Holonomic Gradient Method for Multivariate Normal Distribution Theory. Akimichi Takemura, Univ. of Tokyo August 2, 2013
|
|
- Kevin Giles Hopkins
- 6 years ago
- Views:
Transcription
1 Holonomic Gradient Method for Multivariate Normal Distribution Theory Akimichi Takemura, Univ. of Tokyo August 2, 2013
2 Items 1. First example (non-central t-distribution, by T.Sei) 2. Principles of HGM for normal distribution theory 3. Orthant probability 4. Wishart distribution and hypergeometric function of a matrix argument 5. Sum of weighted independent non-central χ 2 (1) variables (in preparation) 1
3 Joint works with many people, in particular Tamio Koyama, T.Sei, N.Takayama. Orthant probability discussed in Koyama and T. arxiv: (Submitted) Wishart discussed in arxiv: v3. Published in Journal of Multivariate Analysis. Our first paper (N 3 OST 2 ): Nakayama-Nishiyama-Noro-Ohara-Sei- Takayama-Takemura: Holonomic gradient descent and its application to the Fisher-Bingham integral, Advances in Applied Mathematics, 47, ,
4 The first example: Non-central t distribution by T.Sei X: χ 2 variable with k degrees of freedom Y : N(λ, 1) variable Let T = Y/ X/k. Compute Pr(T q). From standard textbooks (e.g. Johnson-Kotz Vol.2), Pr(T q) is usually expressed as an infinite series with each term involving incomplete beta function. 3
5 Regard q and λ as variables (and leave k fixed). Let F (q, λ) = Pr(T q). Our strategy: derive partial differential equations satisfied by F. Then integrate the differential equation to evaluate F. We call this method Holonomic Gradient Method (HGM). 4
6 We can obtain three partial differential equations: λf 2 = q q F λ λ F (1) ( ) qλ q 2 2 k q(k + 1) F = (k + q 2 ) 2 k + q 2 q F λk2 (k + q 2 ) λf (2) 2 q λ F = λk k + q qf qk 2 k + q λf (3) 2 Consider any higher-order derivative i λ j qf of F. Repeatedly applying the above three partial differential equations, any i λ j qf can be reduced to a rational function combination of q F, λ F and F. 5
7 The Pfaffian system for F (q, λ) is given as F F q q F = qλ 0 2 k q(k+1) λk2 (k+q 2 ) 2 k+q 2 (k+q 2 ) 2 q F λ F λ F λ 0 λk k+q 2 qk k+q 2, (4) F F q F = λk qk 0 k+q 2 k+q 2 q F. (5) λ F 0 q λ λ F (there is no singularity in this Pfaffian since k > 0) 6
8 With initial values ( q F )(0, 0) = 1 Γ( k+1 kπ Γ( k), 2 ( λ F )(0, 0) = 1 2π, 2 ) (4) and (5) can be solved by standard numerical solver for differential equations. 7
9 Maybe our first example is not very convincing. The standard method already works OK. The point is that our method is general. There is a general theory and algorithm to obtain the partial differential equations (1) (3) and the Pfaffian system (4), (5), based on Gröbner bases theory for D modules. In this example, although we derived (1)-(5) by hand, it is very important that the general theory tells us what to expect. 8
10 Principles of HGM for normal distribution theory Let X 1,..., X n be i.i.d. random vectors from the multivariate normal distribution N m (µ, Σ). Let R(q) R m n be a parameterized region. Problem: compute F (q, µ, Σ) def = Pr((X 1,..., X n ) R(q)) exp( 1 n (X i µ) T Σ 1 (X i µ))dx 1... dx n R(q) 2 i=1 9
11 FACT: If the family of indicator functions I A(θ) is nice (such as polyhedral region, region defined by polynomial inequalities), i.e., if they are holonomic as Schwartz distributions, then F (q, µ, Σ) is holonomic, i.e., it satisfies a system of partial differential equations with rational function coefficients. This fact means, that the whole sampling distribution theory under the multivariate normal distribution can be treated by HGM. Total rewriting of the sampling theory is in order. 10
12 Orthant probability Let R m + = {x R m x i 0} be the positive orthant. Compute Φ m (Σ, µ) exp ( 1 ) 2 (z µ)t Σ 1 (z µ) dz R m + Define x = 1 2 Σ 1, y = Σ 1 µ and ( m g(x, y) = exp x ij t i t j + R m + i,j=1 m i=1 y i t i ) dt. 11
13 For a subset J of [m] = {1,..., m} let ( g J (x, y) = exp x ij t i t j + ) y k t k dt j. R J + i J j J k J j J Furthermore for J = {j 1,..., j s } write x J = (x jk j l ) 1 k,l s, y J = (y j1,..., y js ) T Σ J = 1 2 x 1 J = (σ J ij), µ J = Σ J y J = (µ J j 1,..., µ J j s ). 12
14 Then g J (x, y) satisfies µ J i g J + j J yi g J = σj ijg J\{j} i J 0 i / J 2 yi yj g J {i, j} J, i < j xij g J = y 2 i g J {i} J, i = j 0 else. (6) (7) These are basically the same as the recurrence relations by Plackett (1954). But our recurrence relations are more suitable for HGM. 13
15 Numerical performance of HGM is good: Table 1: Averages of computational times dim Miwa HGM dim Miwa HGM
