Initial estimates for R 0 and Γ0 R Γ. Γ k. State estimate R Error covariance. k+1 k+1
|
|
- Kathryn Cross
- 6 years ago
- Views:
Transcription
1 HERA-B -32 VDET -1 CATS track fitting algorithm based on the discrete Kalman filter D. Emeliyanov 12 I. Kisel 34 1 Institute of Control Sciences, Profsoyuznaya ul., 65, Moscow, Russia 2 Deutsches Elektronen-Synchrotron DESY, Notkestrasse 85, D-2267, Hamburg, Germany 3 Joint Institute for Nuclear Research, Dubna, Moscow region, Russia 4 Max-Planck-Institut für Physik, Werner Heisenberg Institut, Föhringer Ring 6, D-885 München, Germany Abstract The Kalman filter is a set of mathematical equations that provides an efficient recursive solution of the least squares method. This paper provides a brief introduction to the discrete Kalman filter equations of which are derived for the case of track parameter estimation with measurements from the vertex detector. To reduce computational cost of the algorithm an optimized numerical implementation is proposed. The algorithm was tested on simulated data. Test results regarding an accuracy of estimates and computing time are presented. 1 The discrete Kalman filter The Kalman filter addresses the general problem of trying to estimate the state vector R of a discrete-time process that is governed by the linear difference equation [1] R k+1 = A k R k + ν k (1) with a measurement u that is the linear function of the vector R u k = H k R k + k : (2) The matrix A k in the difference equation (1) relates the state at step k to the state at step k +1, the matrix H k in the measurement equation (2) relates the state to the measurement u k. The random variables k and ν k represent the measurement noise and process noise, respectively. They are assumed to be independent with Gaussian distributions. Let's define b R k to be a state estimate at step k after processing measurement u k. The main idea of the Kalman filter is that the optimal (in mean-square sense) estimate b R k should be a sum of an extrapolated estimate e R k and a weighted difference between an actual measurement u k and a measurement prediction H k e Rk br k = e R k + K k (u k H k e Rk ) where e R k = A k 1 b R k 1 : The difference (u k H k e Rk ) is called the residual. The residual reflects the discrepancy between the predicted measurementvalue H k e Rk and the actual measurement u k. A residual of zero means that they are in complete agreement. 1
2 The matrix K k is called the filter gain and is chosen to minimize the sum of diagonal elements of an estimation error covariance matrix b k. By definition, h b k = E (R k R b k )(R k R b i k ) T where E denotes mathematical expectation, T denotes transposition. measurements, the minimization leads to the following formula for K k In the case of scalar K k = e k V k + H k e k where V k is the covariance of the measurement error k, e k is an extrapolated estimation error covariance matrix b k 1. The formula for e k follows from equation (1): e k = A k 1 b k 1 AT k 1 + Q k 1 where Q k 1 is the covariance matrix of the process noise ν. The new minimized value of the error covariance matrix b k is defined by the equation b k =(I K k H k ) e k where I is aunity matrix of the same dimensions as matrix. In fact, this equation describes an update of the matrix b k after each measurement. The computational algorithm of the discrete Kalman filter consists of two steps (Fig. 1): ~ ~ R Γ k+1 k+1 Initial estimates for R and Γ Prediction step Filtering step R k Γ k State estimate R Error covariance Figure 1: The operation diagram of the discrete Kalman filter. Γ 1. Prediction step extrapolation of the estimate b R and error covariance matrix b to the next step of the algorithm. 2. Filtering step er k = A k 1 b R k 1 e k = A k 1 b k 1 AT k 1 + Q k 1: (3) (a) The gain matrix calculation K k = e k V k + H k e k (4) 2
3 (b) The updated estimate br k = e R k + K k (u k H k e Rk ) (5) (c) The updated error covariance matrix b k =(I K k H k ) e k : (6) After each Prediction/Filtering pair, the process is repeated until all measurements are processed. 2 A Kalman filter for estimating track parameters In context of this paper, a track through the vertex detector is described as a straight-line motion in the presence of Gaussian disturbances: x k+1 = x k + t xk (z k+1 z k ) y k+1 = y k + t yk (z k+1 z k ) t xk+1 = t xk + xk t yk+1 = t yk + yk T where R k = x k y k t xk t yk is the vector of the track parameters taken at a plane z = zk, T denotes transposition, xk, yk are non-correlated random values normally distributed with parameters (ffx(z 2 k )), (ffy(z 2 k )). These random values are introduced in order to describe the multiple scattering which affects the track when it passes through detector planes z = z k, k = 1::: For description of the multiple scattering process we use the model shown in Fig.2. It is is (7) Y Wall Detector planes Wall σ 1 σ 2 σ 2 σ3 σ 3 σ 4 Z z min z min z max z min z max z min Figure 2: Illustration of the multiple scattering model used for track fit. similar to the model proposed in [2] and takes into account a two-sided structure of the VDS detector planes. A particle enters the detector volume on the surface of z min, produces the hit and undergoes the scattering. Then it produces hit on the other plane at z max, undergoes the scattering again and, finally, leaves the volume. In the case of insensitive walls, the particle undergoes the scattering only once on z min surface. The model of the measurements (hits) in the VDS has the following form: u k = u(z k )=x(z k )cosff k y(z k ) sin ff k + ff k k = 1::: (8) where ff k rotation angle of the strips in the plane z = z k, k a random value normally distributed with parameters ( 1), ff r.m.s of the measurement noise. 3
