Electromechanical Motion Devices, Second Edition by Paul Krause, Oleg Wasynczuk and Steven Pekarek Copyright 2012 Institute

e of Electrical and Electronics Engineers, Inc.

Appendix B MATRIX ALGEBRA

Basic Definitions A rectangular array of numbers or functions an
A=

&i2 ' '

in

:
I ^ml

:
^m2

...
' * '

:
Q"mn \

t5"1)

is known as a matrix and is denoted in the text by capital boldface letters. The numbers or functions di j are the elements of the matrix and the subscript i denotes the row and j the column. A matrix with m rows and n columns is of order (m, n) or an m x n (m by n) matrix. If m = n, the matrix is a square matrix. If a matrix is an m x 1 matrix, it is a column vector. If it is a 1 x n matrix, it is a row vector. Generally, lower-case boldface letters are used to denote column or row vectors. A square matrix in which all elements are zero except those on the main diagonal, a n , a 2 2 r ianm is a diagonal matrix. If all elements of a diagonal matrix are unity, then the matrix is the identity matrix and is denoted as I. When aij = aji, the matrix is called a symmetrical matrix. A null or zero matrix is one in which all elements are zero. Addition and Subtraction Two matrices can be added or subtracted only if they are of the same order. Thus, if A has elements a^ and B has elements 6y, then, if 481

482 C = A+ B C has elements Q,-, where

Cij = aij + bij

APPENDIX B

(B-2)

(B-3)

or if C = A - B then
C{j Q*ij vij \D-0)

(B-4)

Also, addition is commutative, A+ B - B+ A and associative, (A + B) + C = A + (B + C) Obviously, A+ O = A where O is the zero matrix. Multiplication If the matrix A is multiplied by a scalar, every element of the matrix is multiplied by the scalar. For example, kA means all elements of A are multiplied by the constant k; tA means that all elements of A are multiplied by time t To multiply two matrices, say A B , it is necessary that the number of columns of A equal the numbers of rows of B. If A is of order raxn and B of order nxp, then the order of A B is mxp. The elements of C = AB (B-9) are obtained by multiplying the elements by the zth row of A by the corresponding elements of the j t h column of B and adding these products. In particular, Cij = anbij + ai2b2j + . . . + ainbnj (B-10) In (B-9), A is said to premultiply B, whereas B is said to postmultiply A. Let (B-8) (B-7) (B-6)

MATRIX ALGEBRA

483 12 31 (B-ll)

4 5 6J
and B Then AB = However, BA = I" 11 - 2 1 1 [ 17 - 4 8 1 39 - 5 4 49 68 44 - 5 5 AB ^ BA -69 87 --66 -7 -8 " 9 10 0 -11 _

(B-12)

(B-13)

(B-14)

We see that, in general, matrix multiplication is not commutative; thus, (B-15) Multiplying a matrix of mxn by a column vector of n x l yields a column vector of m x l . Multiplication of a row vector of l x n and a column vector of n x l yields a function (scalar) that is the sum of the product of specific elements of each vector. Multiplying the identity matrix by A yields A, that is, AI - IA = A We can show that
A{tHJ) = {Ati)*J [O-U)

(B-16)

A(B + C) = A B + A C (B + C)A = B A + C A inally, consider the simultaneous linear equations bx + 3y-2z = 14

(B-18) (B-19) (B-20) (B-21)

x + y Az = 7

484 6x + 3z 1 Let us write the above equations in the form Ax = b Here A = 5 3-2 1 1 -4 6 0 3 x
y

APPENDIX B

(B-22) (B-23)

(B-24)

(B-25)

b =

14 -7 1

(B-26)

In this case, A is called the coefficient matrix. Care must be taken not to confuse the column vector x and the variable x. Transpose The transpose of a matrix A is denoted as A T . The transpose of A is obtained by interchanging the rows and columns of A. Thus, if A = 12 3 4 5 6 7 8 (B-27)

AT

ri

2 6 3 7 4 8J

51

(B-28)

The transpose possesses the following properties: (A T ) T = A (A + B + C ) T = A T + B r + C T (AB) T = B T A T (B-29) (B-30) (B-31)

MATRIX ALGEBRA

485 (ABC) T = C T B r A a (B-32)

Partitioning Partitioning of matrices is used throughout the text. It is helpful in matrix multiplication. For example, let A and B be partitioned as

A= B=

r c D1 [E FJ

(B-33) (B-34)

where C through K are submatrices. The product of A B is AB = CG + DJ E G + FJ CH + DK EH + FK (B-35)

[ JG H 1 K [

Determinants Every square matrix has a scalar associated with it called its In particular, if A is a square matrix, say, A =
on
21 31 12 22 32 13 23 33

determinant

(B-36)

then the determinant of A is denoted det A or |A| detA =

11 021 31 12 22 32 13 23 33

(B-37)

It is important to note the difference between (B-36) and (B-37); (B-36) represents a matrix that is a square array of elements, whereas (B-37) represents a scalar associated with the matrix A. The det A is determined by obtaining the minors and cofactors. Given a matrix A, a minor is the determinant of any square submatrix of A. The cofactor of the element CLI j is a scalar obtained by multiplying (l)l+J times the minor obtained from A by removing the zth row and j t h column. To find the determinant of the square matrix A: 1. Pick any one row or any one column of the matrix. 2. For each element in the row or column chosen, find its cofactor.

486

APPENDIX B

3. Multiply each element in the row or column chosen by its cofactor and sum the results. The sum is the determinant of the matrix. For example, find det A, where A = 3 5 0 -1 1 1 3-6 4 (B-38)

Expanding in the second column, detA = ( 5 ) ( - l ) 1 + 2

3 0 3 0 -1 1 3+2 2+2 3 4 + (-6)(-l) -1 1 3 4 + (1)(-1)

= ( - 5 ) ( - 7 ) + (l)(12) + (6)(3) = 65 (B-39) Adjoint The adjoint matrix of a square matrix A, denoted adjoint A or A a , is formed by replacing each element a^ by the cofactor a^ and transposing. Thus, the adjoint of (B-36) is adjoint A =
an a2l #31 au #22 a32

#13 #23 #33 #31 #32 #33

#11 #12 #13

#21 #22 #23

(B-40)

Inverse The inverse of a square matrix A is written as A - 1 and is defined as A" X A = A A " 1 = I (B-41) In the text, parentheses are used to avoid confusion with superscripts, that is, the inverse is denoted ( A ) - 1 . The inverse is defined only for square matrices. In particular, A-* = adjoint A det A (B-42)

MATRIX ALGEBRA

487

If det A is zero, A does not possess an inverse and is said to be singular. Consider a 2x2 matrix: A = adjoint A =
an
&21

&i2
a

22

(B-43) (B-44) (B-45)

-a2i

&22

12

an _

det A = ana 2 2 - ^12^21 Find the inverse of (B-38). The cofactor of an is 1 1 = (3) (10) = 30 -6 4

a n = (3X-1)1*1

(B-46)

Finding the cofactor of each element and transposing yields the adjoint of A. Thus, adjoint A = 30 20 15 35 12 18 0 33 32 (B-47)

The det A is given by (B-39), hence, A-* = adjoint A det A 1 65 " 30 20 15 " 35 12 18 0 33 32 (B-48)

Derivatives The derivative of the matrix A, denoted (d/dt) A or pA, is the derivative of each element of the matrix. The derivative of A, given by (B-l), is
pan pa2\ panl pa12 pa 2 2 pan2 pain pa2n pann

pA =

(B-49)

where p is the operator d/dt.

488 M a t r i x Formulation In the text, we deal with equations of the form A _ 1 y = A - V l x + p[A _1 z]

APPENDIX B

(B-50)

where A is a square, nonsingular matrix, the elements of which may be functions of time, and p is the operator d/dt. Solving (B-50) for y is achieved by premultiplying by A. Thus, A A - 1 y = A A V l x + Ap[A _1 z] = r A A _ 1 x + AfpA-^z + A A _ 1 [pz] Since A A _ 1 = I, (B-51) may be written as y = r l x + A[pA _1 ]z + pz (B-52) (B-51)