16 Wishart distribution and hypergeometric function of a matrix argument W : m m symmetric positive definite (W > 0) W = n i=1 X ix T i, X i N m (0, Σ), i.i.d. Density of Wishart distribution: f(w ) = C n m 1 W 2 Σ n 2 exp( 1 2 trw Σ 1 ) C is known (containing gamma functions). 15
17 l 1 : the largest root of W We want to evaluate the probability Pr(l 1 < x). l 1 < x W < xi m, where I m : m m is the identity matrix Hence the probability is given in the incomplete gamma form: Pr(l 1 < x) = C n m 1 2 W 0<W <xi m Σ n 2 exp( 1 2 trw Σ 1 )dw 16
18 Pr(l 1 < x) is written as C exp ( x ) 2 trσ 1 x 1 2 nm 1F 1 ( m ; n + m + 1 ; x ) 2 2 Σ 1 Hypergeometric function of a matrix argument (Herz(1955)): Γ m (c) 1F 1 (a; c; Y ) = exp(trxy ) Γ m (a)γ m (c a) 0<X<I m X a (m+1)/2 I m X c a (m+1)/2 dx, where Γ m (a) = π 1 4 m(m 1) m i=1 Γ ( a i 1 ). 2 17
19 1 F 1 (a; c; Y ) is a symmetric function of characteristic roots of Y its series expression is written in terms of symmetric polynomials. Zonal polynomials (A.T.James) C κ (Y ), κ k homogeneous symmetric polynomial of degree k in the characteristic roots of Y. 18
20 Series expansion of 1 F 1 (Constantine(1963)) 1F 1 (a; c; Y ) = k=0 κ k (a) κ C κ (Y ). (c) κ k! This is a beautiful mathematical result. However for numerical computation, zonal polynomials have enormous combinatorial difficulties and statisticians pretty much forgot zonal polynomials. 19
21 The partial differential equations satisfied by F (Y ) = 1 F 1 (a; c; y 1,..., y m ) were obtained by Muirhead(1970). g i F = 0, i = 1,..., m, where g i = y i 2 i + (c y i ) i j i y j y i y j ( i j ) a. We can derive Pfaffian system from these equations. 20
22 Can we use this PDE for numerical computation? (People never tried this for 40 years). Works! HGM works very well up to dimension m = 10. Plot of the cumulative distribution: 1.2 by hg
23 Sum of weighted non-central χ 2 (1) variables Let X N m (µ, Σ). Problem: compute Pr( X r), where X = (X Xm) 2 1/2 is the Euclidean norm. The distribution of X 2 is the distribution of the sum of weighted independent non-central χ 2 (1) variables. 22
24 Given r = X, the conditional density of X on is proportional to S m 1 (r) = {x x = r} exp ( 1 2 (x µ)t Σ 1 (x µ) ) It is the Fisher-Bingham distribution on S m 1 (r) (cf. our first paper N 3 OST 2.) Hence we can just integrate the holonomic system given in N 3 OST 2 w.r.t. r and evaluate Pr( X r). Directional statistics multivariate normal 23
25 Summary HGM is very relevant for the whole sampling distribution theory under the multivariate normal distribution. HGM is practical if we implement it efficiently. HGM is general and can be applied to many problems. We stand at the beginning of applications of D-module theory to statistics! 24
Holonomic gradient method for hypergeometric functions of a matrix argument Akimichi Takemura, Shiga University
Holonomic gradient method for hypergeometric functions of a matrix argument Akimichi Takemura, Shiga University (joint with H.Hashiguchi and N.Takayama) June 20, 2016, RIMS, Kyoto Items 1. Summary of multivariate
More informationHolonomic gradient method for distribution function of a weighted sum of noncentral chi-square random variables
Holonomic gradient method for distribution function of a weighted sum of noncentral chi-square random variables T.Koyama & A.Takemura March 3, 2015 1 The Ball Probability (1). Let X be a d-dimensional
More informationUse of Asymptotics for Holonomic Gradient Method
Use of Asymptotics for Holonomic Gradient Method Akimichi Takemura Shiga University July 25, 2016 A.Takemura (Shiga U) HGM and asymptotics 2015/3/3 1 / 30 Some disclaimers In this talk, I consider numerical
More informationEcon 508B: Lecture 5
Econ 508B: Lecture 5 Expectation, MGF and CGF Hongyi Liu Washington University in St. Louis July 31, 2017 Hongyi Liu (Washington University in St. Louis) Math Camp 2017 Stats July 31, 2017 1 / 23 Outline
More information2 Lie Groups. Contents
2 Lie Groups Contents 2.1 Algebraic Properties 25 2.2 Topological Properties 27 2.3 Unification of Algebra and Topology 29 2.4 Unexpected Simplification 31 2.5 Conclusion 31 2.6 Problems 32 Lie groups
More informationRandom Variables and Their Distributions
Chapter 3 Random Variables and Their Distributions A random variable (r.v.) is a function that assigns one and only one numerical value to each simple event in an experiment. We will denote r.vs by capital
More informationBIOS 2083 Linear Models Abdus S. Wahed. Chapter 2 84
Chapter 2 84 Chapter 3 Random Vectors and Multivariate Normal Distributions 3.1 Random vectors Definition 3.1.1. Random vector. Random vectors are vectors of random variables. For instance, X = X 1 X 2.
More informationc 2005 Society for Industrial and Applied Mathematics
SIAM J. MATRIX ANAL. APPL. Vol. XX, No. X, pp. XX XX c 005 Society for Industrial and Applied Mathematics DISTRIBUTIONS OF THE EXTREME EIGENVALUES OF THE COMPLEX JACOBI RANDOM MATRIX ENSEMBLE PLAMEN KOEV
More informationVarious algorithms for the computation of Bernstein-Sato polynomial
Various algorithms for the computation of Bernstein-Sato polynomial Applications of Computer Algebra (ACA) 2008 Notations and Definitions Motivation Two Approaches Let K be a field and let D = K x, = K
More informationStatistics for scientists and engineers
Statistics for scientists and engineers February 0, 006 Contents Introduction. Motivation - why study statistics?................................... Examples..................................................3
More informationAlgorithms for D-modules, integration, and generalized functions with applications to statistics
Algorithms for D-modules, integration, and generalized functions with applications to statistics Toshinori Oaku Department of Mathematics, Tokyo Woman s Christian University March 18, 2017 Abstract This
More informationA Few Special Distributions and Their Properties
A Few Special Distributions and Their Properties Econ 690 Purdue University Justin L. Tobias (Purdue) Distributional Catalog 1 / 20 Special Distributions and Their Associated Properties 1 Uniform Distribution
More informationPart IA Probability. Definitions. Based on lectures by R. Weber Notes taken by Dexter Chua. Lent 2015
Part IA Probability Definitions Based on lectures by R. Weber Notes taken by Dexter Chua Lent 2015 These notes are not endorsed by the lecturers, and I have modified them (often significantly) after lectures.
More information1 Quantum states and von Neumann entropy
Lecture 9: Quantum entropy maximization CSE 599S: Entropy optimality, Winter 2016 Instructor: James R. Lee Last updated: February 15, 2016 1 Quantum states and von Neumann entropy Recall that S sym n n
More informationTube formula approach to testing multivariate normality and testing uniformity on the sphere
Tube formula approach to testing multivariate normality and testing uniformity on the sphere Akimichi Takemura 1 Satoshi Kuriki 2 1 University of Tokyo 2 Institute of Statistical Mathematics December 11,
More information2 Chance constrained programming
2 Chance constrained programming In this Chapter we give a brief introduction to chance constrained programming. The goals are to motivate the subject and to give the reader an idea of the related difficulties.
More informationComputational Algebraic Statistics, Theories and Applications CASTA 2014, Jan , 2014, Kyoto, Japan Abstracts
Computational Algebraic Statistics, Theories and Applications CASTA 2014, Jan. 21-24, 2014, Kyoto, Japan Abstracts Day 1: Tuesday, Jan. 21st Time: 9:00-9:40 Registration Time: 9:40-10:40 Hidefumi Ohsugi
More informationAnderson-Darling Type Goodness-of-fit Statistic Based on a Multifold Integrated Empirical Distribution Function
Anderson-Darling Type Goodness-of-fit Statistic Based on a Multifold Integrated Empirical Distribution Function S. Kuriki (Inst. Stat. Math., Tokyo) and H.-K. Hwang (Academia Sinica) Bernoulli Society
More informationwhere r n = dn+1 x(t)
Random Variables Overview Probability Random variables Transforms of pdfs Moments and cumulants Useful distributions Random vectors Linear transformations of random vectors The multivariate normal distribution
More informationBayesian Methods with Monte Carlo Markov Chains II
Bayesian Methods with Monte Carlo Markov Chains II Henry Horng-Shing Lu Institute of Statistics National Chiao Tung University hslu@stat.nctu.edu.tw http://tigpbp.iis.sinica.edu.tw/courses.htm 1 Part 3
More informationVector spaces. DS-GA 1013 / MATH-GA 2824 Optimization-based Data Analysis.
