Author |
Topic: Re: eigenvalues property ? (Read 2217 times) |
|
MonicaMath
Newbie
Gender:
Posts: 43
|
|
Re: eigenvalues property ?
« on: Mar 28th, 2009, 8:29pm » |
Quote Modify
|
maybe you can start with: since r is not an eigenvalue for A then det(r*I - A) not zero, so (A-r*I) is nonsingular, and use the fact that Bx=r*x, then show that || inv(r*I - A) (B-A) ||<1 to obtain a contradiction.
|
|
IP Logged |
|
|
|
trusure
Newbie
Gender:
Posts: 24
|
|
Re: eigenvalues property ?
« Reply #1 on: Mar 28th, 2009, 9:33pm » |
Quote Modify
|
I tried your method but there is no result ?! can anyone give me a hint !
|
|
IP Logged |
|
|
|
Eigenray
wu::riddles Moderator Uberpuzzler
Gender:
Posts: 1948
|
|
Re: eigenvalues property ?
« Reply #2 on: Mar 29th, 2009, 5:31pm » |
Quote Modify
|
Well that's a funny problem. Note that (r I - A)-1(B-A) = (r I - A)-1( (r I - A) + (B - r I) ) = I + C, where C = (r I - A)-1(B - r I) is a singular matrix. Do you know the following result: if || I + C || < 1, then C is non-singular?
|
|
IP Logged |
|
|
|
|