0
$\begingroup$

I have to show that the rows and columns of a unitary matrix are orthonormal sets. I have not been provided with a specific matrix, just 'a unitary matrix.' I understand how to show that given vectors are orthonormal, and I know that if the rows of a unitary matrix are orthonormal, then the columns are as well, but I can't think of how to prove this with an arbitrary matrix. Any help is appreciated.

$\endgroup$
  • $\begingroup$ How are the the entries of a product of two matrices related to the rows and columns of the two matrices? $\endgroup$ – Aweygan Feb 17 '17 at 23:46
  • $\begingroup$ The real case of this result. $\endgroup$ – Git Gud Feb 17 '17 at 23:52
1
$\begingroup$

Two (complex) vectors are orthogonal iff $$\sum_i a_i^*b_i = 0$$ A matrix $U$ is unitary iff $U^\dagger U = I$.

A set of vectors is orthonormal if each vector in the set has norm of $1$, that is, $\sum a_i^*a_i = 1$, and for all pairs of two distinct vectors in the set, $\sum_i a_i^*b_i = 0$.

Suppose, now, that $U$ is unitary, and let $\vec{a}$ be the $k$-th column of $U$, and let $\vec{b}$ be the $j$-th column of $U$, with $k\neq j$. Then by the rules of matrix multiplication $$\sum_i a_i^*b_i = (U^\dagger U)_{kj} = I_{kj} == 0 $$ so $\vec{a}$ and $\vec{b}$ are orthogonal.

Now let $\vec{b}$ be the $k$-th column of $U$ so that the two columns are the same column. Then $$\sum_i a_i^*a_i = (U^\dagger U)_{kk} = I_{kk} == 1 $$ so $\vec{a}$ is of norm $1$, that is to say, it is normal.

Thus the set of vectors is orthonormal.

$\endgroup$
  • $\begingroup$ Up voted with a warning. I could do without the double equal signs $\ddot \smile$ $\endgroup$ – Git Gud Feb 18 '17 at 0:05
  • $\begingroup$ Thank you, that was very helpful! It's been a few years since I took linear algebra, this was a great refresher. $\endgroup$ – Andrew Jones Feb 18 '17 at 0:40

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.