Vector spaces DS-GA 1013 / MATH-GA 2824 Optimization-based Data Analysis http://www.cims.nyu.edu/~cfgranda/pages/obda_fall17/index.html Carlos Fernandez-Granda Vector space Consists of: A set V A scalar
More informationNotes on Random Vectors and Multivariate Normal
MATH 590 Spring 06 Notes on Random Vectors and Multivariate Normal Properties of Random Vectors If X,, X n are random variables, then X = X,, X n ) is a random vector, with the cumulative distribution
More informationA distinguisher for high-rate McEliece Cryptosystems
A distinguisher for high-rate McEliece Cryptosystems JC Faugère (INRIA, SALSA project), A Otmani (Université Caen- INRIA, SECRET project), L Perret (INRIA, SALSA project), J-P Tillich (INRIA, SECRET project)
More informationGAUSSIAN PROCESS REGRESSION
GAUSSIAN PROCESS REGRESSION CSE 515T Spring 2015 1. BACKGROUND The kernel trick again... The Kernel Trick Consider again the linear regression model: y(x) = φ(x) w + ε, with prior p(w) = N (w; 0, Σ). The
More informationDS-GA 1002 Lecture notes 0 Fall Linear Algebra. These notes provide a review of basic concepts in linear algebra.
DS-GA 1002 Lecture notes 0 Fall 2016 Linear Algebra These notes provide a review of basic concepts in linear algebra. 1 Vector spaces You are no doubt familiar with vectors in R 2 or R 3, i.e. [ ] 1.1
More informationSEA s workshop- MIT - July 10-14
Matrix-valued Stochastic Processes- Eigenvalues Processes and Free Probability SEA s workshop- MIT - July 10-14 July 13, 2006 Outline Matrix-valued stochastic processes. 1- definition and examples. 2-
More informationZonal Polynomials and Hypergeometric Functions of Matrix Argument. Zonal Polynomials and Hypergeometric Functions of Matrix Argument p.
Zonal Polynomials and Hypergeometric Functions of Matrix Argument Zonal Polynomials and Hypergeometric Functions of Matrix Argument p. 1/2 Zonal Polynomials and Hypergeometric Functions of Matrix Argument
More informationOptimization. The value x is called a maximizer of f and is written argmax X f. g(λx + (1 λ)y) < λg(x) + (1 λ)g(y) 0 < λ < 1; x, y X.
Optimization Background: Problem: given a function f(x) defined on X, find x such that f(x ) f(x) for all x X. The value x is called a maximizer of f and is written argmax X f. In general, argmax X f may
More informationRandom Eigenvalue Problems Revisited
Random Eigenvalue Problems Revisited S Adhikari Department of Aerospace Engineering, University of Bristol, Bristol, U.K. Email: S.Adhikari@bristol.ac.uk URL: http://www.aer.bris.ac.uk/contact/academic/adhikari/home.html
More informationSTAT 450. Moment Generating Functions
STAT 450 Moment Generating Functions There are many uses of generating functions in mathematics. We often study the properties of a sequence a n of numbers by creating the function a n s n n0 In statistics
More informationMATH 260 Homework assignment 8 March 21, px 2yq dy. (c) Where C is the part of the line y x, parametrized any way you like.
MATH 26 Homework assignment 8 March 2, 23. Evaluate the line integral x 2 y dx px 2yq dy (a) Where is the part of the parabola y x 2 from p, q to p, q, parametrized as x t, y t 2 for t. (b) Where is the
More informationProgress in the method of Ghosts and Shadows for Beta Ensembles
Progress in the method of Ghosts and Shadows for Beta Ensembles Alan Edelman (MIT) Alex Dubbs (MIT) and Plamen Koev (SJS) Oct 8, 2012 1/47 Wishart Matrices (arbitrary covariance) G=mxn matrix of Gaussians
More informationSampling Distributions
Sampling Distributions In statistics, a random sample is a collection of independent and identically distributed (iid) random variables, and a sampling distribution is the distribution of a function of
More informationContents. I Basic Methods 13
Preface xiii 1 Introduction 1 I Basic Methods 13 2 Convergent and Divergent Series 15 2.1 Introduction... 15 2.1.1 Power series: First steps... 15 2.1.2 Further practical aspects... 17 2.2 Differential
More informationThe Johnson-Lindenstrauss Lemma
The Johnson-Lindenstrauss Lemma Kevin (Min Seong) Park MAT477 Introduction The Johnson-Lindenstrauss Lemma was first introduced in the paper Extensions of Lipschitz mappings into a Hilbert Space by William
More informationAdvanced Machine Learning & Perception
Advanced Machine Learning & Perception Instructor: Tony Jebara Topic 6 Standard Kernels Unusual Input Spaces for Kernels String Kernels Probabilistic Kernels Fisher Kernels Probability Product Kernels
More informationRegularized Least Squares
Regularized Least Squares Ryan M. Rifkin Google, Inc. 2008 Basics: Data Data points S = {(X 1, Y 1 ),...,(X n, Y n )}. We let X simultaneously refer to the set {X 1,...,X n } and to the n by d matrix whose
More informationProbability and Distributions
Probability and Distributions What is a statistical model? A statistical model is a set of assumptions by which the hypothetical population distribution of data is inferred. It is typically postulated
More informationRandom Vectors and Multivariate Normal Distributions
Chapter 3 Random Vectors and Multivariate Normal Distributions 3.1 Random vectors Definition 3.1.1. Random vector. Random vectors are vectors of random 75 variables. For instance, X = X 1 X 2., where each
More informationLecture 22. m n c (k) i,j x i x j = c (k) k=1
Notes on Complexity Theory Last updated: June, 2014 Jonathan Katz Lecture 22 1 N P PCP(poly, 1) We show here a probabilistically checkable proof for N P in which the verifier reads only a constant number
More informationFormulas for probability theory and linear models SF2941
Formulas for probability theory and linear models SF2941 These pages + Appendix 2 of Gut) are permitted as assistance at the exam. 11 maj 2008 Selected formulae of probability Bivariate probability Transforms
More informationSTAT 450: Statistical Theory. Distribution Theory. Reading in Casella and Berger: Ch 2 Sec 1, Ch 4 Sec 1, Ch 4 Sec 6.
STAT 45: Statistical Theory Distribution Theory Reading in Casella and Berger: Ch 2 Sec 1, Ch 4 Sec 1, Ch 4 Sec 6. Basic Problem: Start with assumptions about f or CDF of random vector X (X 1,..., X p
More informationMultivariate Analysis and Likelihood Inference
Multivariate Analysis and Likelihood Inference Outline 1 Joint Distribution of Random Variables 2 Principal Component Analysis (PCA) 3 Multivariate Normal Distribution 4 Likelihood Inference Joint density
More informationStability of Feedback Solutions for Infinite Horizon Noncooperative Differential Games
Stability of Feedback Solutions for Infinite Horizon Noncooperative Differential Games Alberto Bressan ) and Khai T. Nguyen ) *) Department of Mathematics, Penn State University **) Department of Mathematics,
More informationUnit 2: Modeling in the Frequency Domain Part 2: The Laplace Transform. The Laplace Transform. The need for Laplace
Unit : Modeling in the Frequency Domain Part : Engineering 81: Control Systems I Faculty of Engineering & Applied Science Memorial University of Newfoundland January 1, 010 1 Pair Table Unit, Part : Unit,
More information0.0.1 Moment Generating Functions
0.0.1 Moment Generating Functions There are many uses of generating functions in mathematics. We often study the properties of a sequence a n of numbers by creating the function a n s n n0 In statistics
More informationRandom Methods for Linear Algebra
Gittens gittens@acm.caltech.edu Applied and Computational Mathematics California Institue of Technology October 2, 2009 Outline The Johnson-Lindenstrauss Transform 1 The Johnson-Lindenstrauss Transform
More information1 Uniform Distribution. 2 Gamma Distribution. 3 Inverse Gamma Distribution. 4 Multivariate Normal Distribution. 5 Multivariate Student-t Distribution
A Few Special Distributions Their Properties Econ 675 Iowa State University November 1 006 Justin L Tobias (ISU Distributional Catalog November 1 006 1 / 0 Special Distributions Their Associated Properties
More informationStable Process. 2. Multivariate Stable Distributions. July, 2006
Stable Process 2. Multivariate Stable Distributions July, 2006 1. Stable random vectors. 2. Characteristic functions. 3. Strictly stable and symmetric stable random vectors. 4. Sub-Gaussian random vectors.
More informationSampling Distributions
In statistics, a random sample is a collection of independent and identically distributed (iid) random variables, and a sampling distribution is the distribution of a function of random sample. For example,
More informationPetter Mostad Mathematical Statistics Chalmers and GU
Petter Mostad Mathematical Statistics Chalmers and GU Solution to MVE55/MSG8 Mathematical statistics and discrete mathematics MVE55/MSG8 Matematisk statistik och diskret matematik Re-exam: 4 August 6,
More informationSTAT 801: Mathematical Statistics. Moment Generating Functions. M X (t) = E(e tx ) M X (u) = E[e utx ]
Next Section Previous Section STAT 801: Mathematical Statistics Moment Generating Functions Definition: The moment generating function of a real valued X is M X (t) = E(e tx ) defined for those real t
More informationA Probability Review
A Probability Review Outline: A probability review Shorthand notation: RV stands for random variable EE 527, Detection and Estimation Theory, # 0b 1 A Probability Review Reading: Go over handouts 2 5 in
More informationRandom Matrix Eigenvalue Problems in Probabilistic Structural Mechanics
Random Matrix Eigenvalue Problems in Probabilistic Structural Mechanics S Adhikari Department of Aerospace Engineering, University of Bristol, Bristol, U.K. URL: http://www.aer.bris.ac.uk/contact/academic/adhikari/home.html
More informationMinicourse on Complex Hénon Maps
Minicourse on Complex Hénon Maps (joint with Misha Lyubich) Lecture 2: Currents and their applications Lecture 3: Currents cont d.; Two words about parabolic implosion Lecture 5.5: Quasi-hyperbolicity
More informationAPPENDIX A. Background Mathematics. A.1 Linear Algebra. Vector algebra. Let x denote the n-dimensional column vector with components x 1 x 2.
APPENDIX A Background Mathematics A. Linear Algebra A.. Vector algebra Let x denote the n-dimensional column vector with components 0 x x 2 B C @. A x n Definition 6 (scalar product). The scalar product
More informationChapter 5. Chapter 5 sections
1 / 43 sections Discrete univariate distributions: 5.2 Bernoulli and Binomial distributions Just skim 5.3 Hypergeometric distributions 5.4 Poisson distributions Just skim 5.5 Negative Binomial distributions
More informationCS598 Machine Learning in Computational Biology (Lecture 5: Matrix - part 2) Professor Jian Peng Teaching Assistant: Rongda Zhu
CS598 Machine Learning in Computational Biology (Lecture 5: Matrix - part 2) Professor Jian Peng Teaching Assistant: Rongda Zhu Feature engineering is hard 1. Extract informative features from domain knowledge
More informationOn Example of Inverse Problem of Variation Problem in Case of Columnar Joint
Original Paper Forma, 14, 213 219, 1999 On Example of Inverse Problem of Variation Problem in Case of Columnar Joint Yoshihiro SHIKATA Department of Mathematics, Meijo University, Tempaku-ku, Nagoya 468-8502,
More information1 + z 1 x (2x y)e x2 xy. xe x2 xy. x x3 e x, lim x log(x). (3 + i) 2 17i + 1. = 1 2e + e 2 = cosh(1) 1 + i, 2 + 3i, 13 exp i arctan
Complex Analysis I MT333P Problems/Homework Recommended Reading: Bak Newman: Complex Analysis Springer Conway: Functions of One Complex Variable Springer Ahlfors: Complex Analysis McGraw-Hill Jaenich:
More informationDe Finetti theorems for a Boolean analogue of easy quantum groups
De Finetti theorems for a Boolean analogue of easy quantum groups Tomohiro Hayase Graduate School of Mathematical Sciences, the University of Tokyo March, 2016 Free Probability and the Large N limit, V
More informationBessel s and legendre s equations
Chapter 12 Bessel s and legendre s equations 12.1 Introduction Many linear differential equations having variable coefficients cannot be solved by usual methods and we need to employ series solution method
More information1 Basics of vector space
Linear Algebra- Review And Beyond Lecture 1 In this lecture, we will talk about the most basic and important concept of linear algebra vector space. After the basics of vector space, I will introduce dual
More informationChapter 5 continued. Chapter 5 sections
Chapter 5 sections Discrete univariate distributions: 5.2 Bernoulli and Binomial distributions Just skim 5.3 Hypergeometric distributions 5.4 Poisson distributions Just skim 5.5 Negative Binomial distributions
More information100 CHAPTER 4. SYSTEMS AND ADAPTIVE STEP SIZE METHODS APPENDIX
100 CHAPTER 4. SYSTEMS AND ADAPTIVE STEP SIZE METHODS APPENDIX.1 Norms If we have an approximate solution at a given point and we want to calculate the absolute error, then we simply take the magnitude
More informationChapter 2. Discrete Distributions
Chapter. Discrete Distributions Objectives ˆ Basic Concepts & Epectations ˆ Binomial, Poisson, Geometric, Negative Binomial, and Hypergeometric Distributions ˆ Introduction to the Maimum Likelihood Estimation
More informationNotes, March 4, 2013, R. Dudley Maximum likelihood estimation: actual or supposed
18.466 Notes, March 4, 2013, R. Dudley Maximum likelihood estimation: actual or supposed 1. MLEs in exponential families Let f(x,θ) for x X and θ Θ be a likelihood function, that is, for present purposes,
