The alternating groups and \(K3\) surfaces (Q2502920)

It is well known [see e.g. \textit{V. V. Nikulin}, Trans. Mosc. Math. Soc. 38, 71--135 (1980; Zbl 0454.14017)] that there are exactly 80 abstract finite groups which can act symplectically (over the complex number field) on \(K3\) surfaces. Among these 80 there are exactly four perfect groups: \(A_5, L_2(7), A_6, M_{20} \). The bigger perfect groups \(G_N = L_2(7), A_6, M_{20}\) can be extended to a bigger perfect group \(G\) which is an extension over the multiplicative group of the fourth-roots of unity such that it acts faithfully on a \(K3\) surface after e.g. \textit{S. Kondo} [Am. J. Math. 121, 1245--1252 (1999; Zbl 0978.14043)]. Moreover, the pair \((X,G)\) is unique. The author proves the following theorems: Theorem A. Suppose that a finite group \(G\) acts faithfully on a \(K3\) surface. Suppose that \(G\) contains \(A_5\) as a normal subgroup. Then \(G\) is one of the following four groups, each realizable: \( A_5, S_5, A_5 \times \mu_2 , S_5 \times \mu_2 \). Theorem B. Suppose that a finite group \(G\) acts faithfully on a \(K3\) surface. Suppose further that \(G\) contains \(A_5\) as a normal subgroup. Then the transcendental value \(I(G)\) (see notation below) is 1 or 2. Theorem C. Suppose that a finite group \(G\) acts faithfully on a \(K3\) surface \(X\). Suppose that \(G\) contains a non-trivial perfect group \(H\) as a subgroup (not necessarily normal). Then: a) \(I(G) \leq 4 \). b) If \(I(G) =4 \) then \(G = L_2(7) \cdot \mu_4 , A_6 \cdot \mu_4 \) or \( M_{20} \cdot \mu_4 \) and the pair \((X,G)\) is unique, up to isomorphisms, in all three cases. The paper is organized as follows. In section one the author defines for a group \(G\) the subgroup \(G_N\) of the group of automorphisms which act trivially on the non-zero 2-form of the \(K3\) surface (section 1.0). The resulting group \(G/G_N\) is isomorphic to the multiplicative cyclic group of order \(I\), called the transcendental value of \(G\),~denoted as \(I(G)\). That is we have an extension \(G=G_N \cdot \mu_I\). He states facts 1.0 A, 1.0 B and 1.0 C which are elementary but of frequent use in the paper. For \(h \) a non-symplectic involution on a smooth algebraic surface \(Y\) in lemma 1.2 gives useful information about \(Y^h \) as a disjoint union of smooth curves, \( \chi_{\text{top}} (Y^h) \leq 18\) whenever \(A_5 \subset \text{Aut}(Y) \) and in fact 1.3 this locus is decomposed as a disjoint union of the 1-dimensional part \({Y^h}_{1-\dim} \) and the isolated part \( {Y^h}_{\text{isol}} \), calculating in proposition 1.4 for the values \(I =3,4 \) , \( \chi_{\text{top}}(Y^h) \) in terms of \( m_h = |{Y^h}_{\text{isol}}|, n_h = 1/2 \chi_{\text{top}}({Y^h}_{1-\dim} ) \) and giving upper and lower bounds for \( n_h, |Y^{h \delta}|, |Y^{h^2 \delta}|\) for the case that \( \delta \) is symplectic of order three and five, which commutes with \(h\). He then assumes that \(G = G_N \cdot \mu_I \) acts on a \(K3\) surface \(X\) when \(G_N = A_5\) and determines the action of \(G_N\) on the Néron-Severi lattice \(S_X\) of \(X\) in lemma 1.6, in particular he gives the irreducible decomposition: \( S_X \otimes C = \chi_1 \oplus {\chi_1}' \oplus \chi_4 \oplus {\chi_4}' \oplus \chi_5 \oplus {\chi_5}'\) where the \('\) denotes a copy of the given character. He then states lemmas 1.8 and lemma 1.9 which are well known and used in subsequent sections. In the examples of 1.10 he shows the existence of the groups stated in theorems A and B. In section 2 he assumes that \(X\) is a \(K3\) surface with a faithful action by a group of the form \(G = A_5: \mu_4 \) that \(g\) is a generator of \(\mu_4\) and \( \tau = (345)\). He proves key proposition 2.2 used to prove theorems A, B and C giving precise values for \[ (n_g, m_g; \chi_{\text{top}}(X^g), \chi_{\text{top}}(X^{g \tau}), \chi_{\text{top}}(X^{g^2 \tau}), \chi_{\text{top}}(X^{g^2})). \] He also proves lemma 2.3 which is used in 2.4 to determine the representation of \(S_3 \times \mu_4 \). He determines in lemma 2.5 the matrix representation of \(S_3 \times \mu_4 \) on \( \chi_4 \oplus \chi_4 ' \) and on \(\chi_5 \oplus \chi_5 '\). To prove 2.2 he uses lemma 2.6 and lemma 2.9. In section 3, he proves theorems A,B and C stated in the introduction of the paper. He starts proving theorem C which in the course of the proof, reduces to the case that \(G_N \) is either one of \( A_5, S_5, A_6, M_{20} \) and considers each of the previous cases separately and proves each of theorems A, B and C.