A fake projective plane constructed from an elliptic surface with multiplicities \((2, 4)\) (Q763659)

A fake projective plane is a compact complex surface which shares the Betti numbers of the projective plane \(\mathbb P^2\) while not being isomorphic to \(\mathbb P^2\). These surfaces, automatically projective and of general type, have been known to exist since the first construction due to Mumford using \(p\)-adic uniformisation. Recently \textit{G. Prasad} and \textit{S.-K. Yeung} [Invent. Math. 168, No. 2, 321--370 (2007; Zbl 1253.14034)] classified all possible fundamental groups of fake projective planes. This gives rise to 4 alternatives of non-trivial automorphism groups: the abelian groups \(C_3, C_3^2, C_7\), and the unique non-abelian group \(G_{21}\) of order 21. Subsequently the author determined the possible structures of the quotient surfaces \(X/\text{Aut}(X)\) and their minimal resolutions. For \(G_{21}\), for instance, \(X/\text{Aut}(X)\) has three singular points of type \(\frac 13(1,2)\) and one of type \(\frac 17(1,5)\). The minimal resolution is a specific elliptic surface, namely of type \((2,3), (2,4),\) or \((3,3)\). The present paper gives a converse statement. Starting from a normal surface \(Z\) with the above singularities and the same Betti numbers as \(\mathbb P^2\), assume that minimal resolution of \(Z\) is a \((2,4)\) elliptic surface \(V\). Then the author computes the possible singular fibers of \(V\) and derives a fake projective plane by composing cyclic covers of \(Z\) of degree 3 and 7. It is pointed out that the explicit construction of a \((2,4)\) elliptic surface remains an open problem.