4 The problem is to obtain an estimation of the vector's R components using the measurements fu k g, k = 1 2 :::. It is assumed that the track recognition problem is already solved before fitting, and the measurements fu k g are collected in a downstream order: z k >z k 1, k>. To solve the stated problem, we propose to use the discrete Kalman filter, equations of which are derived below. It should be noted, that the Kalman filter is a natural choice for track fitting in the vertex detector due to the linearity of the track and measurement models in this detector. According to the track model (7), the matrix A in equation (1) has the form A k = 1 z k 1 z k C A z k = z k z k 1 k =1 2:::: The detector volumes and insensitive walls are treated as so-called thin scatterers [2]. In case of such scatterers, the non-zero elements q ij of the process noise covariance matrix Q k equal to q 33 = p 2 ff 2 MS (1+t2 x)(1 + t 2 x + t 2 y) q 1+t 2 x + t 2 y q 34 = q 43 = p 2 ff 2 MS t xt y (1+t 2 x + t 2 y) q 1+t 2 x + t 2 y q 44 = p 2 ff 2 MS (1+t2 y)(1 + t 2 x + t 2 y) q 1+t 2 x + t 2 y where p is an external estimate for inverse momentum of the particle, ffms 2 is the variance of the multiple scattering angle for 1 GeV particle. The latter parameter is obtained from ARTE GEDE and GWAL table for each material the particle passes through. According to the measurement equation (8), the matrix H is: h i H k = cos ff k sin ff k : (9) Using obtained formulas, we can write algorithm of the Kalman filter in the following form. 1. Prediction step, k =1 2:::: ex k = b x k 1 + z k b t x k 1 e xk = b y k 1 + z k b t yk 1 et x k = b t x k 1 e tyk = b t yk 1 e k = A k 1 b k 1 AT k 1 + Q k 1: 2. Filtering step, k = 1::: (a) The gain matrix of the filter K k = e k ff 2 + H k e k (b) The updated estimate of the track parameters at z = z k br k = R e k + K k (u k H e krk ) (c) The updated error covariance matrix b k =(I K k H k ) e k : 4
5 3 Computational refinements of the algorithm It should be noted that any efficient numerical implementation of the Kalman filter algorithm is always application-dependent. There are several ways of reducing computational cost of the algorithm: ffl calculation of the filter covariance matrix in the upper-triangle form, ffl elimination of multiplication by one and zero during prediction and filtering step if the track and measurement models allow such elimination, ffl optimization of the covariance matrix update procedure [3]. Following these directions, we propose a fast numerical implementation of the Kalman filter algorithm in which the number of arithmetical operations is significantly reduced. Calculating the covariance matrix in the upper-triangle form, we obtain the following equation for the prediction step: R 1 = zr b 3 R 2 = zr b 4 R 3 = R 4 = 11 = z 2 b 13 + z b = z 2 b 24 + z b = q = z b + b 23 + z b = z b = q 34 (1) 13 = z b = z b = q = z b 34 where z = z k z k 1, matrix = e k b k 1,vector R = e R k b R k 1. The most time-consuming parts of the filtering step are calculations of the gain matrix and the updated covariance matrix. Substitution of the measurement matrix (9) into (4) yields the following formula for the gain matrix K k = B k s k where vector B and scalar value s are B 1 = e 11 cos ff k e 12 sin ff k B 2 = e 12 cos ff k e 22 sin ff k B 3 = e 13 cos ff k e 23 sin ff k B 4 = e 14 cos ff k e 24 sin ff k (11) s = ff 2 + e 11 cos 2 ff k e 12 sin 2ff k + e 22 sin 2 ff k : (12) In terms of B and s, the filtering step of the filter can be simplified. The formula for updated estimate takes the form br k = e R k + C k k where k = u k e R 1 cos ff k + e R 2 sin ff k (13) and C k = s 1 k B k. The covariance matrix is updated as follows b 11 = e 11 B 1 C 1 b 22 = e 22 B 2 C 2 b 33 = e 33 B 3 C 3 b 12 = e 12 B 1 C 2 b 23 = e 11 B 2 C 3 b 34 = e 11 B 3 C 4 b 13 = e 13 B 1 C 3 b 24 = e 24 B 2 C 4 b 44 = e 44 B 4 C 4 (14) b 14 = e 14 B 1 C 4 The equations (1)-(14) describe the fast implementation of the Kalman filter algorithm for track fitting in the VDS. The number of arithmetical operations needed for the fast algorithm 5
6 Operation: Multiplication Division Summation Classical scheme Fast scheme Table 1: Computational costs of the fast and classical implementation of the algorithm. in comparison with the classical implementation defined by (3)-(6) is presented in Table 1. The state vector is assumed to be four-dimensional one. It can be easily seen that the number of operations for proposed implementation is almost 5 times less than for the classical scheme. Usually the actual implementation of the classical scheme is even more slow due to the use of libraries for matrix operations, in particular, for matrix multiplication. Note that the fast version of the algorithm needs no matrix multiplications. 4 The initialization of the Kalman filter To obtain the initial estimates of vector R b and covariance matrix b = G two space points are used. Let's their coordinates be (x I y I z I ) for the first point and (x II y II z II ) for the second one (Fig. 3). The components of vector R b can be easily obtained from an equations of the Y II Z X I α ΙΙ α Ι Figure 3: The space points used for initialization of the filter. straight line drawn through the space points: br 1 = x I(z II z ) x II (z I z ) z II z I br 2 = y I(z II z ) y II (z I z ) z II z I br 3 = x II x I z II z I b R4 = y II y I z II z I : The errors of estimates of the space point coordinates are assumed to be non-correlated random values normally distributed with parameters (ff 2 ). It is also assumed that angles ff I, ff II between strips which form the space points are approximately ±ß=2. Then the initial covariance matrix can be represented in the following upper-triangle form G 11 = G 22 = ff 2 (z (z II z I ) 2 II z ) 2 +(z I z ) 2 +2ff 2 (z II z I ) 2 6
