experimental Data Professor: Sam Kassegne, PhD, PE.
Vaida Bartkutė, Leonidas Sakalauskas
Ice Investigation with PPC Dmitry Chirkin, UW (photon propagation code)
Data collected during the year 2006 by the first 9 strings of IceCube can be used to measure the energy spectrum of the atmospheric muon neutrino flux.
Detection of electromagnetic showers along muon tracks Salvatore Mangano (IFIC)
July 11, 2006Bayesian Inference and Maximum Entropy Probing the covariance matrix Kenneth M. Hanson T-16, Nuclear Physics; Theoretical Division Los.
Boosted Particle Filter: Multitarget Detection and Tracking Fayin Li.
Ice model update Dmitry Chirkin, UW Madison IceCube Collaboration meeting, Calibration session, March 2014.
MCMC reconstruction of the 2 HE cascade events Dmitry Chirkin, UW Madison.
Feature Extractor: overview and history of recent changes Dmitry Chirkin, UW Madison Goal: Given an ATWD or FADC waveform, determine arrival times of some.
Photonics Tables Bin Optimization Kyle Mandli Paolo Desiati University of Wisconsin – Madison Wuppertal AMANDA Collaboration Meeting.
CMSSM and NUHM Analysis by BayesFITS Leszek Roszkowski* National Centre for Nuclear Research (NCBJ) Warsaw, Poland Leszek Roszkowski *On leave of absence.
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)
Kalanand Mishra June 29, Branching Ratio Measurements of Decays D 0  π - π + π 0, D 0  K - K + π 0 Relative to D 0  K - π + π 0 Giampiero Mancinelli,
G. Cowan Lectures on Statistical Data Analysis Lecture 10 page 1 Statistical Data Analysis: Lecture 10 1Probability, Bayes’ theorem 2Random variables and.
Richard Kass/F02P416 Lecture 6 1 Lecture 6 Chi Square Distribution (  2 ) and Least Squares Fitting Chi Square Distribution (  2 ) (See Taylor Ch 8,
Kevin Stevenson AST 4762/5765. What is MCMC?  Random sampling algorithm  Estimates model parameters and their uncertainty  Only samples regions of.
Tony Hyun Kim 9/26/ MW Probability theory 1.Independent events 2.Poisson distribution 2. Experimental setup 3. Results 1.Comparison to.
Muon Energy reconstruction in IceCube and neutrino flux measurement Dmitry Chirkin, University of Wisconsin at Madison, U.S.A., MANTS meeting, fall 2009.
R. Kass/Sp07P416/Lecture 71 More on Least Squares Fit (LSQF) In Lec 5, we discussed how we can fit our data points to a linear function (straight line)
Monte Carlo Sampling to Inverse Problems Wojciech Dębski Inst. Geophys. Polish Acad. Sci. 1 st Orfeus workshop: Waveform inversion.
A Method of Shower Reconstruction from the Fluorescence Detector M.Giller, G.Wieczorek and the Lodz Auger group GZK-40 Moscow Workshop, May 2006.
Generalization Performance of Exchange Monte Carlo Method for Normal Mixture Models Kenji Nagata, Sumio Watanabe Tokyo Institute of Technology.
Light Propagation in the South Pole Ice
ESTIMATION.
South Pole Ice model Dmitry Chirkin, UW, Madison.
South Pole Ice (SPICE) model
SPICECUBE.
Probability & Statistics Probability Theory Mathematical Probability Models Event Relationships Distributions of Random Variables Continuous Random.
Experimental setup (SPICE)
Unfolding performance Data - Monte Carlo comparison
CHAPTER- 3.1 ERROR ANALYSIS.
Reconstruction of the SC with rime
Sampling Distributions (§ )
Dilepton Mass. Progress report.
Stochastic Methods.
Presentation transcript:

DirectFit reconstruction of the Aya’s two HE cascade events Dmitry Chirkin, UW Madison Method of the fit: exhaustive search simulate cascade events with various x,y,z, ,E,t 0, and compare them to the data event. the simulated event that looks most like the data event is the result Advantages: simple and robust (is not stuck in local minima) most precise description of ice can be used in reconstruction (SPICE Lea: including tilt and anisotropy) Drawbacks: can be very slow (~1 day/Aya’s event)

Comparing a simulation event with a data event Use the same likelihood function as in the SPICE Lea fit: includes Poisson fluctuations in data, simulation, and a 20% allowance for non-Poisson errors (in description of ice and others). All feature-extracted waveforms (charge histogram vs. time in a DOM) are binned in 25 ns bins, and then processed with a Bayesian blocks procedure, which combines low-count or nearly-same-charge bins. This is the same procedure as was used in the SPICE Lea fit. So, we are using exactly the same comparison procedure as was used in the SPICE Lea model fit.

Search algorithm 1.Start with x,y,z of COG,  =0,  =0, E=10 5 GeV, t 0 =0 2.Propose 25 sets of cascade parameters x,y,z, ,  from a gaussian distribution with rms=10 m in x,y,z and rms=30 degrees in , . Keep the values of E and t 0. 3.For each proposed simulated event find the best E (by scaling the simulated event) and t 0 (by time-shifting the hits in the simulated event); calculate the likelihood L. 4.Out of these 25 event select the one with the best value of L and update the 7 cascade parameters; remember the best value of L: L *. 5.Repeat steps times. Use 20 events resulting from step 4 with the best values of L * to update the rms in x,y,z, and rms in , , and to establish correlation between these (important since the brightest point of the cascade is some distance away from the starting point along the cascade direction; also the Cherenkov light is emitted predominantly forward). 6.Repeat steps times; The final result is calculated by averaging simulated events with the best 160 values of L *. The rms of x,y,z, and the rms in ,  are also computed to provide a measure of uncertainties.

Other search algorithms and uncertainties The algorithm described on the previous slide is an optimized variant of Localized random search. Other methods that I tried are: Simultaneous perturbation stochastic approximation with and without the estimate of the second derivative (Newton-like method). Markov chain with transitional probability defined by condition L i+1 <L i. Although the rms values in cascade parameters obtained in the localized random search and Markov chain methods are probably related to the uncertainties of the measurement, the well-defined values of the uncertainties should probably be calculated by applying the reconstruction to a few (dozen?) cascade events simulated with the same parameters.

Results: llh vs. step number Event Event

Results: z vs. llh Event Event

Results: llh Event Event …200 all 201…400

Results: z Event Event …200 all 201…400

Results: t 0 vs. E Event Event

Results: E Event Event …200 all 201…400

Results: t 0 Event Event …200 all 201…400

Results: y vs. x Event Event

Results: x Event Event …200 all 201…400

Results: y Event Event …200 all 201…400

Results:  vs.  Event Event

Results:  Event Event …200 all 201…400

Results:  Event Event …200 all 201…400

Summary Event Event llh = x = m y = m z = m  = 76.0 o  = o E = 1.115e %  r = 2.7 m  = 10.4 o dist. to pivot = m llh = x = m y = m z = 25.1 m  = 28.8 o  = o E = 1.336e %  r = 3.1 m  = 8.5 o dist. to pivot = m SPICE Lea

Summary Event Event llh = E = 1.115e % llh = E = 1.336e % SPICE Lea llh = E = 1.168e % llh = E = 1.398e % SPICE Mie llh = E = % llh = E = 1.127e % SPICE 1 llh = E = 1.247e % llh = E = 1.565e % WHAM!

Simulation at best point vs. data with SPICE Lea

Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Simulation at best point vs. data with SPICE Lea Event Event

Comparison SPICE Lea vs. SP1 for event

SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

Comparison SPICE Lea vs. SP1 for event SPICE LeaSPICE 1

More plots at