More informationPhysics 110. Electricity and Magnetism. Professor Dine. Spring, Handout: Vectors and Tensors: Everything You Need to Know
Physics 110. Electricity and Magnetism. Professor Dine Spring, 2008. Handout: Vectors and Tensors: Everything You Need to Know What makes E&M hard, more than anything else, is the problem that the electric
More informationRandom Variables (Continuous Case)
Chapter 6 Random Variables (Continuous Case) Thus far, we have purposely limited our consideration to random variables whose ranges are countable, or discrete. The reason for that is that distributions
More informationX n D X lim n F n (x) = F (x) for all x C F. lim n F n(u) = F (u) for all u C F. (2)
14:17 11/16/2 TOPIC. Convergence in distribution and related notions. This section studies the notion of the so-called convergence in distribution of real random variables. This is the kind of convergence
More informationApproximating the matrix exponential of an advection-diffusion operator using the incomplete orthogonalization method
Approximating the matrix exponential of an advection-diffusion operator using the incomplete orthogonalization method Antti Koskela KTH Royal Institute of Technology, Lindstedtvägen 25, 10044 Stockholm,
More informationLast Update: March 1 2, 201 0
M ath 2 0 1 E S 1 W inter 2 0 1 0 Last Update: March 1 2, 201 0 S eries S olutions of Differential Equations Disclaimer: This lecture note tries to provide an alternative approach to the material in Sections
More informationKernel Method: Data Analysis with Positive Definite Kernels
Kernel Method: Data Analysis with Positive Definite Kernels 2. Positive Definite Kernel and Reproducing Kernel Hilbert Space Kenji Fukumizu The Institute of Statistical Mathematics. Graduate University
More informationHypergeometric series and the Riemann zeta function
ACTA ARITHMETICA LXXXII.2 (997) Hypergeometric series and the Riemann zeta function by Wenchang Chu (Roma) For infinite series related to the Riemann zeta function, De Doelder [4] established numerous
More informationSTAT 801: Mathematical Statistics. Distribution Theory
STAT 81: Mathematical Statistics Distribution Theory Basic Problem: Start with assumptions about f or CDF of random vector X (X 1,..., X p ). Define Y g(x 1,..., X p ) to be some function of X (usually
More information2 Functions of random variables
2 Functions of random variables A basic statistical model for sample data is a collection of random variables X 1,..., X n. The data are summarised in terms of certain sample statistics, calculated as
More informationSolving Dual Problems
Lecture 20 Solving Dual Problems We consider a constrained problem where, in addition to the constraint set X, there are also inequality and linear equality constraints. Specifically the minimization problem
More informationSometimes can find power series expansion of M X and read off the moments of X from the coefficients of t k /k!.
Moment Generating Functions Defn: The moment generating function of a real valued X is M X (t) = E(e tx ) defined for those real t for which the expected value is finite. Defn: The moment generating function
More informationELE539A: Optimization of Communication Systems Lecture 15: Semidefinite Programming, Detection and Estimation Applications
ELE539A: Optimization of Communication Systems Lecture 15: Semidefinite Programming, Detection and Estimation Applications Professor M. Chiang Electrical Engineering Department, Princeton University March
More informationNew Classes of Multivariate Survival Functions
Xiao Qin 2 Richard L. Smith 2 Ruoen Ren School of Economics and Management Beihang University Beijing, China 2 Department of Statistics and Operations Research University of North Carolina Chapel Hill,
More informationLecture 2: Vector Spaces, Metric Spaces
CCS Discrete II Professor: Padraic Bartlett Lecture 2: Vector Spaces, Metric Spaces Week 2 UCSB 2015 1 Vector Spaces, Informally The two vector spaces 1 you re probably the most used to working with, from
More informationMathematics Department Stanford University Math 61CM/DM Inner products
Mathematics Department Stanford University Math 61CM/DM Inner products Recall the definition of an inner product space; see Appendix A.8 of the textbook. Definition 1 An inner product space V is a vector
More informationConvex Optimization & Parsimony of L p-balls representation
Convex Optimization & Parsimony of L p -balls representation LAAS-CNRS and Institute of Mathematics, Toulouse, France IMA, January 2016 Motivation Unit balls associated with nonnegative homogeneous polynomials
More informationHypergeometric systems II: GKZ systems
Hypergeometric systems II: GKZ systems Uli Walther Aachen summer school, September 2007 Outline 1 Hypergeometric systems 2 Toric language 3 Solutions of A-hypergemetric systems The hypergeometric differential
More informationLecture Notes DRE 7007 Mathematics, PhD
Eivind Eriksen Lecture Notes DRE 7007 Mathematics, PhD August 21, 2012 BI Norwegian Business School Contents 1 Basic Notions.................................................. 1 1.1 Sets......................................................
