We know that for a curve $X$, any object $\mathcal{E}^{\bullet}$ in the derived category $D^b_{\text{coh}}(X)$ is formal, i.e. $\mathcal{E}^{\bullet}$ is quasi-isomporphic to the direct sum of its cohomology sheaves. The reason is that the cohomological dimension of $X$ is $1$. We can see Corollary 3.15 of Daniel Huybrechts' book "Fourier–Mukai transforms in algebraic geometry" for details.

Now could we find an "easy" example of object in $D^b_{\text{coh}}(\mathbb{P}^2)$ which is not formal? In particular, could we find a complex of sheaves on $\mathbb{P}^2$ of length $2$ with coherent cohomology which is not quasi-isomorphic to the direct sum of its cohomology sheaves?


Yes; whenever you have two objects in an abelian category such that $Ext^2(M,N)$ is not equal to 0, we have a nonformal object given by coning with this morphism. More down-to-earthly, the element of $Ext^2(M,N)$ is given by some complex $N \to K \to L\to M$; the non-formal complex is just $\cdots \to 0\to K \to L \to 0\to \cdots$.

EDIT: Thanks for the example below. I was too lazy to provide one, but I also think it risks camouflaging the actual point here, since there's nothing special about coherent sheaves on $\mathbb{P}^2$, this happens in any abelian category with global dimension $>1$.

  • $\begingroup$ Thank you! Maybe I need a more explicit construction. For example we know that $Ext^2_{\mathbb{P}^2}(\mathcal{O},\mathcal{O}(-3))\neq 0$ hence as you pointed out we have a complex $\mathcal{O}(-3)\rightarrow K \rightarrow L\rightarrow \mathcal{O}$. Now could we find an explicit expression of the $K$ and $L$? $\endgroup$ – Zhaoting Wei Feb 7 '15 at 16:46
  • 7
    $\begingroup$ Just take the Koszul complex associated to the surjective map $\ \mathscr{O}_{\mathbb{P}^2}(-1)^3\rightarrow \mathscr{O}_{\mathbb{P}^2}\ $ given by multiplication by $X,Y,Z$. The corresponding morphism is $\ \mathscr{O}_{\mathbb{P}^2}(-1)^3\rightarrow \mathscr{O}_{\mathbb{P}^2}^3\;$, given by the matrix $\pmatrix{0 & Z & -Y\\ -Z & 0 & X\\ Y & -X & 0}$. $\endgroup$ – abx Feb 7 '15 at 16:59
  • 1
    $\begingroup$ @abx: $O^3$ should be $O(-1)^3$. $\endgroup$ – Sasha Feb 7 '15 at 18:04
  • $\begingroup$ Oops -- right, thanks, sorry for the typo. $\endgroup$ – abx Feb 7 '15 at 18:18

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.