Automorphisms of even lattices that arise in connection with automorphisms of algebraic K3-surfaces (Q788780): Difference between revisions

Let X be a K3 surface over \({\mathbb{C}}\) and \(S(X)\subseteq H_ 2(X,{\mathbb{Z}})\) the sublattice of rational cycles on X. Let T(X) be the orthogonal complement of S(X) in \(H_ 2(X,{\mathbb{Z}})\). The group Aut X of automorphisms of X acts on \(H_ 2(X,{\mathbb{Z}})\) and preserves both T(X) and S(X). Set \(H(X):=Ker(Aut X\to Aut S(X))\). Then \(H(X)\simeq {\mathbb{Z}}/m{\mathbb{Z}}\) with \(m=m(X)\in {\mathbb{Z}}\). \textit{V. V. Nikulin} [Tr. Mosk. Mat. O.-va 38, 75-137 (1979; Zbl 0433.14024)] has established deep arithmetic restrictions on H(X). The author uses these restrictions to describe lattices which might appear as T(X) with an action of a cyclic group. The results obtained are very detailed. The best result is the following theorem. If T(X) is not unimodular then \(m(X)\in \{2,3,4,5,7,8,9,11,13,16,19,25,27\}\). Moreover if \(m\in {\mathbb{Z}}\) is odd then there exists a K3 surface with \(m(X)=m.\) For the case when T(X) is unimodular the results of the paper are still very strong. The results for T(X) are obtained as applications of general results on even lattices with automorphisms of special kind. The study of the latter case uses properties of units in cyclotomic extensions. No proofs or indications of proofs are given.