# Confused about choice of vector in a proof.

1. Oct 19, 2013

### Luna=Luna

This is probably going to be a very simple question, i just need justification for a seemingly simple step in a proof.

The statement is as follows:

An endomorphism $T$ of an inner product space is ${0}$ if and only if $\langle b|T|a\rangle = 0$ for all $|a\rangle$ and $|b\rangle$.

Now it is obvious if $T$ is $0$ then $\langle b|T|a\rangle = 0$

For the converse proof if $\langle b|T|a\rangle = 0$ for all $|a\rangle$ and $b\rangle$ then $T = 0$, it starts by choosing $|b\rangle = T|a\rangle$.

Why is this valid, i guess a very naive reasoning would be doesn't this only prove it for the case that $|b\rangle = T|a\rangle$.

2. Oct 20, 2013

### fzero

In the converse we're already given that $\langle b|T|a\rangle = 0$ for all $|a\rangle$ and $b\rangle$. We don't have to prove anything for all $|b\rangle$. We just consider $T|a\rangle$ and find that its norm must be zero for all $|a\rangle$. Therefore $T|a\rangle$ is the zero vector for all $|a\rangle$, hence $T$ is the zero map.

3. Oct 21, 2013

### CompuChip

If the assumption in a proof is a "for all x in X" statement, you are always allowed to set x to any value you like. Often we are quite happy to take any allowed value ("let $x \in X$ arbitrary") as long as we know that it satisfies some condition, but you can always use that since the statement holds for all x in X, it holds for some particular convenient value.

Of course, if you are trying to prove a statement of the form "for all x in X", it is not sufficient to prove the statement for a single particular x. For example, you wrote
Actually what you meant is
It would have been wrong to write
Since that would not have been a statement about all |a⟩ and ⟨b|.