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

# 79 b73 because a satisfies t he c haracteristic eq b77

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: or I f we mUltiply b oth sides by A, t he l eft-hand side is An +l, a nd t he r ight-hand side contains t he t erms An, A n-I, . .. , A. Using Eq. (B.80), i f we s ubstitute An in terms of An -1, An -2, . .. , A, t he h ighest power on t he r ight-hand side is reduced t o n - l. C ontinuing in this way, we see t hat An +k c an b e e xpressed in terms of A n-I, A n-2, . .. , A for a ny k. Hence, t he infinite series on t he r ight-hand s ide of Eq. (B.79) can always b e expressed in t erms of An -1, An -2, . .. , A as (B.75) (B.81) T he s olution f or t his set of homogeneous equations exists if a nd only if IA - >.11 = 1>.1 - AI = 0 (B.80) (B.76a) I f we assume t hat t here a re n d istinct eigenvalues A I, A2, . .. , An, t hen Eq. (B.81) holds for these n values of A. T he s ubstitution o f these values in Eq. (B.81) yields n s imultaneous equations 44 Background • Al At An - 1 [30 A2 1 f (>'l) A~ An - 1 45 B.6 Vectors a nd M atrices E xample B .l3 Let us consider the case where [31 f (A2) 1 2 (B.82a) . ..................... 1 / (An) . .. A2 n An n An - 1 [3n-1 The eigenvalues are a nd [30 [31 Al A2 1 An - 1 - 1 1 /(A1) A2 1 A2 2 An - 1 /(A2) 2 (B.82b) Hence, A1 = - 1, >'2 = - 2, and . ..................... [3n-1 1 An A2 n ... n An - 1 / (An) in which Since A also satisfies Eq. (B.80), we m ay advance a similar argument t o show t hat if / (A) is a f unction o f a s quare m atrix A expressed as a n infinite power series in A , t hen [30] = [ (31 00 f (A) = a ol + a1A + a 2A2 + ... + ... = L :aiAi = t -1 1 (B.83a) i =O [1 -1] [e[2 -1] [e1 a nd -1 e - 2t -2 t e - 2t [2e- et = 2t _ e-t _ e - 2t (B.83b) and in which t he coefficients [3is a re found from Eq. (B.82b). I f some of t he eigenvalues are r epeated ( multiple roots), t he r esults are somewhat modified. We shall d emonstrate t he u tility of t his r esult with the following two examples. B.6-5 2 e - t - e - 2t Computation o f an Exponential and a Power o f a Matrix = [ _ 2e- t L et us c ompute e At defined by (B.84) + 2 e- 2t • Computation o f A k As Eq. (B.83b) indicates, we c an express A k as From Eq. (B.83b), we c an express n -1 e At = A k = [30 1 + [31 A L: [3i(A)i i =l in which t he !3is a re given by Eq. (B.82b), w ith / (Ai) = e A". + ., . + [3n-1 A n -1 in which t he [3iS a re given by Eq. (B.82b) with / (Ai) = A~. F or a completed exampl, of t he c omputation of A k b y t his m ethod, see Example 13.12. 46 Background B.7 Miscellaneous 47 B.7 Miscellaneous B.7-5 Complex Numbers B.1-1 L'Hopital's Rule I f l imf(x)/g(x) r esults in t he i ndeterministic form % or 00/00, t hen n even n odd , f (x) - I' j (x) I1m g{x) - 1m g{x) B.7-2 e ±j9 a +jb=re j9 The Taylor and Maclaurin Series f {x) = f {a) f {x) = f(O) B.7-3 (x - a) , I! { re j9 )k (x - a)2 .. 2! + - - f ( a ) + - - f ( a ) + ... x2 x. = cos IJ ± j s in IJ r =v'a 2 +b 2 , = r k e jk9 ( rl ej9, ) (r2 ej92 ) + -! f{O) + -2! f{O) + ... I B. 7-6 Trigonometric Identities Power Series eX x2 2! x3 3! e ±jx xn n! + - + - +...
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