8
$\begingroup$

Let $V$ be a vector space of $\dim n$ over $K$. Let $P$ be the set of all pure products of the form $v_1 \bigwedge v_2$. How to prove that there is a one-one correspondence between the one dimensional subspaces in $P$ and the two dimensional subspaces of $V$.

Try: the correspondence is: subspace spanned by $v_1 \bigwedge v_2$ <-------> subspace spanned by $(v_1,v_2)$. I showed that, subspace spanned by $(v_1,v_2)$ ------> subspace spanned by $v_1 \bigwedge v_2$ is well defined. I cannot show that the reverse arrow is well defined. Please help.

$\endgroup$
  • 1
    $\begingroup$ Hint: If $(v_1,v_2)$ is a basis for a subspace of $V$, then you can get to any other basis by the operation $(v_1,v_2)\mapsto (v_1,rv_1+v_2)$, scalar multiplication of $v_i$, and switching the order. What effect do these operations have on the wedge $v_1\wedge v_2$? $\endgroup$ – Cheerful Parsnip Mar 27 '13 at 19:52
  • 3
    $\begingroup$ @JimConant I strongly suspect you are proving the wrong implication. You show $sp(v_1,v_2) = sp(v_3,v_4) \implies v_1\wedge v_2 = v_3\wedge v_4$. OP is interested in the reverse. $\endgroup$ – Lord_Farin Mar 27 '13 at 20:00
  • $\begingroup$ Welcome user1 and +1 , since your question is well-formulated and interesting. You are asking about a particular case of the Plücker embedding: if you google this two-word expression you will find many interesting online documents about the subject. $\endgroup$ – Georges Elencwajg Mar 27 '13 at 20:31
  • $\begingroup$ @Lord_Farin: Ah, you are right. I read it too quickly. Thanks. $\endgroup$ – Cheerful Parsnip Mar 27 '13 at 20:46
4
$\begingroup$

$P$ is stable by multiplication by elements of the multiplicative group $K^\times$, and so it is enough to produce a bijection between $K^\times$-orbits of non-zero elements of $P$ and 2-d subspaces of $V$. In what follows I assume all the basics of multilinear algebra.

You have already defined a candidate function, sending $K\{v_1,v_2 \}$ to the $K^\times$ orbit of $v_1 \wedge v_2$. This map is surjective since any non-zero wedge $v_1 \wedge v_2$ has $v_1$ and $v_2$ linearly independent. It remains to show that it is injective. So let's suppose that $v_1 \wedge v_2=x(w_1 \wedge w_2)$ for some non-zero $x \in K^\times$. It follows that $v_1 \wedge w_1 \wedge w_2=0$, and hence $v_1 \in K\{w_1,w_2\}$ (I consider this assertion part of the "basics" of multilinear algebra, but understand I am unusual here, so am going to include a proof sketch below anyway). By symmetry $v_2 \in K\{w_1,w_2\}$, and now it follows that $K\{v_1,v_2\}=K\{w_1,w_2\}$.

Proof that $v_1 \wedge w_1 \wedge w_2=0$ implies $v_1$ in the span of $w_1$ and $w_2$: We may complete $w_1,w_2$ to a basis of $V$. Write $v_1=\sum a_i w_i$ and compute $$0=v_1 \wedge w_1 \wedge w_2=\sum_{i=3}^n a_i w_i \wedge w_1 \wedge w_2.$$ Now the vectors on the RHS are linearly independent, implying the assertion.

$\endgroup$
  • $\begingroup$ Very nice! ${}{}$ $\endgroup$ – Georges Elencwajg Mar 27 '13 at 22:31

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.