Let $V$ be a finite dimensional real vector space, let $\mathrm{Alt}^k(V)$ denote the space of alternating $k$-linear forms on $V$ and let $\wedge^k(V)$ denote the $k^{th}$ exterior power of $V$. I am trying to see why the algebraic dual $\wedge^k(V)^* := (\wedge^k(V))^*$ is isomorphic to $\mathrm{Alt}^k(V)$. Here are my thoughts:

By the universal property of the exterior power, for any alternating $k$-linear form $f$ with domain $V^k$ there exists a unique linear form $\phi$ with domain $\wedge^k(V)$ such that $$ \phi(v_1 \wedge \cdots \wedge v_k) = f(v_1, \dots, v_k). $$ The universal property thus provides a mechanism to produce elements in $\wedge^k(V)^*$ from elements in $\mathrm{Alt}^k(V)$ and this mechanism of production is unique.

Thus, we have an injection $$ \Phi: \mathrm{Alt}^k(V) \longrightarrow \wedge^k(V)^* $$

What I'm not sure about is how to argue surjectivity; what is the best way to approach this?

  • $\begingroup$ Since everything is finite dimensional, you can just count dimensions to show surjectivity. $\endgroup$ – Joe Johnson 126 Jun 12 '12 at 21:22
  • $\begingroup$ @JoeJohnson126 Yes, I see this, but counting the dimensions of these spaces is a fair amount of work. One has to pick bases, show that they are, in fact, bases and then count....Is there a more direct way from the universal property to do this that doesn't involve counting? $\endgroup$ – ItsNotObvious Jun 12 '12 at 21:25
  • $\begingroup$ I think the two are switched up in the paragraph that begins "The universal property..." $\endgroup$ – Dylan Moreland Jun 12 '12 at 21:34
  • $\begingroup$ @ItsNotObvious I can see what the dimension of the exterior power is immediately, but not the alternating forms. I imagine that it's standard though, and if the dimensions match you are already done. $\endgroup$ – rschwieb Jun 12 '12 at 21:46
  • $\begingroup$ @rschwieb Yes, the dimension of both spaces, assuming $V$ is of dimension $n$, is given by $\frac{n!}{k!(n - k)!}$, but would like to prove this result without using dimension arguments, if possible... $\endgroup$ – ItsNotObvious Jun 12 '12 at 21:50

The canonical map $\psi\colon V^k \to \bigwedge^k V$ given by $\psi(v_1, \ldots, v_k) = v_1 \wedge \cdots \wedge v_k$ is alternating. If you have an element $g \in (\bigwedge^k V)^*$ then $g \circ \psi\colon V^k \to \mathbf R$ is also alternating, and I believe that this assignment $g \mapsto g \circ \psi$ gives you an inverse to $\Phi$.

This is simpler than you might fear, and has little to do with the ground ring or the finiteness of $V$. And maybe we should expect that, since the universal property of $\bigwedge^k V$ more or less says that it represents the functor $W \mapsto L^k_a(V, W)$.

Extra trouble and hypotheses seem to enter once you try to write things such as an isomorphism $(\bigwedge^k V)^* \approx \bigwedge^k (V^*)$, an embedding $\bigwedge^k V \hookrightarrow V^{\otimes k}$, or the wedge product induced on alternating forms. Some examples of this are discussed in the back of Fulton and Harris.

  • $\begingroup$ There's a bit more to say (and check!). I'll try to find time after walking the dog/doing other stuff. $\endgroup$ – Dylan Moreland Jun 12 '12 at 21:54
  • 1
    $\begingroup$ What you have said is already correct and interesting : +1. (And don't forget the ball for the dog !) $\endgroup$ – Georges Elencwajg Jun 12 '12 at 22:06
  • $\begingroup$ Nice; thanks for your help. $\endgroup$ – ItsNotObvious Jun 13 '12 at 13:41

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.