7 G 13 = G 24 = G 33 = G 44 = ff 2 z (z II z I ) 2 II + z I 2z 2ff 2 (z II z I ) 2 +2ff2 G 12 = G 14 = G 23 = G 34 =: In these formulas, the additional terms with ff 2 are included in order to take the influence of the multiple scattering on the initial data into account. In the numerical implementation of the initialization procedure, ff is equal to 2. mrad that seems to be a quite reasonable approximation. 5 Results and discussion The developed algorithm was implemented numerically in CATS program [4] and its performance has been tested using simulated inelastic events. Some results of the test are presented below. Table 2 provides the reconstruction efficiencies (see definitions in the CATS note [4]) of reference and all tracks, the clone rate, the ghost rate and the time needed by CATS (total, reconstruction and fitting only) for different number of mixed Poisson distributed interactions for fully instrumented VDS. While mean interaction rate is expected to be 4, we investigated CATS up to 1 interactions. The reconstruction efficiency for reference tracks slowly goes down from about 98% to 89% demonstrating stability of the program. The clone rate is constantly abut 3%. The ghost rate grows from 7% up to 3%. Ghost tracks are mainly very short tracks which should be accepted or rejected at the next step of tracking using information from other subdetectors. The total time used by CATS grows almost linearly from 6 to 174 ms per event running on PC hb-af3 at DESY, Hamburg. This time is mainly determined by the reconstruction itself. The Kalman refit procedure takes about 1 ms per interaction. The rest of time spent for storing the ARTE tables. It should be emphasized that such low computing time is achieved due to optimized numerical implementation of the filter. Interactions Refset Allset Clone Ghost t, total t, reco t, fit Table 2: Results of a tracking efficiency study using different number of mixed Poisson distributed inelastic events generated in a Monte Carlo simulation. Tracking efficiency is measured in per cent and time unit is ms. For each track the algorithm for Kalman refit has estimated track parameters in two planes: z = z f and z = z l, where z f, z l are z-coordinates of the first and the last hits of a simulated track, respectively. The reliability of parameter error estimation was studied by investigating normalized residual distributions, which use the estimated errors for normalization. The normalized residual (also called the pull) of a track parameter, for instance x-coordinate at the first 7
8 hit, is defined by P(x) = xrec x MC p 11 where x REC is the reconstructed x position of a track and x MC the corresponding Monte Carlo value, while 11 is the estimate for the corresponding covariance matrix diagonal element. In the ideal case the pull distribution should be unbiased and have gaussian core of unity. The distribution of residuals and pulls for the plane of the first hit are plotted in Fig. 4 and in Fig. 5. The results of testing on Monte Carlo simulated inelastic events show that 1. The algorithm provides high accuracy of the estimation of track parameters (see Table 3). ff of residual first hit plane last hit plane x, μm y, μm t x,mrad t y,mrad Table 3: Sigma of track parameter residuals. 2. All observed residual distributions have shapes close to Gaussian without heavy tails. 3. The non-gaussian tails in the pull distributions are small, the gaussian cores of the pulls agree with unity, indicating reliable estimate of the covariance matrix. 4. The comparison with the ideal case shows that if particle momentum is known the non- Gaussian tails in the distributions completely disappear. It means that the multiple scattering model embedded in the filter corresponds quite accurately to the way the particle's interaction with the detector is simulated. 6 Summary An algorithm for track parameter estimation in the vertex detector has been developed. The algorithm is based on the Kalman filter technique and takes effects from multiple scattering into account. Due to an optimized numerical implementation the algorithm shows very low computing time. Tested on simulated data, a good performance regarding estimate of track parameters and their errors is observed. References [1] R.K. Bock et al., Data analysis techniques for high-energy physics experiments, Cambridge Univ. Press (199). [2] R. Mankel and A. Spiridonov, A Pattern Recognition Algorithm for the HERA-B Main Tracking System, Part IV: The Object-Oriented Track Fit, HERA-B Note (1998). [3] K. Brammer and G. Siffling, Kalman-Bucy Filter: Deterministische Beobachtung und Stochastische Filterung, (1974). [4] I. Kisel and S. Masciocchi, CATS: A Cellular Automaton for Tracking in Silicon for the HERA-B Vertex Detector, HERA-B Note (1999). 8
9 Mean E E / 84 Constant 39.2 Mean -.221E-4 Sigma.13E Mean.3991E E / 86 Constant Mean.1556E-4 Sigma.15E x first residual ideal y first residual ideal Mean.522E E / 97 Constant 279. Mean.4428E-5 Sigma.591E Mean -.582E-5.136E / 97 Constant Mean.1417E-4 Sigma.5826E dpx/dpz first residual ideal dpy/dpz first residual ideal Mean -.31E / 77 Constant Mean E-2 Sigma x first pull ideal Mean E / 78 Constant 37.1 Mean.1575E-2 Sigma dpx/dpz first pull ideal Mean.285E / 81 Constant 383. Mean.163E-1 Sigma y first pull ideal Mean E / 82 Constant Mean.8215E-2 Sigma dpy/dpz first pull ideal Figure 4: Residuals and normalized residuals (pulls) of tracks parameters at the first hit obtained on simulated inelastic events with known momentum value. 9
10 Mean E E / 85 Constant Mean E-4 Sigma.142E x first residual Allset Mean.622E E / 85 Constant Mean.2355E-4 Sigma.174E y first residual Allset Mean E E / 97 Constant Mean E-5 Sigma.5346E Mean E-5.147E / 97 Constant 218. Mean.925E-5 Sigma.638E dpx/dpz first residual Allset dpy/dpz first residual Allset Mean -.167E / 89 Constant Mean E-2 Sigma x first pull Allset Mean.3966E / 85 Constant Mean.225E-1 Sigma y first pull Allset Mean -.268E / 94 Constant Mean.6385E-2 Sigma Mean -.492E / 96 Constant Mean.4615E-2 Sigma dpx/dpz first pull Allset dpy/dpz first pull Allset Figure 5: Residuals and normalized residuals (pulls) of tracks parameters at the first hit obtained on simulated inelastic events with approximate momentum value. 1
Track Reconstruction and Muon Identification in the Muon Detector of the CBM Experiment at FAIR
Track Reconstruction and Muon Identification in the Muon Detector of the CBM Experiment at FAIR ab, Claudia Höhne a, Ivan Kisel ac, Anna Kiseleva a and Gennady Ososkov b for the CBM collaboration a GSI
More informationGEANT4 simulation of the testbeam set-up for the ALFA detector
GEANT4 simulation of the testbeam set-up for the detector V. Vorobel a and H. Stenzel b a Faculty of Mathematics and Physics, Charles University in Prague, Czech Republic b II. Physikalisches Institut,
More informationarxiv:hep-ex/ v1 19 Sep 1998
The Concurrent Track Evolution Algorithm: Extension for Track Finding in the Inhomogeneous Magnetic Field of the HERA-B Spectrometer arxiv:hep-ex/98921v1 19 Sep 1998 Rainer Mankel 1 Institut für Physik,
More informationA fast introduction to the tracking and to the Kalman filter. Alberto Rotondi Pavia
A fast introduction to the tracking and to the Kalman filter Alberto Rotondi Pavia The tracking To reconstruct the particle path to find the origin (vertex) and the momentum The trajectory is usually curved
More informationCS 532: 3D Computer Vision 6 th Set of Notes
1 CS 532: 3D Computer Vision 6 th Set of Notes Instructor: Philippos Mordohai Webpage: www.cs.stevens.edu/~mordohai E-mail: Philippos.Mordohai@stevens.edu Office: Lieb 215 Lecture Outline Intro to Covariance
More informationFirst Year Presentation
Friday 4 th July 2003 First Year Presentation Development of HzTool and a new event display for ZEUS Christopher Targett-Adams, University College London 1 Outline esentation outline... Brief introduction
More informationCosmic Muons and the Cosmic Challenge. data taken with B 3.5Tesla, red lines are reconstructed chamber segments
Cosmic Muons and the Cosmic Challenge data taken with B 3.5Tesla, red lines are reconstructed chamber segments 0 Preamble Warning: Data analyzed in this talk are just a few days old, so all interpretations
More informationAPPLICATION OF KALHAN FILTERING TO TRACK FITTING IN THE DELPHI DETECTOR. R. Fruhwirth
DELPHI 87-23 PROG 70 24 March 1987 APPLICATION OF KALHAN FILTERING TO TRACK FITTING IN THE DELPHI DETECTOR R. Fruhwirth APPLICATION OF KALMAN FILTERING TO TRACK FITTING IN THE DELPHI DETECTOR R_Frohwirth
More informationNon-collision Background Monitoring Using the Semi-Conductor Tracker of ATLAS at LHC
WDS'12 Proceedings of Contributed Papers, Part III, 142 146, 212. ISBN 978-8-7378-226-9 MATFYZPRESS Non-collision Background Monitoring Using the Semi-Conductor Tracker of ATLAS at LHC I. Chalupková, Z.
More informationFATRAS. A Novel Fast Track Simulation Engine for the ATLAS Experiment. Sebastian Fleischmann on behalf of the ATLAS Collaboration
A Novel Fast Track Engine for the ATLAS Experiment on behalf of the ATLAS Collaboration Physikalisches Institut University of Bonn February 26th 2010 ACAT 2010 Jaipur, India 1 The ATLAS detector Fast detector
More informationarxiv: v1 [hep-ex] 6 Jul 2007
Muon Identification at ALAS and Oliver Kortner Max-Planck-Institut für Physik, Föhringer Ring, D-005 München, Germany arxiv:0707.0905v1 [hep-ex] Jul 007 Abstract. Muonic final states will provide clean
More informationSome studies for ALICE
Some studies for ALICE Motivations for a p-p programme in ALICE Special features of the ALICE detector Preliminary studies of Physics Performances of ALICE for the measurement of some global properties
More informationBose-Einstein correlations in hadron-pairs from lepto-production on nuclei ranging from hydrogen to xenon
in hadron-pairs from lepto-production on nuclei ranging from hydrogen to xenon Yerevan Physics Institute, A. Alikhanyan Br. 2, Yerevan, Armenia DESY, Notkestrasse 85, Hamburg, Germany E-mail: gevkar@email.desy.de
More informationUsing the Kalman Filter to Estimate the State of a Maneuvering Aircraft
1 Using the Kalman Filter to Estimate the State of a Maneuvering Aircraft K. Meier and A. Desai Abstract Using sensors that only measure the bearing angle and range of an aircraft, a Kalman filter is implemented
More informationMultiple Scattering of a π beam in the HERA-B detector.
Multiple Scattering of a π beam in the HERA-B detector. Antonios Garas Aristotle University of Thessaloniki Department of Physics Thessaloniki, Greece. Email: agara@skiathos.physics.auth.gr We study the
More informationAlignment of the ATLAS Inner Detector
Alignment of the ATLAS Inner Detector Roland Härtel - MPI für Physik on behalf of the ATLAS inner detector alignment group Outline: LHC / ATLAS / Inner Detector Alignment approaches Results Large Hadron
More informationParticle Filters. Outline
Particle Filters M. Sami Fadali Professor of EE University of Nevada Outline Monte Carlo integration. Particle filter. Importance sampling. Degeneracy Resampling Example. 1 2 Monte Carlo Integration Numerical
More informationKalman Filter. Predict: Update: x k k 1 = F k x k 1 k 1 + B k u k P k k 1 = F k P k 1 k 1 F T k + Q
Kalman Filter Kalman Filter Predict: x k k 1 = F k x k 1 k 1 + B k u k P k k 1 = F k P k 1 k 1 F T k + Q Update: K = P k k 1 Hk T (H k P k k 1 Hk T + R) 1 x k k = x k k 1 + K(z k H k x k k 1 ) P k k =(I
More informationTracking and Fitting. Natalia Kuznetsova, UCSB. BaBar Detector Physics Series. November 19, November 19, 1999 Natalia Kuznetsova, UCSB 1
Tracking and Fitting Natalia Kuznetsova, UCSB BaBar Detector Physics Series November 19, 1999 November 19, 1999 Natalia Kuznetsova, UCSB 1 Outline BaBar tracking devices: SVT and DCH Track finding SVT
More informationAbsolute energy calibration
E coeffs 1 Absolute energy calibration V. Morgunov DESY, Hamburg and ITEP, Moscow HCAL main meeting, DESY, January 19, 26 The copy of this talk one can find at the http://www.desy.de/ morgunov Jan 19,
More informationCMS Note Mailing address: CMS CERN, CH-1211 GENEVA 23, Switzerland
Available on CMS information server CMS NOTE 21/17 The Compact Muon Solenoid Experiment CMS Note Mailing address: CMS CERN, CH-1211 GENEVA 23, Switzerland 2 March 21 Study of a Level-3 Tau Trigger with
More informationValidation of Geant4 Physics Models Using Collision Data from the LHC
Journal of Physics: Conference Series Validation of Geant4 Physics Models Using Collision from the LHC To cite this article: S Banerjee and CMS Experiment 20 J. Phys.: Conf. Ser. 33 032003 Related content
More informationNeutrino-Nucleus Scattering at MINERvA
1 Neutrino-Nucleus Scattering at MINERvA Elba XIII Workshop: Neutrino Physics IV Tammy Walton Fermilab June 26, 2014 2 MINERvA Motivation Big Picture Enter an era of precision neutrino oscillation measurements.
More informationCharm with ZEUS HERA-II data
Charm with ZEUS HERA-II data Falk Karstens on behalf of the ZEUS collaboration University of Freiburg/ Germany Outline: DIS 26 21th April 26 HF-2 D ± mesons in deep inelastic
More informationPATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 13: SEQUENTIAL DATA
PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 13: SEQUENTIAL DATA Contents in latter part Linear Dynamical Systems What is different from HMM? Kalman filter Its strength and limitation Particle Filter
More informationBeauty jet production at HERA with ZEUS
DESY and University of Hamburg E-mail: mlisovyi@mail.desy.de wo recent measurements of beauty production in deep inelastic ep scattering with the detector are presented. In the first analysis, beauty production
More informationNonlinear State Estimation! Extended Kalman Filters!
Nonlinear State Estimation! Extended Kalman Filters! Robert Stengel! Optimal Control and Estimation, MAE 546! Princeton University, 2017!! Deformation of the probability distribution!! Neighboring-optimal
More information45-Degree Update of The Wic Fitter Algorithm. Rob Kroeger, Vance Eschenburg
45-Degree Update of The Wic Fitter Algorithm Rob Kroeger, Vance Eschenburg Abstract This note describes changes in the calculation of the inverse weight matrix due to multiple scattering for the 45-Degree
More informationMeasurements of the Proton F L and F 2 Structure Functions at Low x at HERA
Measurements of the Proton F L and F 2 Structure Functions at Low x at HERA Andrei Nikiforov DESY on behalf of the H1 and ZEUS collaborations Moriond QCD, La Thuile, 13 March 2008 Andrei Nikiforov DESY
More informationA Novel Maneuvering Target Tracking Algorithm for Radar/Infrared Sensors
Chinese Journal of Electronics Vol.19 No.4 Oct. 21 A Novel Maneuvering Target Tracking Algorithm for Radar/Infrared Sensors YIN Jihao 1 CUIBingzhe 2 and WANG Yifei 1 (1.School of Astronautics Beihang University
More informationarxiv: v1 [hep-ex] 6 Aug 2008
GLAS-PPE/28-9 24 th July 28 arxiv:88.867v1 [hep-ex] 6 Aug 28 Department of Physics and Astronomy Experimental Particle Physics Group Kelvin Building, University of Glasgow, Glasgow, G12 8QQ, Scotland Telephone:
More informationthe robot in its current estimated position and orientation (also include a point at the reference point of the robot)
CSCI 4190 Introduction to Robotic Algorithms, Spring 006 Assignment : out February 13, due February 3 and March Localization and the extended Kalman filter In this assignment, you will write a program
More informationNA62: Ultra-Rare Kaon Decays
NA62: Ultra-Rare Kaon Decays Phil Rubin George Mason University For the NA62 Collaboration November 10, 2011 The primary goal of the experiment is to reconstruct more than 100 K + π + ν ν events, over
More informationCalorimeter energy calibration using the energy conservation law
Abs Calibr 1 Calorimeter energy calibration using the energy conservation law V. Morgunov DESY, Hamburg and ITEP, Moscow LCWS26, Bangalore, India, 26. The copy of this talk one can find at the http://www.desy.de/
More informationA New Detector for Physics at HERA - III
A New Detector for Physics at HERA - III CIPANP New York May 2003 Iris Abt MPI für Physik isa@mppmu.mpg.de Content Detector Concept Interaction Region Silicon Tracker Calorimetry Acceptance Momentum Resolution
More informationAlignment of the ATLAS Inner Detector. Oscar Estrada Pastor (IFIC-CSIC)
Alignment of the ATLAS Inner Detector Oscar Estrada Pastor (IFIC-CSIC) 1 Summary table Introduction. ATLAS: the detector. Alignment of the Inner Detector. Weak modes. Weak modes using J/Psi resonance.
More informationImpact of the Velo 2 half misalignment on physical quantities
LHCb-INT-216-36 August 19, 216 Impact of the Velo 2 half misalignment on physical quantities Matej Roguljić 1, Silvia Borghi 2, Lucia Grillo 2,3, Giulio Dujany 2. 1 CERN, Geneva, Switzerland 2 The University
More informationTrack reconstruction for the Mu3e experiment Alexandr Kozlinskiy (Mainz, KPH) for the Mu3e collaboration DPG Würzburg (.03.22, T85.
Track reconstruction for the Mu3e experiment Alexandr Kozlinskiy (Mainz, KPH) for the Mu3e collaboration DPG 2018 @ Würzburg (.03.22, T85.1) Mu3e Experiment Mu3e Experiment: Search for Lepton Flavor Violation
More informationπ p-elastic Scattering in the Resonance Region
New Results on Spin Rotation Parameter A in the π p-elastic Scattering in the Resonance Region I.G. Alekseev Λ, P.E. Budkovsky Λ, V.P. Kanavets Λ, L.I. Koroleva Λ, B.V. Morozov Λ, V.M. Nesterov Λ, V.V.
