next up previous


Powers of a Matrix and Inverse Matrix


Proposition 7.1.1. If tex2html_wrap_inline9653 and tex2html_wrap_inline9655 then I-F is a regular matrix and
displaymath9629
while

displaymath9630

Proof. If we suppose the contrary to the assertion that the matrix I-F is singular, then there exists such a nonzero vector tex2html_wrap_inline7157 that tex2html_wrap_inline9663 i.e., tex2html_wrap_inline9665 and tex2html_wrap_inline9667, tex2html_wrap_inline9669 Hence the matrix I-F is a regular matrix. To find the matrix (I-F)-1, we consider the identity
displaymath9631
Since
displaymath9632
then
displaymath9633
that implies
displaymath9634
and
displaymath9635
which was to be proved. tex2html_wrap_inline7949

Proposition 7.1.2. Let QHAQ=T=D+N be the Schur factorization of the matrix tex2html_wrap_inline8415, while D is a diagonal matrix and N is a strictly upper triangular matrix (on the leading diagonal there are zeros). Let tex2html_wrap_inline9685 and tex2html_wrap_inline9687 be respectively the greatest and the least by modulus eigenvalues of the matrix A. If tex2html_wrap_inline9691 then for all tex2html_wrap_inline9693
displaymath9636
If A is a regular matrix and the number tex2html_wrap_inline9697 is such that
displaymath9637
then for all tex2html_wrap_inline9693

displaymath9638

Proof. See Golub, Loan (1996, pp 336-337). tex2html_wrap_inline7853

The formula
displaymath9639
which is easily checked shows how the inverse matrix changes when the matrix A is substituted by the matrix B. The modification of this formula is a formula of Sherman-Morrison-Woodbury given in the following proposition.

Proposition 7.1.3. If tex2html_wrap_inline7067 and tex2html_wrap_inline9709 while matrices A and I+VTA-1U are regular,
displaymath9640

Proof. See Golub, Loan (1996, pp. 50). tex2html_wrap_inline7853