Signal Processing and Linear Systems-B.P.Lathi copy

# 2 2t e e cos t 2e 2t t e t similarly we define

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 211 = 1, ID311 = 1, I Dul a nd IAI e - t +e- 2t O ne of t he c onditions necessary for a unique solution of Eq. (B.61) is t hat t he n umber of e quations m ust equal the number of unknowns. This implies t hat t he m atrix A m ust b e a s quare matrix. In addition, we observe from t he solution as given in Eq. (B.62) t hat if t he solution is t o exist, IAI "lO.t Therefore, t he inverse exists only for a s quare m atrix a nd only under t he condition t hat t he d eterminant of t he m atrix b e nonzero. A m atrix whose determinant is nonzero is a n onsingular m atrix. T hus, a n inverse exists only for a nonsingular (square) matrix. By definition, we h ave • 41 B.6 Vectors a nd Matrices = - 4. = 8, ID221 = - 1, ID321 = - 5, ID121 = -4 ID231 = - 1 ID331 = 3 ID131 j J e -2t dt A(t)dt= [ J et dt J sin dt J (e- t + 2 e- 2t ) dt We can readily prove t he following identities: Therefore, -4 A -1 = -~ 8 [ -4 • t These two c onditions imply t hat t he n umber of equations is equal t o t he n umber of unknowns a nd t hat all t he e quations a re independent. d dA dB (B.71a) ; U(A+B) = d i + d i d dA (B.71b) ;U(cA) = c di d ; U(AB) dA dB· . = d i B + A di = A B + A B (B.71c) 42 B ackground 43 B.6 Vectors a nd M atrices T he proofs of i dentities (B.71a) a nd (B.71b) are trivial. We c an prove Eq. (B.71c) as follows. Let A b e a n m x n m atrix a nd B a n n x p m atrix. T hen, if a ll - A C =AB (B.76b) =0 from Eq. (B.56), we have n C ik = L . .. j -1 a nd E quation (B.76a) [or (B.76b)] is known as t he c haracteristic e quation o f t he m atrix A a nd c an b e expressed as n C ik = L n a ijbjk j -1 + L a ijbjk (B.72) (B.77) j-1 ------- ------dik or a nn - A a ijbjk e ik E quation ( B.72) along w ith t he m ultiplication rule clearly indicate t hat d ik is t he i kth e lement o f m atrix AB a nd e ik is t he i kth element of m atrix AB. E quation (B.71c) t hen follows. I f we l et B = A -I i n Eq. (B.71c), we o btain Q(A) is called t he c haracteristic p olynomial o f t he m atrix A . T he n zeros of t he c haracteristic polynomial are t he eigenvalues of A and, corresponding t o each eigenvalue, t here is a n eigenvector t hat satisfies Eq. (B.74). T he C ayley-Hamilton t heorem s tates t hat every n X n m atrix A satisfies its own characteristic equation. I n o ther words, Eq. (B.77) is valid if A is r eplaced by A: Q (A) = A n + a n -1 A n -1 + ... + a lA + a oA 0 = 0 (B.78) Functions o f a Matrix B ut since T he C ayley-Hamilton t heorem c an be used t o e valuate functions of a square m atrix A , as shown below. Consider a function f(A) in t he form of a n infinite power series: dId - (AA- )=-1=0 dt dt we have (B.79) (B.73) Because A satisfies t he c haracteristic Eq. (B.77), we c an write 8 .6-4 The Characteristic Equation o f a Matrix: The Cayley-Hamilton Theorem For a n (n X n) s quare m atrix A , any vector x (x t 0) t hat satisfies t he e quation A X=AX (B.74) is a n e igenvector (or c haracteristic v ector), a nd A is t he c orresponding e igenvalue (or c haracteristic v alue) of A. E quation (B.74) can b e expressed as ( A - AI)x = 0...
View Full Document

## This note was uploaded on 04/14/2013 for the course ENG 350 taught by Professor Bayliss during the Spring '13 term at Northwestern.

Ask a homework question - tutors are online