Rosenberg's reconstruction theorem (Q1747829)

In very short words, the article gives a precise, correct, detailed and readable proof of \textit{P. Gabriel} [Bull. Soc. Math. Fr. 90, 323--448 (1962; Zbl 0201.35602)] and \textit{A. L. Rosenberg}'s famous result [Noncommutative algebraic geometry and representations of quantized algebras. Dordrecht: Kluwer Academic Publishers (1995; Zbl 0839.16002)]: Theorem (Gabriel, Rosenberg). Let $X$ and $Y$ be two quasi-separated schemes. If the categories $\mathsf{Qcoh}(X)$ and $\mathsf{Qcoh}(Y)$ are equivalent, then $X$ and $Y$ are isomorphic. All conditions for the theorem to be true are given in details. Quasi-separatedness means that the intersection of two open affines is again affine, so that for instance separated schemes are clearly separated. The reconstruction theorem is interesting in itself, but is of importance for the definition of noncommutative algebraic geometry using abelian categories as noncommutative schemes, as is done in work by Rosenberg [loc. cit.], \textit{M. Artin} and \textit{J. J. Zhang} [Adv. Math. 109, No. 2, 228--287 (1994; Zbl 0833.14002)], and \textit{M. Van den Bergh} [Blowing up of non-commutative smooth surfaces. Providence, RI: American Mathematical Society (AMS) (2001; Zbl 0998.14002)]. The proof works by associating to each suitable abelian category $\mathcal A$ a ringed space $\text{Spec}(\mathcal A)$ called the spectrum of $\mathcal A$, such that for every quasi-separated scheme $X$, $X\cong\text{Spec}(\mathsf{Qcoh}(X))$. An application of the explicit proof of the reconstruction theorem is the classification of the equivalences of categories $\mathsf{Qcoh}(X)\simeq\mathsf{Qcoh}(Y).$ Theorem. Let $X$ and $Y$ be two quasi-separated schemes. Every equivalence of categories $\mathsf{Qcoh}(X)\simeq\mathsf{Qcoh}(Y)$ is isomorphic to $f^\ast(-)\otimes\mathcal L$ for some isomorphism $f:Y\overset\sim\rightarrow X$ and some invertible sheaf $\mathcal L$ on $Y$. Moreover, this construction yields an equivalence of groupoids \[ \left\{\begin{matrix}f:Y\overset\sim\rightarrow X\text{ isomorphism,}\\\mathcal L\text{ invertible sheaf on }Y\end{matrix}\right\}\simeq\{\text{equivalences }\mathsf{Qcoh}(X)\overset\sim\rightarrow\mathsf{Qcoh}(Y)\}. \] From this it follows that if $X$ is a quasi-separated scheme, then the automorphism class group of $\mathsf{Qcoh}(X)$, the group of isomorphism classes of auto-equivalences of $\mathsf{Qcoh}(X)$, is isomorphic to the semidirect product $\text{Aut}(X)\ltimes\text{Pic}(X)$, where $\text{Aut}(X)$ acts on the Picard group $\text{Pic}(X)$ via direct images. This is proved by Zhu for quasi-compact separated schemes, and Calabrese and Groechenig have proved the corresponding result for quasi-compact separated algebraic spaces which are flat over some commutative ring. Also, a similar result is obtained in the derived setting by \textit{A. Bondal} and \textit{D. Orlov} [Compos. Math. 125, No. 3, 327--344 (2001; Zbl 0994.18007)], where translations provide the only further class of autoequivalences if the canonical bundle of $X$ is ample or anti-ample. In this case they prove that a smooth irreducible projective variety may be reconstructed from its derived category of coherent sheaves. The author of this article claims that Rosenberg's published proofs of the main theorem are quite long and maybe incomplete. He uses Rosenberg's spectrum construction with a modification due to Gabber to give a clear and correct proof. The construction of the spectrum is self-contained and complete with regard to its application to the reconstruction theorem. From the tensor structure on quasi-coherent sheaves appearing in the construction, more can be done: Balmer has reconstructed a Noetherian scheme from its tensor triangulated category of perfect complexes, and this is generalized to quasi-compact quasi-separated schemes by \textit{A. B. Buan} et al. [Homology Homotopy Appl. 9, No. 1, 45--74 (2007; Zbl 1118.18005)]. Also, \textit{J. Lurie} [``Tannaka duality for geometric stacks'', Preprint, \url{arXiv:math/0412266}] has reconstructed an arbitrary semi-separated quasi-compact algebraic stack from its tensor category of quasi-coherent sheaves. This can be seen as a generalization of Tannaka duality. In the case of algebraic stacks the tensor structure is essential; the classifying stack $BC_2$ of a cyclic group of order $2$ over a nontrivial ring $R$ is not isomorphic to the scheme $\text{Spec}(R)\coprod\text{Spec}(R),$ but have equivalent categories of quasi-coherent sheaves provided $2\in R^\times.$ Also, the tensor structure is a main ingredient for reconstructing morphisms of schemes, resp. stacks, from their pullback functors between categories of quasi-coherent sheaves. The article is very explicit and achieves its goal in making a clear and correct proof of the reconstruction theorem. The general construction of the spectrum of an abelian category and the conditions the category needs to fulfil to be able to define a Zariski topology is very valuable, and so is the construction of the structure sheaf making the spectrum to a ringed space. The proof of the reconstruction theorem that follows from these very good preparations, is impressively clear and readable. Then the appendix with the classification of topologizing reflexive subcategories is a very nice contribution by itself.