1
$\begingroup$

I'm really stuck on the following quesiton.

Let $U$ and $V$ be finite dimensional vector spaces over the complex numbers with bases $e_1,..,e_n$ of $U$ and $f_1,...,f_m$ of $V$. They also have dual spaces $U^*$ and $V^*$ with bases $e^i$ and $f^i$ respectively.

Then assume that both spaces are Hermitian.

Let $T_U:U\to U^*$ be defined by $T_U(w)(u) = \langle w,u\rangle \forall w,u\in U$.

Prove that if the basis $e_i$ is orthonormal then $T_U(e_i) = e^i$.

I have tried showing that

$T_U(e_i)(u) = \langle e_i, u\rangle $ $ = \langle e_i,x_1e_1+...+x_ne_n\rangle $ $ = \langle e_i, x_1e_1\rangle + \langle e_i, x_2e_2\rangle +\cdots +\langle e_i, x_ne_n\rangle$ $ =x_1\langle e_i,e_1\rangle + x_2\langle e_i,e_2\rangle +\cdots +x_n\langle e_i,e_n\rangle$

So we are left with, $x_i\langle e_i,e_i \rangle $ since the others are mutually orthogonal.

I'm having trouble understanding why this implies the desired result.

$\endgroup$
  • 1
    $\begingroup$ Please use \langle and \rangle to get $\langle\cdots\rangle$. (I fixed it for you this time.) Oh, and what role is $V$ playing here? $\endgroup$ – Harald Hanche-Olsen Nov 28 '13 at 18:09
1
$\begingroup$

To see that

$T_U(e_i) = e^i, \tag{1}$

first compute

$T_U(e_i)(e_j) = \langle e_i, e_j \rangle = \delta_{ij} \tag{2}$

by orthonormality of the $e_i$. Then note that

$e^i(e_j) = \delta_{ij} \tag{3}$

as well, this time by the duality if the bases $e_i$, $e^j$. Then since $T_U(e_i)$ and $e^i$ agree on the basis elements $e_j$ of $U$, it follows by linearity thet agree on every $u = \sum u_j e_j \in U$:

$T_U(e_i)(\sum u_j e_j) = \sum u_j T_U(e_i)(e_j) = \sum u_j e^i(e_j) = e^i(\sum u_j e_j); \tag{4}$

thus

$T_U(e_i) = e^i, \tag{5}$

and we are done!QED

Hope this helps. Holiday Cheers,

and as always,

Fiat Lux!!!

$\endgroup$
  • $\begingroup$ Thanks, I almost had it. I think my troubles stemmed from not entirely understanding the definition of the map. I got a bit worried because I had $T_U$ defined as $T_U(w)(u)$ but I had to prove something about $T_U(w)$. I didn't realise that this meant I had to prove it for all $u$. Thanks everyone $\endgroup$ – benjiebob Nov 28 '13 at 18:30
  • $\begingroup$ @ benjiebob: Yup, you sure did! Glad to help out on this bee-a-ooo-ti-ful Turkey Day here in Beserkeley, California; and thanks for the "acceptance"! $\endgroup$ – Robert Lewis Nov 28 '13 at 18:34
2
$\begingroup$

The property of the dual basis is that $e^{i}(e_{j}) = \delta_{i,j}$. So, why don't you consider how $T_{U}(e_{i})$ and $e^{i}$ act on the basis of $U$?

$\endgroup$
0
$\begingroup$

Hint: First of all you don't need $V$ for your question, right? So let's first talk about your Hermitian. If $U$ is a Hermitian space, then there is a Hermitian form $$\langle, \rangle: U \times U \longrightarrow \Bbb C$$

Now your map $T_U$ is not defined in a clear way. It should go from $U \to U^*$, but then, you have to define

$$T_U: U \to U^*, u \mapsto \langle -,u \rangle$$

and then $T_U(u) \in U^*$ for $u \in U$. So to prove that $T_U(e_i) = e^i$, you just have to know how the $e^i$ is defined, namely via the Kronecker Delta: $$e^i (e_j) = \delta_{ij}$$

So start with $u = \sum_{i=1}^{n} e_i$ and show what happens if you let $T_U$ act on $u$.

$\endgroup$

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.