More informationExperiments on deflection of charged particles using silicon crystals at. Proton Synchrotron (KEK(
Experiments on deflection of charged particles using silicon crystals at REFER ring (Hiroshima( University) and Proton Synchrotron (KEK( KEK) S. Sawada KEK High Energy Accelerator Research Organization
More informationTrack measurement in the high multiplicity environment at the CBM Experiment
Track measurement in the high multiplicity environment at the CBM Experiment Pradeep Ghosh 1,, for the CBM Collaboration 1 Goethe University, Max-von-Laue-Straße 1, D-60438 Frankfurt am Main GSI Helmholtz
More informationAlignment of the ATLAS Inner Detector tracking system
Alignment of the ALAS Inner Detector tracking system Oleg BRAND University of Oxford and University of Göttingen E-mail: oleg.brandt@cern.ch he Large Hadron Collider (LHC) at CERN is the world largest
More informationThe Detector Design of the Jefferson Lab EIC
The Detector Design of the Jefferson Lab EIC Jefferson Lab E-mail: mdiefent@jlab.org The Electron-Ion Collider (EIC) is envisioned as the next-generation U.S. facility to study quarks and gluons in strongly
More informationSpacal alignment and calibration
Spacal alignment and calibration Sebastian Piec AGH University of Science and Technology Al. Mickiewicza 3, Cracow, Poland Email: sepiec@poczta.onet.pl The main purpose of my work was alignment and calibration
More informationStatistics. Lent Term 2015 Prof. Mark Thomson. 2: The Gaussian Limit
Statistics Lent Term 2015 Prof. Mark Thomson Lecture 2 : The Gaussian Limit Prof. M.A. Thomson Lent Term 2015 29 Lecture Lecture Lecture Lecture 1: Back to basics Introduction, Probability distribution
More informationMeasurement of Muon Momentum Using Multiple Coulomb Scattering for the MicroBooNE Experiment
Measurement of Muon Momentum Using Multiple Coulomb Scattering for the MicroBooNE Experiment Polina Abratenko August 5, 2016 1 Outline What is MicroBooNE? How does the TPC work? Why liquid argon? What
More informationThe experiment at JINR: status and physics program
The 3rd International Conference on Particle Physics and Astrophysics Volume 2018 Conference Paper The BM@N experiment at JINR: status and physics program D. Baranov, M. Kapishin, T. Mamontova, G. Pokatashkin,
More informationPAMELA satellite: fragmentation in the instrument
PAMELA satellite: fragmentation in the instrument Alessandro Bruno INFN, Bari (Italy) for the PAMELA collaboration Nuclear Physics for Galactic CRs in the AMS-02 era 3-4 Dec 2012 LPSC, Grenoble The PAMELA
More informationDESY Summer Students Program 2008: Exclusive π + Production in Deep Inelastic Scattering
DESY Summer Students Program 8: Exclusive π + Production in Deep Inelastic Scattering Falk Töppel date: September 6, 8 Supervisors: Rebecca Lamb, Andreas Mussgiller II CONTENTS Contents Abstract Introduction.
More informationDirect Measurement of the W Total Decay Width at DØ. Outline
Direct Measurement of the W Total Decay Width at DØ Introduction Junjie Zhu University of Maryland On behalf of the DØ Collaboration Monte Carlo Simulation Event Selection Outline Determination of the
More informationResults from HARP. Malcolm Ellis On behalf of the HARP collaboration DPF Meeting Riverside, August 2004
Results from HARP Malcolm Ellis On behalf of the HARP collaboration DPF Meeting Riverside, August 2004 The HAdRon Production Experiment 124 physicists 24 institutes2 Physics Goals Input for precise calculation
More informationCalorimeter ECAL (tungsten silicon) HCAL (tile, iron - scintillator)
Calorimeter ECAL (tungsten silicon) HCAL (tile, iron - scintillator) Vasiliy Morgunov DESY, Hamburg and ITEP, Moscow Workshop, Padova - May 00 Vasiliy Morgunov Padova, - May 00 Calorimeter parameters ECAL
More informationCMS Note Mailing address: CMS CERN, CH-1211 GENEVA 23, Switzerland
Available on CMS information server CMS OTE 997/064 The Compact Muon Solenoid Experiment CMS ote Mailing address: CMS CER, CH- GEEVA 3, Switzerland 3 July 997 Explicit Covariance Matrix for Particle Measurement
More informationExperimental results on nucleon structure Lecture I. National Nuclear Physics Summer School 2013
Experimental results on nucleon structure Lecture I Barbara Badelek University of Warsaw National Nuclear Physics Summer School 2013 Stony Brook University, July 15 26, 2013 Barbara Badelek (Univ. of Warsaw
More informationof the gauge eld U which one obtains after integrating through K = (8 + 2m) 1. It is easy to show that the lattice Dirac operator
A new simulation algorithm for lattice QCD with dynamical quarks B. Bunk a, K. Jansen b, B. Jegerlehner c, M. Luscher b, H. Simma b and R. Sommer d a Humboldt Universitat, Institut fur Physik Invalidenstrasse
More informationPoS(ICHEP 2010)170. D +, D 0 and Λ + c production in deep inelastic scattering at HERA
D +, D and Λ + c production in deep inelastic scattering at HERA ab a Deutsches Elektronen-Synchrotron DESY, Notkestr. 85, D-67 Hamburg, Germany b University of Hamburg, Institute of Experimental Physics,
More informationBayes Filter Reminder. Kalman Filter Localization. Properties of Gaussians. Gaussians. Prediction. Correction. σ 2. Univariate. 1 2πσ e.
Kalman Filter Localization Bayes Filter Reminder Prediction Correction Gaussians p(x) ~ N(µ,σ 2 ) : Properties of Gaussians Univariate p(x) = 1 1 2πσ e 2 (x µ) 2 σ 2 µ Univariate -σ σ Multivariate µ Multivariate
More informationThe Kalman Filter (part 1) Definition. Rudolf Emil Kalman. Why do we need a filter? Definition. HCI/ComS 575X: Computational Perception.
The Kalman Filter (part 1) HCI/ComS 575X: Computational Perception Instructor: Alexander Stoytchev http://www.cs.iastate.edu/~alex/classes/2007_spring_575x/ March 5, 2007 HCI/ComS 575X: Computational Perception
More informationMeasurement of charged particle spectra in pp collisions at CMS
Measurement of arged particle spectra in pp collisions at CMS for the CMS Collaboration Eötvös Loránd University, Budapest, Hungary E-mail: krisztian.krajczar@cern. We present the plans of the CMS collaboration
More informationarxiv:hep-ex/ v1 5 Apr 2000
Track Fit Hypothesis Testing and Kink Selection using Sequential Correlations Robert V. Kowalewski and Paul D. Jackson arxiv:hep-ex/48v 5 Apr 2 Abstract Department of Physics and Astronomy, University
More informationPoS(Bormio2012)013. K 0 production in pp and pnb reactions. Jia-Chii Berger-Chen
Excellence Cluster Universe, Technische Universität München E-mail: jia-chii.chen@tum.de The kaon nucleus (KN) interaction in nuclear matter is predicted to be repulsive and dependent from the density.
