Eigen-decomposition of a class of Infinite dimensional tridiagonal matrices G.V. Moustakides: Dept. of Computer Engineering, Univ. of Patras, Greece B.

Slides:



Advertisements
Similar presentations
Vector Spaces A set V is called a vector space over a set K denoted V(K) if is an Abelian group, is a field, and For every element vV and K there exists.
Advertisements

5.1 Real Vector Spaces.
Chapter 28 – Part II Matrix Operations. Gaussian elimination Gaussian elimination LU factorization LU factorization Gaussian elimination with partial.
Algebraic, transcendental (i.e., involving trigonometric and exponential functions), ordinary differential equations, or partial differential equations...
Lecture 17 Introduction to Eigenvalue Problems
Matrix Theory Background
Matrices & Systems of Linear Equations
Solution of linear system of equations
Principal Component Analysis
MECH300H Introduction to Finite Element Methods Lecture 2 Review.
Matrices and Systems of Equations
The Terms that You Have to Know! Basis, Linear independent, Orthogonal Column space, Row space, Rank Linear combination Linear transformation Inner product.
MOHAMMAD IMRAN DEPARTMENT OF APPLIED SCIENCES JAHANGIRABAD EDUCATIONAL GROUP OF INSTITUTES.
Digital Image Processing Final Project Compression Using DFT, DCT, Hadamard and SVD Transforms Zvi Devir and Assaf Eden.
Discrete Time Periodic Signals A discrete time signal x[n] is periodic with period N if and only if for all n. Definition: Meaning: a periodic signal keeps.
MATH 685/ CSI 700/ OR 682 Lecture Notes Lecture 6. Eigenvalue problems.
Copyright © Cengage Learning. All rights reserved. 7.6 The Inverse of a Square Matrix.
1 February 24 Matrices 3.2 Matrices; Row reduction Standard form of a set of linear equations: Chapter 3 Linear Algebra Matrix of coefficients: Augmented.
Eigenvalue Problems Solving linear systems Ax = b is one part of numerical linear algebra, and involves manipulating the rows of a matrix. The second main.
8.1 Vector spaces A set of vector is said to form a linear vector space V Chapter 8 Matrices and vector spaces.
Linear Algebra/Eigenvalues and eigenvectors. One mathematical tool, which has applications not only for Linear Algebra but for differential equations,
Lecture 22 MA471 Fall Advection Equation Recall the 2D advection equation: We will use a Runge-Kutta time integrator and spectral representation.
Eigenvalues and Eigenvectors
Matrices CHAPTER 8.1 ~ 8.8. Ch _2 Contents  8.1 Matrix Algebra 8.1 Matrix Algebra  8.2 Systems of Linear Algebra Equations 8.2 Systems of Linear.
Chapter 5 Eigenvalues and Eigenvectors 大葉大學 資訊工程系 黃鈴玲 Linear Algebra.
Fourier Analysis of Discrete Time Signals
The Fast Fourier Transform and Applications to Multiplication
Solving Linear Systems Solving linear systems Ax = b is one part of numerical linear algebra, and involves manipulating the rows of a matrix. Solving linear.
What is the determinant of What is the determinant of
1 Solving the algebraic equations A x = B =. 2 Direct solution x = A -1 B = = Applicable only to small problems For the vertical in the spectral technique.
KEY THEOREMS KEY IDEASKEY ALGORITHMS LINKED TO EXAMPLES next.
5.1 Eigenvectors and Eigenvalues 5. Eigenvalues and Eigenvectors.
1. Systems of Linear Equations and Matrices (8 Lectures) 1.1 Introduction to Systems of Linear Equations 1.2 Gaussian Elimination 1.3 Matrices and Matrix.
Advanced Computer Graphics Spring 2014 K. H. Ko School of Mechatronics Gwangju Institute of Science and Technology.
STROUD Worked examples and exercises are in the text Programme 5: Matrices MATRICES PROGRAMME 5.
Singular Value Decomposition and Numerical Rank. The SVD was established for real square matrices in the 1870’s by Beltrami & Jordan for complex square.
3.5 Perform Basic Matrix Operations Add Matrices Subtract Matrices Solve Matric equations for x and y.
STROUD Worked examples and exercises are in the text PROGRAMME 5 MATRICES.
Boot Camp in Linear Algebra TIM 209 Prof. Ram Akella.
Linear Algebra Engineering Mathematics-I. Linear Systems in Two Unknowns Engineering Mathematics-I.
Matrices. Variety of engineering problems lead to the need to solve systems of linear equations matrixcolumn vectors.
ALGEBRAIC EIGEN VALUE PROBLEMS
Copyright © Cengage Learning. All rights reserved. 8 Matrices and Determinants.
Chapter 5 Eigenvalues and Eigenvectors
MAT 322: LINEAR ALGEBRA.
CS479/679 Pattern Recognition Dr. George Bebis
Integral Transform Method
ECE 3301 General Electrical Engineering
Matrices and vector spaces
Section 4.1 Eigenvalues and Eigenvectors
Complex Eigenvalues Prepared by Vince Zaccone
Matrix Operations SpringSemester 2017.
CHE 391 T. F. Edgar Spring 2012.
Numerical Analysis Lecture 16.
Derivative of scalar forms
One dimensional Poisson equation
1.3 Vector Equations.
Chapter 3 Linear Algebra
Numerical Analysis Lecture10.
Symmetric Matrices and Quadratic Forms
Numerical Analysis Lecture 17.
Linear Algebra Lecture 32.
Chapter 8 The Discrete Fourier Transform
Chapter 8 The Discrete Fourier Transform
Linear Vector Space and Matrix Mechanics
Matrix Operations SpringSemester 2017.
Linear Algebra Lecture 35.
Eigenvalues and Eigenvectors
Linear Algebra: Matrix Eigenvalue Problems – Part 2
Linear Algebra Lecture 28.
Presentation transcript:

Eigen-decomposition of a class of Infinite dimensional tridiagonal matrices G.V. Moustakides: Dept. of Computer Engineering, Univ. of Patras, Greece B. Philippe: IRISA-INRIA, France

2 Outline  Definition of the problem.  From finite to infinite dimensions.  Reduction of the problem to a finite dimensional d.e. and eigen-decomposition problem, using Fourier transform.  Numerical aspects regarding the solution of the d.e. and the computation of the final (infinite dimensional) eigenvectors.  Conclusion.

