4.II.39C

Numerical Analysis
Part II, 2007

(a) Suppose that AA is a real n×nn \times n matrix, and that wRnw \in \mathbb{R}^{n} and λ1R\lambda_{1} \in \mathbb{R} are given so that Aw=λ1wA w=\lambda_{1} w. Further, let SS be a non-singular matrix such that Sw=ce(1)S w=c e^{(1)}, where e(1)e^{(1)} is the first coordinate vector and c0c \neq 0. Let A^=SAS1\widehat{A}=S A S^{-1}. Prove that the eigenvalues of AA are λ1\lambda_{1} together with the eigenvalues of the bottom right (n1)×(n1)(n-1) \times(n-1) submatrix of A^.\widehat{A} .

(b) Suppose again that AA is a real n×nn \times n matrix, and that two linearly independent vectors v,wRnv, w \in \mathbb{R}^{n} are given such that the linear subspace L{v,w}L\{v, w\} spanned by vv and ww is invariant under the action of AA, i.e.,

xL{v,w}AxL{v,w}x \in L\{v, w\} \quad \Rightarrow \quad A x \in L\{v, w\}

Denote by VV an n×2n \times 2 matrix whose two columns are the vectors vv and ww, and let SS be a non-singular matrix such that R=SVR=S V is upper triangular, that is,

R=SV=S×[v1w1v2w2v3w3::vnwn]=[r11r120r2200::00]R=S V=S \times\left[\begin{array}{cc} v_{1} & w_{1} \\ v_{2} & w_{2} \\ v_{3} & w_{3} \\ : & : \\ v_{n} & w_{n} \end{array}\right]=\left[\begin{array}{cc} r_{11} & r_{12} \\ 0 & r_{22} \\ 0 & 0 \\ : & : \\ 0 & 0 \end{array}\right]

Again let A^=SAS1\widehat{A}=S A S^{-1}. Prove that the eigenvalues of AA are the eigenvalues of the top left 2×22 \times 2 submatrix of A^\widehat{A}together with the eigenvalues of the bottom right (n2)×(n2)(n-2) \times(n-2) submatrix of A^\widehat{A}