1
$\begingroup$

Let $V$ be a $\mathbb{Q}$-vector space and suppose there is a decomposition of $V_{\mathbb{C}}:=V \otimes_{\mathbb{Q}} \mathbb{C}$ into two $\mathbb{C}$-sub-vector spaces i.e., $V_{\mathbb{C}} \cong V^{1,0} \oplus V^{0,1}$, where $V^{i,j}$ are sub-spaces of $V_{\mathbb{C}}$ satisfying $V^{i,j}= \overline{V^{j,i}}$. Note that such a structure is known as a pure Hodge structure of weight $1$. I am looking for an example of $V$ along with such a decomposition such that $$\Bigl(\Bigl({\bigwedge}^2 V^{1,0}\Bigr) \wedge \Bigl({\bigwedge}^2 V^{0,1}\Bigr)\Bigr) \cap {\bigwedge}^4 V \not= {\bigwedge}^2 \Bigl(\Bigl(V^{1,0} \wedge V^{0,1}\Bigr) \cap {\bigwedge}^2 V\Bigr).$$ The question is motivated by the Hodge decomposition for very general abelian varieties, for which the two sides are indeed equal (where $V$ plays the role of the first cohomology of the abelian variety).

$\endgroup$
3
  • $\begingroup$ Isn't this a rather special property of a Hodge structure? E.g. if V is a weight one HS of rank 4 (that is, the H^1 of some 2-dimensional complex torus A) then LHS is always 1-dimensional, while RHS is the exterior square of the space of Hodge classes in Λ^2V=H^2(X) so the equality holds iff rk NS(A)=2. $\endgroup$
    – SashaP
    Commented Jan 11, 2022 at 21:19
  • $\begingroup$ @SashaP Yes, I guess such inequalities will happen when the Picard rank is at least $2$. However, it is not as special as one would hope. This happens for the abelian varieties lying over countably many components of the moduli space (instead of finitely many). The union of such components could be dense in the moduli space. I was trying to understand how would such an "extra" vector arise, in terms of simple linear algebra!! $\endgroup$
    – Chen
    Commented Jan 11, 2022 at 21:23
  • $\begingroup$ Yes, I didn't phrase my claim well: I meant to say that the equality is a special situation and inequality holds for any 2-dimensional torus of Picard rank 0 or 1, so for a very general abelian surface the inequality holds. $\endgroup$
    – SashaP
    Commented Jan 11, 2022 at 21:27

0

You must log in to answer this question.