A remark on Ulrich and ACM bundles (Q1732886)

Let \(X \subseteq \mathbb{P}^n\) be a smooth projective variety over an algebraically closed field \(k\), and let \(E\) be a vector bundle on \(X\). \(E\) is called an Ulrich bundle if \(H^i(X,E(-i))=0\) for \(i>0\) and \(H^j(X,E(-j-1))=0\) for \(j < \dim X\). A key observation of the paper under review is that if \(X\) is a smooth, projective, ordinary (i.e., the Frobenius map \(H^i(X, \mathcal O_X) \to H^i (X, \mathcal O_X)\) is isomorphic for every \(i\)) curve of genus \(g \ge 2\) over \(k\) of characteristic \(p>0\) has an Ulrich bundle of rank \(p-1\), namely, \(B_X^1 (1)\), where \(B_X^1 := im [d:\mathcal O_X \to \Omega_X^1]\) is the sheaf of locally exact differentials. When \(X\) is not ordinary, one requires a further twist by a certain line bundle \(L\) of degree \(0\). For \(\dim X \ge 2\), the sheaf \(B_X^1\) is unlikely to be Ulrich in general. However, when \(X\) is a smooth projective variety with trivial canonical line bundle, then \(B_X^1\) is an ACM bundle on \(X\) if and only if \(X\) is Frobenius split (see Theorem 3.4). In particular, when \(p>2\) and \(X\) is ordinary Calabi-Yau, then \(B_X^1\) is an ACM bundle which does not split as a direct sum of line bundles.