Scribd
Upload
72 views

Mini-Introduction To Matrix Algebra

This document provides a mini-introduction to matrix algebra, covering basic concepts like vectors, matrices, and operations. It discusses determinants and leading principal minors, and their applications in determining linear independence, concavity/convexity of functions, and definiteness of Hessian matrices. Determinants are used to check if a matrix is full rank, while the signs of leading principal minors indicate concavity or convexity of functions. Examples are provided to demonstrate calculating determinants, minors, and checking concavity.

Uploaded by

YHTRTR
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
72 views

Mini-Introduction To Matrix Algebra

This document provides a mini-introduction to matrix algebra, covering basic concepts like vectors, matrices, and operations. It discusses determinants and leading principal minors, and their applications in determining linear independence, concavity/convexity of functions, and definiteness of Hessian matrices. Determinants are used to check if a matrix is full rank, while the signs of leading principal minors indicate concavity or convexity of functions. Examples are provided to demonstrate calculating determinants, minors, and checking concavity.

Uploaded by

YHTRTR
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

Mini-Introduction to Matrix Algebra

The Plan Basic concepts in MA: vectors, matrices, operations application: norm & projection Determinants and Leading Principle Minors application: linear independence application: Hessian matrices and deniteness (concavity, convexity) of functions (SOC)

c Ronald Wendner

Matrix Algebra-I-1

v2.0

Basic Concepts Matrix: rectangular array of terms numbers, functions, variables, other matrices, etc.

a11 a A = 21 an1 Dimension

a12 a22 an2

a1k a2k ank

n k (n rows, k cols) k = 1 (column vector, x) n = 1 (row vector, x )

c Ronald Wendner

Matrix Algebra-I-2

v2.0

Special Matrices n = k (square) n = k and aij = aji (symmetric) aij = 0 (nullmatrix) aij = 0 (for i = j ) and aii = 1 (identity matrix, I )

Operations Operations involving scalars, vectors and matrices (A, B ) addition (subtraction) multiplication division (inversion) conformability requirements: operations are well dened only if dimensions of A, B t addition (subtraction) conformability: dimensions A, B equal! C = A B , cij = aij bij
c Ronald Wendner Matrix Algebra-I-3 v2.0

A+B =B+A A0=A Examples: (i) A I (ii) Give graphical interpret. of vector addition (subtraction)

multiplication: A B , where A lead matrix, B lag matrix B is premultiplied by A A is postmultiplied by B conformability: #columns of A = #rows of B Ank Bkl = Cnl multiplication: row of A with column of B 2 4 1 0 2 Examples. A = , B = 6 0 ; 3 1 1 1 3 A B =?, B A =?, A B =? Can we multiply a vector by a matrix?
c Ronald Wendner Matrix Algebra-I-4 v2.0

AB =BA A 0 = 0, A I = A, I A=A

scalar multiplication cA multiply every element of A by scalar c a11 a12 c a11 c a12 A= , then: c A = a21 a22 c a21 c a22

vector multiplication: conformability: same dimension product of two vectors can be a matrix (scalar) x1n, yn1 outer product: y x = A?? inner product, scalar product, dot product x y = A??

If x y = 0, then x, y are orthogonal (perpendicular) vectors. E.g., x = (2, 0), y = (0, 1).
c Ronald Wendner Matrix Algebra-I-5 v2.0

Application: Norm & Projection vector (graphically) origin (often origin of coordinate system) direction length typically all info given by endpoint; e.g., x = (1, 1)

norm (length / distance from origin) of a vector ||x|| x x x = (1, 2), then ||x|| = 5 projection of y on x: generally: t x (t scalar) t =? (y t x) t x = 0. Why? t=
y x xx

Example. Calculate projection of y = (1, 1) on x = (2, 0)


c Ronald Wendner Matrix Algebra-I-6 v2.0

Determinants and Leading Principle Minors determinant is a scalar linear independence SOC in (constrained) optimization concavity/convexity of functions quasiconcavity/quasiconvexity of functions uniqueness of optimum in optimization dynamics of dynamic systems (e.g., uctuations vs. monotone transitions, stability vs. instability, etc.) equation solving (Cramers rule) determinant of A22: |A| = a11 a22 a12 a21 determinant of A33 (and higher order): Laplace expansion sum of 3 determinants of 2 2 submatrices det. of submatrices: minors Mij e.g.: M11 = a22a33 a32a23, M12 =?, M13 =? a11 a12 a13. .a11 a12 a13. .a11 a12 a13. a21 a22 a23 ; a21 a22 a23 ; a21 a22 a23 ; a31 a32 a33 a31 a32 a33 a31 a32 a33

c Ronald Wendner

Matrix Algebra-I-7

v2.0

|A| = a11 M11 a12 M12 + a13 M13. 1 0 2 Example. Calculate |B |, for B = 0 1 1 1 1 0 leading principal minors of A

a11 a A = 21 ak 1

a12 a22 ak 2

a1k a2k akk

M1 = a11, M2 = a11 a Mk = 21 ak1 a12 a22 ak2

a11 a12 , M3 = a21 a22 a1k a2k = |A|. akk

a11 a12 a13 a21 a22 a23 , ..., a31 a32 a33

Example. Calculate the leading principle minors of B .

c Ronald Wendner

Matrix Algebra-I-8

v2.0

Applications linear independence 6 12 2 4

A=

|A| = 0 ! rank of matrix = # of linearly indept. rows (columns) matrix of full rank |A| = 0

concavity (convexity) of a function f (x1, ..., xn) signs of leading principal minors of Hessian matrix of f (x) f11 f H = 21 f n1 f12 f22 fn2 f1n f2n fnn

c Ronald Wendner

Matrix Algebra-I-9

v2.0

concavity: M1 0, M2 0, M3 0, ... strict concavity: M1 < 0, M2 > 0, M3 < 0, ... convexity: all Mi 0, i = 1, ..., n strict convexity: all Mi > 0, i = 1, ..., n

Examples. Are the following functions concave (convex)?


2 f (x1, x2) = x2 1 + x2 , 2 f (x1, x2) = x2 1 x2 .

Consider u(x1, x2) = ln(x1) + ln(x2), where x1 > 0, x2 > 0. For which values of (, ) is u(x1, x2) strictly concave?

Comments (i) deniteness of H and concavity (convexity) of f (x) (ii) SOC in unconstrained optimization problems
c Ronald Wendner Matrix Algebra-I-10 v2.0

You might also like