2-descent on elliptic curves and rational points on certain Kummer surfaces (Q2577004)

The main object of the paper under review is a Kummer surface \(S\) given as the minimal projective desingularization of \(Z^2=f^{(1)}(X)f^{(2)}(Y)\), where the \(f^{(s)}\) are quartic polynomials over a number field \(k\), having no repeated roots. The authors' goal is to obtain sufficient conditions for the Hasse principle to hold for \(S\). Their main result (Theorem 1) says that this is the case provided the Jacobians \(E^{(1)}\), \(E^{(1)}\) of the genus one curves \(D^{(1)}: U^2=f^{(1)}(X)\), \(D^{(2)}: V^2=f^{(2)}(Y)\) have all their 2-division points defined over \(k\), three additional arithmetic assumptions \(E\), \(Z_1\), and \(Z_2\) hold, and the Tate--Shafarevich groups of the \(E^{(s)}\) and their twists are finite. In the appendix, one can find a relationship between condition E and the algebraic part of the Brauer--Manin obstruction for \(S\) (Theorem A.1) as well as a relationship between conditions \(Z_1\) and \(Z_2\) and the transcendental Brauer-Manin obstruction (Theorem A.2). Note that in contrast with the earlier papers [\textit{J.-L.~Colliot-Thélène, A.~N.~Skorobogatov}, and \textit{P.~Swinnerton-Dyer}, Invent. Math. 134, 579--650 (1998; Zbl 0924.14011); \textit{P.~Swinnerton-Dyer}, Proc. London Math. Soc. 80 (3), 513--544 (2000; Zbl 1066.11029)], Schinzel's hypothesis does not appear among the assumptions. This important improvement continues the strategy developed by \textit{P.~Swinnerton-Dyer} [Ann. Sci. École Norm. Sup. 34 (4), 891--912 (2001; Zbl 1003.11028)]. As in many of the earlier papers, the authors use the fibration method. An interesting feature of the paper under review is that instead of looking at the pencil of genus one curves on the surface under considerations, they lift the surface to a threefold with a pencil of products of two genus one curves. Some refinements of the 2-descent method in sections 3 and 4 may be of independent interest.