More informationMultivariate Gaussian Analysis
BS2 Statistical Inference, Lecture 7, Hilary Term 2009 February 13, 2009 Marginal and conditional distributions For a positive definite covariance matrix Σ, the multivariate Gaussian distribution has density
More informationMatrix orthogonal polynomials and group representations
Matrix orthogonal polynomials and group representations Inés Pacharoni Universidad Nacional de Córdoba, Argentina (Joint work with Juan Tirao) Discrete Systems and Special Functions Cambridge, UK. July
More informationStochastic homogenization 1
Stochastic homogenization 1 Tuomo Kuusi University of Oulu August 13-17, 2018 Jyväskylä Summer School 1 Course material: S. Armstrong & T. Kuusi & J.-C. Mourrat : Quantitative stochastic homogenization
More informationReview (Probability & Linear Algebra)
Review (Probability & Linear Algebra) CE-725 : Statistical Pattern Recognition Sharif University of Technology Spring 2013 M. Soleymani Outline Axioms of probability theory Conditional probability, Joint
More informationDifferential Equations and Associators for Periods
Differential Equations and Associators for Periods Stephan Stieberger, MPP München Workshop on Geometry and Physics in memoriam of Ioannis Bakas November 2-25, 26 Schloß Ringberg, Tegernsee based on: St.St.,
More informationPreliminary Examination in Numerical Analysis
Department of Applied Mathematics Preliminary Examination in Numerical Analysis August 7, 06, 0 am pm. Submit solutions to four (and no more) of the following six problems. Show all your work, and justify
More informationMinimal basis for connected Markov chain over 3 3 K contingency tables with fixed two-dimensional marginals. Satoshi AOKI and Akimichi TAKEMURA
Minimal basis for connected Markov chain over 3 3 K contingency tables with fixed two-dimensional marginals Satoshi AOKI and Akimichi TAKEMURA Graduate School of Information Science and Technology University
More informationOn the Generalised Hermite Constants
On the Generalised Hermite Constants NTU SPMS-MAS Seminar Bertrand MEYER IMB Bordeaux Singapore, July 10th, 2009 B. Meyer (IMB) Hermite constants Jul 10th 2009 1 / 35 Outline 1 Introduction 2 The generalised
More informationStat260: Bayesian Modeling and Inference Lecture Date: February 10th, Jeffreys priors. exp 1 ) p 2
Stat260: Bayesian Modeling and Inference Lecture Date: February 10th, 2010 Jeffreys priors Lecturer: Michael I. Jordan Scribe: Timothy Hunter 1 Priors for the multivariate Gaussian Consider a multivariate
More informationMeasures and Jacobians of Singular Random Matrices. José A. Díaz-Garcia. Comunicación de CIMAT No. I-07-12/ (PE/CIMAT)
Measures and Jacobians of Singular Random Matrices José A. Díaz-Garcia Comunicación de CIMAT No. I-07-12/21.08.2007 (PE/CIMAT) Measures and Jacobians of singular random matrices José A. Díaz-García Universidad
More informationChapter One. The Calderón-Zygmund Theory I: Ellipticity
Chapter One The Calderón-Zygmund Theory I: Ellipticity Our story begins with a classical situation: convolution with homogeneous, Calderón- Zygmund ( kernels on R n. Let S n 1 R n denote the unit sphere
More informationOn the conservative multivariate Tukey-Kramer type procedures for multiple comparisons among mean vectors
On the conservative multivariate Tukey-Kramer type procedures for multiple comparisons among mean vectors Takashi Seo a, Takahiro Nishiyama b a Department of Mathematical Information Science, Tokyo University
More informationPade Approximations and the Transcendence
Pade Approximations and the Transcendence of π Ernie Croot March 9, 27 1 Introduction Lindemann proved the following theorem, which implies that π is transcendental: Theorem 1 Suppose that α 1,..., α k
More information6.4 Incomplete Beta Function, Student s Distribution, F-Distribution, Cumulative Binomial Distribution
6.4 Incomplete Beta Function, Student s Distribution, F-Distribution, Cumulative Binomial Distribution 29 CITED REFERENCES AND FURTHER READING: Stegun, I.A., and Zucker, R. 974, Journal of Research of
More informationHendrik De Bie. Hong Kong, March 2011
A Ghent University (joint work with B. Ørsted, P. Somberg and V. Soucek) Hong Kong, March 2011 A Classical FT New realizations of sl 2 in harmonic analysis A Outline Classical FT New realizations of sl
More information