On the finite dimensionality of a \(K3\) surface (Q2428549)

Let \(X\) be a smooth complex projective \(K3\) surface with an invlution \(\iota\), \(Y:=\widetilde{X\backslash\langle\iota\rangle}\) be the resolution, and \(h(X):=(X, \Delta_X, 0)\) denote the Chow motive of \(X\), where \(\Delta_X\) is a diagonal correspondence. In the paper under review, the finite-dimensionality of \(h(X)\) for certain \(K3\) surface \(X\) conjectured by Kimura is discussed. In case of \(K3\) surfaces, the finite-dimensionality of \(h(X)\) is obtained by the evenly finite-dimensionality of the transcendental part \(t_2(X)\), which appears in the Chow-Künneth decomposition of \(h(X)\). It is known in general that there exists a natural morphism \(t_2(X)\to t_2(Y)\). Criterions for \(t_2(X)\simeq t_2(Y)\) and \(t_2(Y)=0\) are respectively given in the Section 2, using which, the followings are proved: (1) If \(X\) admits a Nikulin involution and \(\rho(X)=19,20\), then \(h(X)\) is finite-dimensional. (2) If \(X\) admits a Nikulin involution and \(h(X)\) is finite-dimensional, then, \(h(X)\simeq h(Y)\). (3) If \(X\) admits a non-symplectic involution acting trivially on algebraic cycles, and the order of \(H_X:=\text{Ker}(\text{Aut}(X)\to O(\text{NS}(X)))\) is not \(3\), then, \(h(X)\) is finite-dimensional. Proof of statement (1) uses the facts that \(X\) admits a Shioda-Inose structure so that \(Y\) is Kummer, that \(h(Y)\) is finite-dimensional, and that \(h(X)=h(Y)\oplus N\), where \(N\) is proved to be \(0\). Proof of statement (3) uses the facts that there exists a surjection \(F_m\to X\) from a Fermat surface \(F_m\) and that \(h(F_m)\) is finite-dimensional. If \(\rho(X)=9\) or \(X\) admits an elliptic fibration with certain automorphism, then, as examples in Section 4 show, \(h(X)\simeq h(Y)\) holds for several \(K3\) surfaces \(X\).