Let $T$ be a linear operator on a finite-dimensional vector space $V$, and let $W$ be a $T-$invariant subspace of $ V$. Suppose that $v_1, v_2, . . . , v_k$ are eigenvectors of $T$ corresponding to distinct eigenvalues. Prove that if $v_1+v_2+· · ·+v_k$ is in $W$, then $v_i ∈ W$ for all $i$.

Use the above lemma and prove the given result.


Prove that the restriction of a diagonalizable linear operator $T$ to any nontrivial $T-$invariant subspace is also diagonalizable.

I could prove the above lemma using induction.

My attempt:- A linear operator $T$ on a finite-dimensional vector space $V$ is diagonalizable if and only if $V$ is the direct sum of the eigenspaces of $T$. Let $\lambda_1, \lambda_2,..., \lambda_k$ are the eigen values of $T$, $V=E_{\lambda_1} \bigoplus E_{\lambda_1}\bigoplus ...\bigoplus E_{\lambda_k}.$ Let $W_{\lambda_j}=E_{\lambda_j}\cap W(1\leq j \leq k)$. Will we get $W=W_{\lambda_1} \bigoplus W_{\lambda_1}\bigoplus ...\bigoplus W_{\lambda_k}?$

Claim 1:-

$W=W_{\lambda_1} \bigoplus W_{\lambda_1}\bigoplus ...\bigoplus W_{\lambda_k}.$

$$V=E_{\lambda_1} + E_{\lambda_1}+ ...+ E_{\lambda_1}.$$ $$V\cap W=W \cap (E_{\lambda_1} + E_{\lambda_1}+ ...+ E_{\lambda_k}).$$

Claim 2:-

$W \cap (E_{\lambda_1} + E_{\lambda_1}+ ...+ E_{\lambda_k})=(W \cap E_{\lambda_1}) + (W\cap E_{\lambda_1})+ ...+ (W\cap E_{\lambda_k})$

$x\in (W \cap E_{\lambda_1}) + (W\cap E_{\lambda_1})+ ...+ (W\cap E_{\lambda_k})\implies$ $x=x_1+x_2+...x_k \implies x \in W \cap (E_{\lambda_1} + E_{\lambda_1}+ ...+ E_{\lambda_k})$ How do I prove the converse? Does there exist any counter example to prove that result is false? How do I use the above lemma and prove the result?


Let $w \in W$ Then $w \in V=E_{\lambda_{1}} \bigoplus E_{\lambda_{2}}\bigoplus ...\bigoplus E_{\lambda_{k}}$. So we can write $w=w_1+w_2+...+w_k$ where $w_i \in E_{\lambda_{i}}$ for each $i$. The lemma now tells you that $w_i \in W$ for each $i$. Claim 2 follows immediately from this.

  • $\begingroup$ Thank you, sir.@KaviRamaMurthy $\endgroup$ – Unknown x Jun 4 '18 at 10:02

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.