Rank 3 ACM bundles on general hypersurfaces in \(\mathbb{P}^5\) (Q2274043)

Given an smooth projective polarized variety \((X,\mathcal{O}_X(1))\), a vector bundle \(\mathcal{E}\) on \(X\) is called arithmetically Cohen-Macaulay (aCM) if it has no intermediate cohomology: \(H^i(E(t))=0\) for \(t\in\mathbb{Z}\) and \(1\leq i\leq \dim(X)-1\). In the case of \(X\subset\mathbb{P}^{n+1}\) being a hypersurface of degree \(d>1\), it is known that \(X\) supports non-splitting (i.e., non-isomorphic to a direct sum of line bundles) aCM bundles of a certain rank \(r>1\) (unless \(X\) is the quadric surface). \textit{R. O. Buchweitz} et al. [Invent. Math. 88, 165--182 (1987; Zbl 0617.14034)] (motivated by the classification of aCM bundles on hyperquadrics) conjectured that if \(r:=\mathrm{rank}(E)<2^e\), where \(e=\lfloor\frac{n-1}{2}\rfloor\) then any aCM bundle \(\mathcal{E}\) splits as a direct sum of line bundles. The authors also showed that this conjectured can not be strengthened further. In the paper under review, the authors point out that the previous (BGS) conjecture can be viewed from the point of view of Lefschetz theorems for Picard groups. Indeed, if an aCM bundle \(\mathcal{E}\) extends to a vector bundle on \(\mathbb{P}^{n+1}\), the latter is also aCM and therefore, by Horrock's theorem, it splits as a direct sum of line bundles. Guided by this principle, the authors suggest that it should hold an analogue to Noether-Lefschetz theorem for aCM bundles on hypersurfaces of sufficiently high degree as well. Therefore they propose the following Conjecture. (Generic BGS conjecture) Let \(X\subset\mathbb{P}^{n+1}\) be a general hypersurface of sufficiently high degree and \(\mathcal{E}\) an aCM bundle of rank \(<2^s\), for \(s=\lfloor\frac{n+1}{2}\rfloor\). Then \(\mathcal{E}\) splits as a sum of line bundles. The main result of this paper is a contribution to the latter conjecture: Theorem. (Theorem \(1.1\)) Let \(X\subset\mathbb{P}^{n+1}\) be a smooth hypersurface. then: (1) If \(n\geq 5\), any aCM rank three bundle on \(X\) splits as a direct sum of line bundles. (2) If \(n=4\) and \(X\) is general of degree \(d\geq 3\), any aCM rank three bundle on \(X\) splits as a direct sum of line bundles. (3) If \(n=3\) and \(X\) is general of degree \(d\geq 5\), any aCM rank three bundle \(\mathcal{E}\) on \(X\) is rigid (i.e., \(H^i(\mathrm{End}(\mathcal{E}))=0\) for \(i=1,2\)).