2
$\begingroup$

Let $\{e_{1},\ldots,e_{n}\}$ be the usual basis for $\mathbb{R}^{n}$ and let $\{\varphi_{i},\ldots,\varphi_{n}\}$ be the dual basis.

Show that $$\varphi_{i_{1}}\wedge\cdots\wedge \varphi_{i_{k}}(e_{i_{1}},\ldots,e_{i_{k}})=1$$

The wedge product is defined by $$\omega\wedge \eta=\frac{(k+l)!}{k!l!}\text{Alt}(\omega \otimes \eta)$$ where, say $\omega\in \Lambda^{k}(V)$, $\eta\in \Lambda^{l}(V)$ and $\omega\wedge \eta\in\Lambda ^{k+l}(V)$

and alternation

$\text{Alt}(T)(v_{1},\ldots,v_{n}):=\dfrac{1}{k!}\displaystyle\sum_{\sigma\in S_{k}}\text{sgn}\;\cdot T(v_{\sigma(1)},\ldots,v_{\sigma(k)})$ with $S_{k}$ being the set of all permutations of the numbers $1$ to $k$.

there are two steps that I don't get in the following solution:

$$ \begin{align} & \varphi_{i_{1}}\wedge\cdots\wedge \varphi_{i_{k}}(e_{1},\ldots,e_{n}) \\[6pt] = & \frac{k!}{1!\cdots 1!}\text{Alt}(\varphi_{i_{1}}\otimes\cdots\otimes\varphi_{i_{k}})(e_{1},\ldots,e_{k}) \\[6pt] = & \sum_{\sigma\in S_{k}}\text{sgn}(\sigma)\varphi_{i_{1}}(e_{\sigma(1)})\cdots \varphi_{i_{k}}(e_{\sigma(k)})=1 \end{align} $$

I am really new to multilinear algebra and will appreciate if someone can explain why in the second line the coefficient becomes $k!$ and why in the last line we get 1? Which elements do we exactly permute?

Thank you

$\endgroup$
2
$\begingroup$

The coefficient $\frac{k!}{1! \cdots 1!}$ comes from the coefficient $\frac{(k + l)!}{k!l!}$ in the wedge product formula. They're generalizing that formula to a wedge product of arbitrarily many alternating tensors:

$$\omega_1 \wedge \omega_2 \wedge \cdots \wedge \omega_r = \frac{(k_1 + k_2 + \cdots + k_r)!}{k_1!k_2!\cdots k_r!}Alt(\omega_1 \otimes \omega_2 \cdots \otimes \omega_r)$$

Where $\omega_i \in \Lambda^{k_i}(V)$. In your case, we have:

$$\varphi_{i_1} \wedge \cdots \wedge \varphi_{i_n} = \frac{(1 + \cdots + 1)!}{1! \cdots 1!}Alt(\varphi_{i_1} \otimes \cdots \otimes \varphi_{i_k})$$

For the last line, let's look at an example where $k = 3$:

$$\sum_{\sigma \in S_3} sgn(\sigma)\varphi_{i_1}(e_{i_{\sigma(1)}})\varphi_{i_2}(e_{i_{\sigma(2)}})\varphi_{i_3}(e_{i_{\sigma(3)}}) $$

$$=\varphi_{i_1}(e_{i_1})\varphi_{i_2}(e_{i_2})\varphi_{i_3}(e_{i_3}) - \varphi_{i_1}(e_{i_2})\varphi_{i_2}(e_{i_1})\varphi_{i_3}(e_{i_3}) - \varphi_{i_1}(e_{i_1})\varphi_{i_2}(e_{i_3})\varphi_{i_3}(e_{i_2}) - \varphi_{i_1}(e_{i_3})\varphi_{i_2}(e_{i_2})\varphi_{i_3}(e_{i_1}) + \varphi_{i_1}(e_{i_2})\varphi_{i_2}(e_{i_3})\varphi_{i_3}(e_{i_1}) + \varphi_{i_1}(e_{i_3})\varphi_{i_2}(e_{i_1})\varphi_{i_3}(e_{i_2}) $$

What you're doing is permuting all the possible indices of the $e's$. But you can see that only one of these terms is going to be nonzero, and that's the first one where $\sigma$ is the identity, because $\phi_i(e_j)$ is nonzero only when $i = j$.

$\endgroup$
  • $\begingroup$ Thank you very much Alex. Really liked your anwer. Just on thing that bothers me. In the equation $$\varphi_{i_1} \wedge \cdots \wedge \varphi_{i_n} = \frac{(1 + \cdots + 1)!}{1! \cdots 1!}Alt(\varphi_{i_1} \otimes \cdots \otimes \varphi_{i_k})$$ where all the $1$'s come from? Is it because each $\varphi\in\Lambda^{1}(V)$ is a linear functional on $V$ i.e belongs to the dual space. $\endgroup$ – user124471 May 5 '14 at 18:27
  • 1
    $\begingroup$ Yep, the 1's come from the fact that each of the $\varphi_i$ are one-forms. $\endgroup$ – Alex Zorn May 5 '14 at 19:02
  • $\begingroup$ Brilliant! many thanks $\endgroup$ – user124471 May 6 '14 at 14:56

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.