I need help getting through this proof

Let $T : R^n \rightarrow R^n$ be a linear transformation. T has an eigenvalue $\lambda$ if there exists some non-zero vector $\vec{x} \in R^n$ such that $T(\vec{x}) = \lambda\vec{x}$. We say that T is similar to another linear transformation $S:R^n \rightarrow R^n$ if there exists an invertible linear transformation $F:R^n\rightarrow R^n$ such that $$ S=F^{-1}\circ T \circ F, $$ which means $S(\vec{x})=(F^{-1} \circ T \circ F)(\vec{x})$ for all $\vec{x} \in R^n$. The following steps will help us prove similar linear transformations have the same eigenvalues.

a) Let $\vec{x}$ be an eigenvector of T with eigenvalue $\lambda$. Show $F^{-1}(\vec{x}) \neq 0$. Hint: Assume $F^{-1}(\vec{x}) = 0$ then apply F to both sides of the equations. Why does this lead to a contradiction?

b) Show $F^{-1}(\vec{x})$ is an eigenvector of S with eigenvalue $\lambda$. (Why is it important that $F^{-1}(\vec{x})\neq 0$?)

c) Show that if $T$ is similar to $S$ then $S$ is similar to $T$.

d) Use the above observations to write a proof of the following statement: If $T$ is similar to $S$ then $T$ and $S$ have the same eigenvalues.

I managed to get the first part (below), but I am having trouble going from there.

Let $x$ be an eigenvector, such that $F^{-1}(\vec{x}) = 0 \rightarrow \vec{x} = 0$, which is a contradiction because eigenvectors are not $0$.


  • $\begingroup$ A related problem. $\endgroup$ – Mhenni Benghorbal Dec 8 '13 at 2:01
  • $\begingroup$ @MhenniBenghorbal I saw that one, but it honestly did not help me; it is not quite the same. $\endgroup$ – lord_sneed Dec 8 '13 at 2:04
  • $\begingroup$ See this problem. It proves the eigenvalue case and it tells you what is the corresponding eigenvector. $\endgroup$ – Mhenni Benghorbal Dec 8 '13 at 2:24

b) $F^{-1}TF\,(F^{-1}x)=F^{-1}Tx=F^{-1}(\lambda x)=\lambda\,(F^{-1}x)$.

c) and d) should be then easy.

  • $\begingroup$ For (c) does "similar" in this case, mean substitute T for S (so wherever there is a T, put a S)? $\endgroup$ – lord_sneed Dec 8 '13 at 2:24
  • $\begingroup$ Yes, that's a way to do it. Specifically, for c), consider $G:=F^{-1}$ then $T=G^{-1}SG$ satisfying the definition of '$S$ similar to $T$'. $\endgroup$ – Berci Dec 8 '13 at 2:55
  • $\begingroup$ so then it becomes: $[F^{-1} \circ S \circ F](F^{-1}(x)) = [F^{-1} \circ S] (x) = F^{-1}(\lambda x) = \lambda F^{-1}(x)$ ? $\endgroup$ – lord_sneed Dec 8 '13 at 3:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.