City Suburbs. : population distribution after m years

Transcription

1 Section 5.3 Diagonalization of Matrices Definition Example: stochastic matrix To City Suburbs From City Suburbs = A City Suburbs probability matrix of a sample person s residence movement p = Ap = A m p : current population of the city and suburbs : population distribution in the next year : population distribution after m years

2 A = PDP 1 where P = ,D = A 3 =(PDP 1 )(PDP 1 )(PDP 1 )=PD 3 P 1 A m = PD m P (.82) = m (.82) m 1 (.82) m = (.82) m 5+(.82) m Question

3 Not all matrices are diagonalizable. A = A 2 = 0. If A = PDP -1 for some invertible P and diagonal D, then A 2 = PD 2 P -1 = 0. D 2 = 0 D = 0 since D is diagonal A = 0. Ö Question Let s observe some examples before answering this question.

4 A = Characteristic polynomial Eigenvalues Eigenspaces B = C = I =

5 A = Characteristic polynomial 3 5 Eigenvalues Eigenspaces B = C =

6 Theorem 5.2

7 Proof A R n n is diagonalizable. an invertible P = [ p 1 p 2 p n ] R n n and a diagonal D = diag[d 1 d 2 d n ]. R n n such that A = PDP -1. AP = PD and P = [ p 1 p 2 p n ] is invertible. [ Ap 1 Ap 2 Ap n ] = Pdiag[d 1 d 2 d n ]. = P d 1 e 1 d 2 e 2 d n e n = P (d 1 e 1 ) P (d 2 e 2 ) P (d n e n ) = d 1 (P e 1 ) d 2 (P e 2 ) d n (P e n ) = d 1 p 1 d 2 p 2 d n p n. and P = [ p 1 p 2 p n ] is invertible. Ap i = d i p i for i = 1, 2,, n and {p 1, p 2,, p n } is L.I.. p i is an eigenvector of A corresponding to the eigenvalue d i and B = {p 1, p 2,, p n } is a basis for R n.

8 Steps to diagonalize a given A R n n : 1. Find n eigenvalues (repeated or not) for A and form a diagonal matrix D with eigenvalues on the diagonal; det (A eigenvalues of A = { 0.82, 1 }, and 2. Find n L.I. eigenvectors corresponding to these eigenvalues, if possible, and form an invertible P R n n ; reduced row p = echelon form A.82I 2 P = A I reduced row p = echelon form A = PDP -1. ti 2 )=det A = t t =(t.82)(t 1) D = invertible

9 Every A R n n has n eigenvalues (counting repeated ones) if complex eigenvalues are allowed. However, some matrices may not have n L.I. eigenvectors even if complex eigenvectors are allowed.

10 Theorem 5.3

11 Theorem 5.3 Proof Let A be an n n matrix with eigenvectors v 1, v 2,, v m having corresponding distinct eigenvalues λ 1, λ 2,, λ m. Suppose the set of eigenvectors are L.D.. As eigenvectors are nonzero, Theorem 1.9 implies v k = c 1 v 1 +c 2 v 2 + +c k-1 v k-1 for some k [2, m] and scalars c 1, c 2,, c k-1. Av k = c 1 Av 1 +c 2 Av 2 + +c k-1 Av k-1 λ k v k = c 1 λ 1 v 1 +c 2 λ 2 v 2 + +c k-1 λ k-1 v k-1 (-λ k ) 0 = c 1 (λ 1 -λ k ) v 1 +c 2 (λ 2 -λ k )v 2 + +c k-1 (λ k-1 -λ k )v k-1 c 1 = c 2 = = c k-1 = 0 v k = 0, a contradiction.

12 Corollary 1: Let S 1, S 2,, S p be subsets of p eigenspaces of a square matrix corresponding to p distinct eigenvalues. If S i is L.I. for all i = 1, 2,, p, then the set S 1 S 2 S p is L.I.. Corollary 2: If A R n n has n distinct eigenvalues, then R n has a basis consisting of eigenvectors of A, i.e., A is diagonalizable.

13 Definition For A 2 R n n, the characteristic polynomial of A may be factored into a product of linear factors if det(a ti n )=( 1) n (t 1)(t 2) (t n). Here, i,i=1, 2,...,n do not have to be distinct, but i 2R. Test for a Diagonalizable Matrix An n x n matrix A is diagonalizable if and only if both the following conditions are met. 1) The characteristic polynomial of A factors into a product of linear factors. 2) For each eigenvalue λ of A, the multiplicity of λ equals the dimension of the corresponding eigenspace (n rank(a-λi n )).

14 Proof if Follow the previous diagonalization stepts. Condition (1) there are n eigenvalues for Step 1. Condition (2) and Theorem 5.3 there are n L.I. eigenvectors in Step 2. only if If Condition (1) fails, then A has less than n eigenvalues (counting repeated ones), and (sum of all geometric multiplicities) (sum of all algebraic multiplicities) < n, which means that there are no enough L.I. eigenvectors to form a basis. If Condition (2) fails, then there are no enough L.I. eigenvectors to form a basis. Note: Condition (1) always holds if complex eigenvalues are allowed.

15 Example: The characteristic polynomial of A = is -(t + 1) 2 (t - 3) eigenvalues: 3, -1, -1 the eigenspaces corresponding to the eigenvalue 3 and -1 have bases B 1 = and A = PDP -1, where P = B 2 = and 1 0 0, 1, respectively D =

16 Example: The characteristic polynomials of A = B = C = M = are -(t+1)(t 2 +4), -(t+1)(t+4) 2, -(t+1)(t+4) 2, -(t+1)(t 2-4), respectively

17 Example: The characteristic polynomials of A = C = B = M = are -(t+1)(t 2 +4), -(t+1)(t+4) 2, -(t+1)(t+4) 2, -(t+1)(t 2-4), respectively A can not be diagonalized as it has only one (real) eigenvalue B can be diagonalized as nullity of B - (-4)I 3 is 2 C can be not diagonalized as nullity of C - (-4)I 3 is 1 M can be diagonalized as M has three distinct eigenvalues -1, -2, 2

18 Homework Set for Section 5.3 Section 5.3: Problems 1, 3, 5, 9, 13, 17, 29, 31, 33, 35, 41, 43, 47