Existence of Ulrich bundle on some surfaces of general type (Q6183571)

Ulrich bundles on a nonsingular, projective variety are certain vector bundles with the simplest possible cohomology. More precisely, they are vector bundles without intermediate cohomology and with the maximal possible numbers of generators. In 80's they appeared in the arena of Commutative Algebra under the name maximally generated maximal Cohen-Macaulay module (i.e. MGMCM module). They appear in the context of Algebraic Geometry with the foundational paper by \textit{D. Eisenbud} and \textit{F.-O. Schreyer} [J. Am. Math. Soc. 16, No. 3, 537--575, appendix 576--579 (2003; Zbl 1069.14019)]. The existence of Ulrich bundles on a smooth projective variety offers significant insight about the variety itself (e.g. existence of such bundles means that the associated `Cayley-Chow' form of the variety is simpler). Noting the importance of the existence of Ulrich bundles on a smooth projective variety, the following fundamental questions were raised by Eisenbud and Schreyer (Nowadays called the `Eisenbud-Schreyer' conjecture): \begin{itemize} \item[(i)] Does every smooth projective variety support an Ulrich bundle? \item[(ii)] If so, what is the smallest possible rank? \end{itemize} For a smooth projective curve, we have a complete answer to the above questions. For higher dimensions, there is no general result known (although several partial results are available in the literature for the surface case). In the present paper, the authors take up the case of certain surfaces of general type. Let \(X\) be a smooth projective surface (/\(\mathbb C\)) such that the canonical bundle \(K_X\) is very ample, \(\text{Pic}(X)= \mathbb Z[K_X]\) and \(q(X)=h^1(\mathcal O_X) < \chi(\mathcal O_X)-1\) (indeed, there are examples of surfaces satisfying the above hypothesis (cf. Remark 1.3)). Then the first main result of the article states that \(X\) supports a rank \(2\) stable bundle (cf. Theorem 1.1). The strategy of the proof is as follows: First, the existence of a zero-dimensional subscheme (say \(Z\)) of a certain length (say \(l\)) on \(X\) is established such that it satisfies Cayley-Bacharch property for the linear system \(|\mathcal O_X(5)|\) (cf. Lemma \(4.1\)). This by Serre correspondence guarantees the non-emptiness of moduli variety of rank \(2\) locally free extensions of \(\mathcal O_X\) by \(I_{Z, X}(4)\), where \(Z \in \text{Hilb}^l(X)\). Also, it is shown that the general element of this variety is stable (cf. Corollary 4.3). This gives non-emptiness of moduli space of slope stable rank \(2\) bundles with appropriate Chern invariants. Finally, using the necessary and sufficient criterion of special Ulrich bundles in terms of first and second Chern classes and initializedness, it is shown that for a general element \(E\) of the moduli space \(E(1)\) is Ulrich. A smooth projective variety is called Ulrich-wild if it supports families of pairwise non-isomorphic, indecomposable, Ulrich bundles of arbitrary dimension. Now consider \(X\) to be the same as in the beginning of the previous paragraph. Then the second main result of this article states that \(X\) is Ulrich-wild (cf. Theorem 1.4). The proof of this uses the following sufficient criterion for Ulrich-wildeness: If \(A\) and \(B\) are simple Ulrich bundles on \(X\) such that \(h^1(A \otimes B^*) \geq 3\) and every non-zero morphism from \(A\) to \(B\) is an isomorphism, then \(X\) is Ulrich-wild.