3 Definition of the problem I:D:A:r:I:D:A:r: identity matrix diagonal matrix general matrix real scalar real matrices of dimensions N  N }

4 Eigen-decomposition Eigenvalues: Eigenvalues: There is an infinite number. Eigenvectors: Eigenvectors: There is an infinite number and each eigenvector is of infinite size. Goal: Goal: To reduce the infinite dimensional eigen-decomposition problem into a finite one.

5 From finite to infinite dimensions Q K has dimensions: (2K+1)N  (2K+1)N, therefore we have (2K+1)N eigenvalue-eigenvector pairs. Typical values: N = , K = 5-10.

6 Q K has dimensions: (2K+1)N  (2K+1)N  i (k) has dimensions: (2K+1)N  1. k = -K,…,K, i = 1,…,N.  i (k,l) has dimensions: N  1. k,l= -K,…,K, i=1,…,N.

7 Consider now the infinite dimensional problem by letting K   A  i (k,l+1) + (D+lrI)  i (k,l) + A t  i (k,l-1) = i (k)  i (k,l) A  i (k,l+1) + D  i (k,l) + A t  i (k,l-1) = ( i (k) -lr)  i (k,l)

8 Reduction to finite dimensions A  i (k,l+1)+D  i (k,l)+A t  i (k,l-1) = ( i (k)-lr)  i (k,l) A,D: N  N  i (k,l): N  1 i=1,…,N, k,l= - ,…,  Key Idea i (k) = i + kr without loss of generality assume 0  i  r  i (k,l) =  i (l-k) A  i (l-k+1)+D  i (l-k)+A t  i (l-k-1) = ( i -(l-k)r)  i (l-k) A  i (n+1)+(D- i I)  i (n)+A t  i (n-1) = -nr  i (n)

9 i, {  i (n)}, i=1,…,N, 0  i  r

10 Fourier Transform Let …, x(-2), x(-1), x(0), x(1), x(2),… be a real sequence. Then we define its Fourier Transform asImportant

11 A  i (n+1)+(D- i I)  i (n)+A t  i (n-1) = -rn  i (n)

12  i (  ) as being the Fourier transform of a (vector) sequence is necessarily periodic with period 2 . We need i and  i (0) to solve it.

13 Theorem Consider the following linear system of d.e. Let Z(  ) be the transition matrix of the d.e., that is then we know that X(  )= Z(  )X 0. The solution X(  ) is periodic if and only if X(2  )=X(0)

14

15 Steps to obtain ( i,{  i (n)}), i=1,…,N  Compute the transition matrix  (  ) from the d.e.  Find the eigenvalue-eigenvector pairs  i,  i (0) of  Form the desired eigenvalue-FT(eigenvector) pairs as  Use Inverse Fourier Transform to recover the final infinite eigenvector {  i (n)} from  i (  ).

16 Numerical aspects  Numerical solution of the d.e.  Eigen-decomposition of  (2  ).  Computation of the Inverse Fourier Transform of  i (  ) where

17 Numerical solution of the d.e. One can show that  (  ) is unitary, therefore any numerical solution should respect this structure. A possible scheme is

18 3 Step Integration. Yoshida scheme 1 Step Integration Pade 1 Pade 2

19 Pade 1, 1 step intgr. Pade 2, 1 step intgr. Pade 2, 3 step intgr. Pade 1, 3 step intgr.

20 Eigen-decomposition of  (2  ) Since  (2  ) is unitary there are special eigen-decomposition algorithms that require lower computational complexity than the corresponding algorithm for the general case. From this problem we obtain the pairs i,  i (0), i=1,…,N. Using the solution  (  ) of the differential equation we can compute the Discrete Fourier Transform of the eigenvectors Notice that we obtain a sampled version of the required Fourier transform.

21 Inverting the Fourier Transform Let …, x(-2), x(-1), x(0), x(1), x(2),… with Fourier Transform If x(n)=0 for n < 0 and n  M, then the Fourier Transform is equal then the finite sequence x(n), n =0,…, M-1, can be completely recovered from a sampled version of the Fourier transform. Specifically we need only the samples

22 Complexity O(M 2 ). For M=2 m popular Fast Fourier Transform (FFT). Complexity O(M log(M)). Apply Inverse Discrete Fourier Transform to  i (  n ), this will yield the desired vectors  i (n). If only a small number of  i (n) is significant, then we apply Inverse Discrete Fourier Transform only to a subset of the vectors  i (  n ) produced by the solution of the d.e. Inverce discrete Fourier Transform

23 ConclusionConclusion  We have presented as special infinite dimensional eigen- decomposition problem.  With the help of the Fourier Transform this problem was transformed into a d.e. followed by an eigen-decomposition both of finite size.  We presented numerical techniques that efficiently solve all subproblems of the proposed solution.

24 E n D Questions please ?