cauchys interlace theorem and lower bounds for the spectral radius柯西交错的定理和下界的谱半径.pdfVIP

Internat. J. Math. Math. Sci. Vol. 23, No. 8 (2000) 563–566 S016117120000257X ©Hindawi Publishing Corp. CAUCHY’S INTERLACE THEOREM AND LOWER BOUNDS FOR THE SPECTRAL RADIUS A. MCD. MERCER and PETER R. MERCER (Received 31 December 1998) Abstract. We present a short and simple proof of the well-known Cauchy interlace theo- rem. We use the theorem to improve some lower bound estimates for the spectral radius of a real symmetric matrix. Keywords and phrases. Eigenvalues, symmetric matrix. 2000 Mathematics Subject Classification. Primary 15A42. 1. Cauchy’s interlace theorem. We begin by presenting a short and simple proof of the Cauchy interlace theorem, which we believe to be new. See [1, 3, 4, 5], for example, for several other proofs. The theorem states that if a row-column pair is deleted from a real symmetric matrix, then the eigenvalues of the resulting matrix interlace those of the original one. Let A be a real symmetric n ×n matrix with eigenvalues (assumed distinct for now) λ1 λ2 ··· λn (1.1) and normalized eigenvectors v1,v2 ,...,vn . (1.2) Let A1 be the matrix obtained from A by deleting the first row and column. We list the eigenvalues of A1 via µ1 ≤ µ2 ≤ ··· ≤ µn−1. Set D(λ) := det(A −λI), D1 (λ) := det(A1 −λI), (1.3) e := [1,0,0,...,0]T , x := [x ,x ,...,x ]T . (1.4) 1 2 n Applying Cramer’s rule to the set of equations (A −λI)x = e yields D1 (λ) x1 =