Register to reply

Invariant spaces and eigenvector problem

by gottfried
Tags: eigenvector, invariant, spaces
Share this thread:
gottfried
#1
Nov12-12, 06:41 AM
P: 119
1. The problem statement, all variables and given/known data
Let W be a 1-dimensional subspace of V that is A-invariant. Show that every non zero vector in W is a eigenvector of A. [A element of Mn(F)]

3. The attempt at a solution

We know W is A-invariant therefore for all w in W A.w is in W. W is one dimensional which implies to me that A must therefore be a one by one matrix with an entry from F. Is this a correct assumption?

If so then A.w=λ.w where λ is an element of F which implies that all w in W are eigenvectors of A.

I'm new to this sort of linear algebra and therefore can't tell if I've made a blatant mistake?
Phys.Org News Partner Science news on Phys.org
Experts defend operational earthquake forecasting, counter critiques
EU urged to convert TV frequencies to mobile broadband
Sierra Nevada freshwater runoff could drop 26 percent by 2100
HallsofIvy
#2
Nov12-12, 06:57 AM
Math
Emeritus
Sci Advisor
Thanks
PF Gold
P: 39,569
Quote Quote by gottfried View Post
1. The problem statement, all variables and given/known data
Let W be a 1-dimensional subspace of V that is A-invariant. Show that every non zero vector in W is a eigenvector of A. [A element of Mn(F)]

3. The attempt at a solution

We know W is A-invariant therefore for all w in W A.w is in W. W is one dimensional which implies to me that A must therefore be a one by one matrix with an entry from F. Is this a correct assumption?
No, in fact it makes no sense at all. Any n by n matrix can have "1-dimensional invarient subspace".

If so then A.w=λ.w where λ is an element of F which implies that all w in W are eigenvectors of A.
How did you arrive at that? If w is in the invariant subspace, W, then Aw is also in W and, since W is "1-dimensional", it must be a multiple of A. But it does not immediately follow that that multiple is the same for all vectors in W.

Suppose that u and v are different vectors in W. Then we can say that [itex]Au= \lambda_1 u[/itex] and that [itex]Av= \lambda v[/itex] but we cannot yet say that [itex]\lambda_1= \lambda_2[/itex]. To do that, we need to use the fact that W is "1-dimensional" again. Because of that, any vector in the subspace is a mutiple of any other (except 0). We can write u= xv so that Au= x(Av). What does that give you?

I'm new to this sort of linear algebra and therefore can't tell if I've made a blatant mistake?
gottfried
#3
Nov12-12, 07:55 AM
P: 119
Thanks for the help. I'm still a little confused though.

Quote Quote by HallsofIvy View Post

Suppose that u and v are different vectors in W. Then we can say that [itex]Au= \lambda_1 u[/itex] and that [itex]Av= \lambda v[/itex] but we cannot yet say that [itex]\lambda_1= \lambda_2[/itex].
Why do we need to show that the two λ's are the same. Isn't that [itex] Au= \lambda_1 u[/itex] sufficient because that is the definition of an eigenvector? Unless eigenvalues have to unique for each matrix?


Register to reply

Related Discussions
Eigenvector problem Precalculus Mathematics Homework 4
Another eigenvector problem Calculus & Beyond Homework 1
Eigenvalue and Eigenvector problem Calculus & Beyond Homework 4
Eigenvector solution problem Advanced Physics Homework 8
Urgent: eigenvector problem Introductory Physics Homework 4