On prime degree isogenies between \(K3\) surfaces (Q1676498)

Let \(X\) and \(Y\) be complex projective \(K3\) surfaces. Denote by \(T_{X,\mathbb{Q}}\) and \(T_{Y,\mathbb{Q}}\) their transcendental lattices, that is, \(T_{X,\mathbb{Q}}\) is the orthogonal complement of the Néron--Severi group \(\mathrm{NS}(X)\) inside \(H^2(X, \mathbb{Q})\). The Hodge conjecture for the product \(X \times Y\) predicts that any isomorphism \({\phi_T} : {T_{Y,\mathbb{Q}}} \to {T_{X,\mathbb{Q}}}\) of rational Hodge structures is algebraic (induced by an algebraic cycle on \(X \times Y\)). The results of \textit{S. Mukai} [in: Vector bundles on algebraic varieties, Bombay, Tata Inst. Fund. Res., 341--413 (1987; Zbl 0674.14023)] and \textit{V. Nikulin} [Izv. Akad. Nauk SSSR Ser. Mat. 51(2), 402--411, 448 (1987; Zbl 0653.14007)] show that \(\phi_T\) is algebraic if \(\phi_T\) is an isometry and \(\mathrm{rank}\, T_{X,\mathbb{Q}} \geq 5\). (Recently, \textit{N. Buskin} [``Every rational Hodge isometry between two \(K3\) surfaces is algebraic'', Preprint, \url{arXiv:1510.02852}] removed the assumption on the rank.) Now assume there exists a dominant rational map \({\gamma} : {X} \dashrightarrow {Y}\) of prime degree \(p\). Then the induced Hodge isomorphism \({\gamma^*_T} : {T_{Y,\mathbb{Q}}} \to {T_{X,\mathbb{Q}}}\) is not an isometry. The main result of the paper determines whether the quadratic spaces \(T_{X,\mathbb{Q}}\) and \(T_{Y,\mathbb{Q}}\) are in fact non-isometric, in terms of the properties of \(T_{Y,\mathbb{Q}}\) as a quadratic space. The theory of quadratic spaces needed in the proof is reviewed in the paper. The authors give several examples to illustrate the result.