23 Matrix Inverses：23逆.pptVIP

8.2 Orthogonal Diagonalization Theorem 1 The following conditions are equivalent for an (n x n) matrix P: 1. P is invertible and P-1 = PT 2. The rows of P are orthonormal. 3. The columns of P are orthonormal. Proof: (1) ?(2) Given P-1 = PT, PPT = I Let X1,…,Xn be rows of P. Then XjT is jth col of PT So (i,j) entry of PPT is Xi?Xj Since PPT=0, we know that all (i,j) entries where i≠j will be 0, so Xi?Xj=0 (rows are orthog) if i≠j and also Xi?Xj=1 if i=j (rows are orthonormal) (1/3 equiv is same idea) Definition An (n x n) mtx P is called an orthogonal matrix if it satisfies one (and so all) of the conditions in Theorem 1. Example Is the following matrix orthogonal for any angle? Show rows are orthonormal. Example If P and Q are orthogonal matrices, show that PQ is also orthogonal, and that P-1=PT is orthogonal. Solution: P and Q are invertible, so PQ is also (PQ)-1 = Q-1P-1=QTPT = (PQ)T so PQ is orthogonal. Also (P-1)-1 = P = (PT)T = (P-1)T so P-1 is orthogonal. Definition An (n x n) matrix A is orthogonally diagonalizable when an orthogonal matrix P can be found such that P-1AP=PTAP is diagonal. (We will see that this only happens with symmetric matrices.) Theorem 2-Principal Axes Theoremor Real Spectral Theorem The following conditions are equivalent for A (n x n): 1. A has an orthonormal set of n eigenvectors. (w/ eigenvalues that we call the spectrum of the matrix. 2. A is orthogonally diagonalizable. 3. A is symmetric. Proof: (1)?(2) P = [X1 … Xn] (n x n) P is orthog iff {X1,…,Xn} is orthonormal set in ?n (by thm1) P-1AP is diag iff {X1,…,Xn} consists of the eigenvectors of A (from 6.2) Proof-cont. (2)?(3) If PTAP = D is diagonal where PT=P-1(given from 2), then A = PDP-1=PDPT So, AT = (PDPT)T=PTTDTPT=PDTPT=PDPT = A (since D diag) So A is symmetric. (3)?(2) very tough to prove -- omitted Definition Principal Axes: A set of orthonormal eigenvectors of a symmetric matrix A. Example Find an ortho