More informationDer Silizium Recoil Detektor für HERMES Ingrid-Maria Gregor
Der Silizium Recoil Detektor für HERMES Introduction HERMES at DESY Hamburg What do we want to measure? Recoil Detector Overview Silicon Recoil Detector Principle First measurements Zeuthen activities
More informationLeast Squares Estimation Namrata Vaswani,
Least Squares Estimation Namrata Vaswani, namrata@iastate.edu Least Squares Estimation 1 Recall: Geometric Intuition for Least Squares Minimize J(x) = y Hx 2 Solution satisfies: H T H ˆx = H T y, i.e.
More informationFeasibility of a cross-section measurement for J/ψ->ee with the ATLAS detector
Feasibility of a cross-section measurement for J/ψ->ee with the ATLAS detector ATLAS Geneva physics meeting Andrée Robichaud-Véronneau Outline Motivation Theoretical background for J/ψ hadroproduction
More informationFeasibility Studies for the EXL Project at FAIR *
* a,b,, S. Bagchi c, S. Diebold d, C. Dimopoulou a, P. Egelhof a, V. Eremin e, S. Ilieva a, N. Kalantar-Nayestanaki c, O. Kiselev a,f, T. Kröll f, Y.A. Litvinov a,g, M. Mutterer a, M.A. Najafi c, N. Petridis
More informationContents. What Are We Looking For? Predicted D
Contents D * e S DS e Souvik Das, Anders Ryd Cornell University What Are We Looking For? Predicted D * S D S e e - Rate Decay Modes of D S Used Fitting Soft Electrons Signal Samples Background Samples
More informationCIS 390 Fall 2016 Robotics: Planning and Perception Final Review Questions
CIS 390 Fall 2016 Robotics: Planning and Perception Final Review Questions December 14, 2016 Questions Throughout the following questions we will assume that x t is the state vector at time t, z t is the
More informationHARP (Hadron Production) Experiment at CERN
HARP (Hadron Production) Experiment at CERN 2nd Summer School On Particle Accelerators And Detectors 18-24 Sep 2006, Bodrum, Turkey Aysel Kayιş Topaksu Çukurova Üniversitesi, ADANA Outline The Physics
More informationTrack Fitting With Broken Lines for the MU3E Experiment
Track Fitting With Broken Lines for the MU3E Experiment Moritz Kiehn, Niklaus Berger and André Schöning Institute of Physics Heidelberg University DPG Frühjahrstagung Göttingen 01 Introduction Overview
More informationDiffusion and cellular-level simulation. CS/CME/BioE/Biophys/BMI 279 Nov. 7 and 9, 2017 Ron Dror
Diffusion and cellular-level simulation CS/CME/BioE/Biophys/BMI 279 Nov. 7 and 9, 2017 Ron Dror 1 Outline How do molecules move around in a cell? Diffusion as a random walk (particle-based perspective)
More informationStudy of the HARPO TPC for a high angular resolution g-ray polarimeter in the MeV-GeV energy range. David Attié (CEA/Irfu)
Study of the HARPO TPC for a high angular resolution g-ray polarimeter in the MeV-GeV energy range David Attié (CEA/Irfu) Outline Motivation of an MeV-GeV polarimeter Scientific case and expected performance
More informationInclusive Deep-Inelastic Scattering at HERA
Inclusive Deep-Inelastic Scattering at HERA Vladimir Chekelian (MPI for Physics, Munich) on behalf of the H and ZEUS Collaborations Completion of the HERA inclusive DIS cross section measurements: HERA
More informationIf we want to analyze experimental or simulated data we might encounter the following tasks:
Chapter 1 Introduction If we want to analyze experimental or simulated data we might encounter the following tasks: Characterization of the source of the signal and diagnosis Studying dependencies Prediction
More informationLuminosity measurement and K-short production with first LHCb data. Sophie Redford University of Oxford for the LHCb collaboration
Luminosity measurement and K-short production with first LHCb data Sophie Redford University of Oxford for the LHCb collaboration 1 Introduction Measurement of the prompt Ks production Using data collected
More informationLecture 2: From Linear Regression to Kalman Filter and Beyond
Lecture 2: From Linear Regression to Kalman Filter and Beyond January 18, 2017 Contents 1 Batch and Recursive Estimation 2 Towards Bayesian Filtering 3 Kalman Filter and Bayesian Filtering and Smoothing
More informationThe TAIGA experiment - a hybrid detector for very high energy gamma-ray astronomy and cosmic ray physics in the Tunka valley
The TAIGA experiment - a hybrid detector for very high energy gamma-ray astronomy and cosmic ray physics in the Tunka valley N. Budnev, Irkutsk State University For the TAIGA collaboration The TAIGA experiment
More informationPROBABILISTIC REASONING OVER TIME
PROBABILISTIC REASONING OVER TIME In which we try to interpret the present, understand the past, and perhaps predict the future, even when very little is crystal clear. Outline Time and uncertainty Inference:
More informationOnline monitoring of MPC disturbance models using closed-loop data
Online monitoring of MPC disturbance models using closed-loop data Brian J. Odelson and James B. Rawlings Department of Chemical Engineering University of Wisconsin-Madison Online Optimization Based Identification
More informationLatest Results from the OPERA Experiment (and new Charge Reconstruction)
Latest Results from the OPERA Experiment (and new Charge Reconstruction) on behalf of the OPERA Collaboration University of Hamburg Institute for Experimental Physics Overview The OPERA Experiment Oscillation
More informationThe ALICE Inner Tracking System Off-line Software
The ALICE Inner Tracking System Off-line Software Roberto Barbera 1;2 for the ALICE Collaboration 1 Istituto Nazionale di Fisica Nucleare, Sezione di Catania Italy 2 Dipartimento di Fisica dell Università
