Since the paper Smooth and proper noncommutative schemes and gluing of DG categories by Orlov, dg categories are considered the non-commutative counterpart of algebraic geometry. More specifically, we call a dg category a non-commutative scheme if it is an admissible dg subcategory of the dg category $\mathfrak{Perf}(X)$ for a smooth projective scheme $X$. Now, many properties of a scheme $X$ defined over a field $k$ can be translated into properties of the category $\mathfrak{Perf}(X)$ (a dg enhancement of $Perf(X)$), e.g.

(1) the scheme $X$ is proper over $k$ if and only if for any $E,F \in \mathfrak{Perf}(X)$ we have (here I am identifying $E$ and $F$ with their image in the homotopy category of $\mathfrak{Perf}(X)$, which by definition is $Perf(X)$) $$ \sum_{n \in \mathbb{Z}} \text{dim} \, \text{Hom}_{Perf(X)}(E,F[n]) < +\infty$$

(2) the scheme $X$ is smooth over $k$ if and only if the diagonal bimodule associated to $\mathfrak{Perf}(X)$ is perfect in the derived category of $\mathfrak{Perf}(X)-\mathfrak{Perf}(X)$ bimodules.

From the above, one can then generalise these notions to that and smooth and proper dg categories. My question is whether there exists a similar analogy for the notion of projective scheme, and therefore the notion of a "projective dg category".

Thanks in advance.

  • 3
    $\begingroup$ If you want an approach to noncommutative projective geometry, this survey arxiv.org/abs/math/9910082 of Stafford-Van den Bergh describes work following the philosophy due to Artin-Tate-van den Bergh of using a category of graded modules modulo torsion to stand as a noncommutative projective space. This ultimately relies on a noncommutative version of Serre's theorem, which in the commutative setting says that this category is equivalent to that of (quasi-) coherent sheaves. I know this doesn't quite answer your question as posed but hope it's helpful. $\endgroup$ Aug 20, 2020 at 10:29
  • $\begingroup$ Have you tried making sense of "very ample line objects" in a dg-category? $\endgroup$
    – user147129
    Aug 20, 2020 at 14:28
  • $\begingroup$ @Riza That is what I was trying to make sense of actually, but I have not had luck so far. $\endgroup$ Aug 20, 2020 at 14:30
  • $\begingroup$ @Jan thanks, I'll take a look at the reference! $\endgroup$ Aug 20, 2020 at 14:30

1 Answer 1


If $X$ is a smooth projective threefold with a flopping curve $C$ then typically the variety $Y$ obtained from $X$ by a flop in $C$ is not projective, but smooth, proper, and derived equivalent to $X$. This shows that projectivity is not invariant under derived equivalence, hence does not correspond to a property of the derived category.


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.