$T: \mathbb{R}^n \to \mathbb{R}^n$ linear with $n > 1$. Then there is $M$ with $\dim M = 2$ and $T(M) \subset M$.

55 Views Asked by At

Let $T: \mathbb{R}^n \to \mathbb{R}^n$ be a linear map, with $n > 1$. Prove that there is a subspace $M \subset \mathbb{R}^n$, with $\dim M = 2$ such that $T(M) \subset M$.

This question is from my list of eigenvalues and eigenvectors. I don't want a solution to the question, my problem is that I really don't know how to approach the question. I think some hints should enough.


Using hints: Is $\lambda = a + bi \in \mathbb{C}^n$ is a non-real eigenvalue, so $$Ah = \lambda h \Longrightarrow A(h_1 + ih_2) = (a + bi)(h_1 + ih_2) \Longrightarrow Ah_1 + iAh_2 = ah_1 - bh_2 + i(ah_2 + bh_1),$$ from where $$Ah_1 = ah_1 - bh_2;\quad Ah_2 = bh_1 + ah_2.$$ Thus, $Ah_1, Ah_2 \in Span(h_1,h_2)$. Therefore, $Span(h_1,h_2)$ is two-dimensional invariant provided that $h_1, h_2$ are LI. How to ensure that?

Now, suppose that $T$ has no complex eigenvalues. If at least two real eigenvalues are distinct, is straighforward. But what about if no?

1

There are 1 best solutions below

1
On BEST ANSWER

You are almost there with your solution using the hints.

Complex eigenvalue case

Note that if $T(v)=\lambda v$, by taking complex conjugates, $T(v^*)=\lambda^* v^*$. Then $\lambda $ and $\lambda^* $ are unequal and therefore $v$ and $v^* $ are independent.

All eigenvalues real case

We have $T(v)=\lambda v$. Similarly, $T$ acting on $\mathbb{R}^n/(v)$ will have a real eigenvalue (which may of course also be $\lambda$). If the corresponding eigenvector is $(v)w$ then the requisite 2-dimensional subspace is generated by $v$ and $w$.