More informationOptimal Linear Unbiased Filtering with Polar Measurements for Target Tracking Λ
Optimal Linear Unbiased Filtering with Polar Measurements for Target Tracking Λ Zhanlue Zhao X. Rong Li Vesselin P. Jilkov Department of Electrical Engineering, University of New Orleans, New Orleans,
More informationIntroduction to Unscented Kalman Filter
Introduction to Unscented Kalman Filter 1 Introdution In many scientific fields, we use certain models to describe the dynamics of system, such as mobile robot, vision tracking and so on. The word dynamics
More information2. How Tracks are Measured Charged particle tracks in high energy physics experiments are measured through a two step process. First, detector measure
Applied Fitting Theory V Track Fitting Using the Kalman Filter Paul Avery CBX 92{39 August 2, 992 Section : Introduction Section 2: How Tracks are Measured Section 3: The Standard Track Fit Section 4:
More informationThe Kalman Filter ImPr Talk
The Kalman Filter ImPr Talk Ged Ridgway Centre for Medical Image Computing November, 2006 Outline What is the Kalman Filter? State Space Models Kalman Filter Overview Bayesian Updating of Estimates Kalman
More informationATLAS detector and physics performance Technical Design Report 25 May 1999
3 Inner Detector 3. Introduction The ATLAS Inner Detector (ID) and its performance were described extensively in the ID TDR [3-][3-2]. In this chapter, a summary of the results which are of most interest
More informationarxiv: v1 [hep-ex] 1 Oct 2015
CIPANP2015-Galati October 2, 2015 OPERA neutrino oscillation search: status and perspectives arxiv:1510.00343v1 [hep-ex] 1 Oct 2015 Giuliana Galati 1 Università degli Studi di Napoli Federico II and INFN
More informationMuon reconstruction performance in ATLAS at Run-2
2 Muon reconstruction performance in ATLAS at Run-2 Hannah Herde on behalf of the ATLAS Collaboration Brandeis University (US) E-mail: hannah.herde@cern.ch ATL-PHYS-PROC-205-2 5 October 205 The ATLAS muon
More informationCALICE Test Beam Data and Hadronic Shower Models
EUDET CALICE Test Beam Data and Hadronic Shower Models Riccardo Fabbri on behalf of the CALICE Collaboration FLC, DESY, Notkestrasse 85, 67 Hamburg, Germany Email: Riccardo.Fabbri@desy.de January 1, 1
More informationLinear-Optimal State Estimation
Linear-Optimal State Estimation Robert Stengel Optimal Control and Estimation MAE 546 Princeton University, 218 Linear-optimal Gaussian estimator for discrete-time system (Kalman filter) 2 nd -order example
More informationGeometry optimization of a barrel silicon pixelated tracker *
Chinese Physics C Vol. 1, No. 8 (017) 086001 Geometry optimization of a barrel silicon pixelated tracker * Qing-Yuan Liu() 1, Meng Wang() 1;1) Marc Winter 1 School of Physics and Key Laboratory of Particle
More informationSearch for heavy neutrinos in kaon decays
Search for heavy neutrinos in kaon decays L. Littenberg (work mainly done by A.T.Shaikhiev INR RAS) HQL-2016 Outline Motivation Previous heavy neutrino searches Experiment BNL-E949 Selection criteria Efficiency
More informationarxiv: v2 [hep-ex] 12 Feb 2014
arxiv:141.476v2 [hep-ex] 12 Feb 214 on behalf of the COMPASS collaboration Technische Universität München E-mail: stefan.huber@cern.ch The value of the pion polarizability is predicted with high precision
More informationLeast Squares and Kalman Filtering Questions: me,
Least Squares and Kalman Filtering Questions: Email me, namrata@ece.gatech.edu Least Squares and Kalman Filtering 1 Recall: Weighted Least Squares y = Hx + e Minimize Solution: J(x) = (y Hx) T W (y Hx)
More informationarxiv: v1 [hep-ph] 30 Dec 2015
June 3, 8 Derivation of functional equations for Feynman integrals from algebraic relations arxiv:5.94v [hep-ph] 3 Dec 5 O.V. Tarasov II. Institut für Theoretische Physik, Universität Hamburg, Luruper
More informationThe HERMES Dual-Radiator Ring Imaging Cerenkov Detector N.Akopov et al., Nucl. Instrum. Meth. A479 (2002) 511
The HERMES Dual-Radiator Ring Imaging Cerenkov Detector N.Akopov et al., Nucl. Instrum. Meth. A479 (2002) 511 Shibata Lab 11R50047 Jennifer Newsham YSEP student from Georgia Institute of Technology, Atlanta,
More informationMobile Robot Localization
Mobile Robot Localization 1 The Problem of Robot Localization Given a map of the environment, how can a robot determine its pose (planar coordinates + orientation)? Two sources of uncertainty: - observations
More informationSusanna Costanza. (Università degli Studi di Pavia & INFN Pavia) on behalf of the ALICE Collaboration
(Università degli Studi di Pavia & INFN Pavia) on behalf of the ALICE Collaboration 102 Congresso della Società Italiana di Fisica Padova, 26-30 settembre 2016 Outline Heavy flavour physics in ALICE The
More informationLecture 6: Multiple Model Filtering, Particle Filtering and Other Approximations
Lecture 6: Multiple Model Filtering, Particle Filtering and Other Approximations Department of Biomedical Engineering and Computational Science Aalto University April 28, 2010 Contents 1 Multiple Model
More informationResonance analysis in pp collisions with the ALICE detector
IL NUOVO CIMENO Vol.?, N.?? Resonance analysis in pp collisions with the ALICE detector A Pulvirenti( 1 )( 2 )( ), for the ALICE Collaboration ( 1 ) INFN, Sezione di Catania ( 2 ) Università di Catania
More informationarxiv: v1 [hep-ex] 18 Feb 2009
GARLIC - GAmma Reconstruciton for the LInear Collider Marcel Reinhard and Jean-Claude Brient arxiv:92.342v1 [hep-ex] 18 Feb 29 LLR - Ecole polytechnique, IN2P3/CNRS Palaiseau - France In order to profit
More information