Please offer a solution to the following problem. It was offered in class by my professor as an additional exercise to try on one's own.

Let $V$ be the inner product space, and assume that $\alpha \in End(V)$. Suppose that $A$ is the representation matrix of $\alpha$ with respect to an orthonormal basis {$v_1,...,v_n$}.

(i) Prove that $\alpha$ is self-adjoint if and only if $A = A^*$.

(ii) Prove that if $B$ is the representation matrix of $\alpha$ with respect to another orthonormal basis {$w_1,...,w_n$}, then $B=U^*AU$ for some matrix $U$ such that $U^*U=I$.

Thank you for your assistance.

  • $\begingroup$ The formula $\alpha v = \alpha (\sum_k x_k v_k) = \sum_k x_k \alpha v_k = \sum_k x_k \sum_i [A]_{ik} v_i = \sum_k \sum_i [A]_{ik} x_k v_i = \sum_i [Ax]_i v_i$ may be helpful. $\endgroup$
    – copper.hat
    Nov 19, 2012 at 0:56

1 Answer 1


For part 1, use Proposition 16.16 from Golan.

Let $V$ and $W$ be finitely-generated inner product spaces, having ONB $B=\{v_1, \ldots, v_n\}$ and $D=\{w_1, \ldots, w_n\}$, respectively. Let $\alpha: V \to W$ be a linear transformation. Then $\Phi_{BD}(\alpha)$ is the matrix $A=[a_{ij}]$, where $a_{ji}=\langle \alpha(v_i),w_j\rangle$ and $\Phi_{DB}(\alpha^*)=A^H$.

The proof of this proposition follows: For all $1 \leq i \leq n$, let $\alpha(v_i)= \displaystyle\sum_{h=1}^k a_{hj} w_h$. Then for all $1 \leq j \leq k$, we have $\langle \alpha(v_i), w_j \rangle= \langle \sum_{h=1}^k a_{hj} w_h, w_j \rangle=a_{ji}$ and also $\langle \alpha^*(w_j),v_i \rangle= \overline{\langle v_i, \alpha^*(w_j) \rangle}= \overline{\langle \alpha(v_i), w_j \rangle}=\overline{a_{ji}}$ as needed.

The proof for this proposition is similar to the proof of Part 1.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .