The classification of isotrivially fibred surfaces with \(p_{g} = q = 2\) (Q644466)

The classification of complex, projective surfaces is one of the fundamental questions of algebraic geometry, investigated initially by the classical Italian school and by many others since then. The structure of surfaces of general type, i.e., surfaces with maximal Kodaira dimension, is particularly subtle. It is known that for fixed value of the volume they are parametrized by a moduli scheme of finite type. However, the structure of this moduli space is very intricate. For example, it can have arbitrarily many connected components even after fixing the homeomorphism type by \textit{F. Catanese} [J. Differ. Geom., 24, No. 3, 395--399, (1986; Zbl 0621.14014)] and it can be arbitrarily singular by \textit{R. Vakil} [Invent. Math., 164, No. 3, 569--590 (2006; Zbl 1095.14006)]. One of the current goals of the theory of surfaces of general type is to classify surfaces with \(\chi =0\) (see for example the survey by \textit{I. Bauer, F. Catanese} and \textit{R. Pignatelli} [in: Global aspects of complex geometry. Berlin: Springer. 1--58 (2006; Zbl 1118.14041)]). Such surfaces have \(\rho_g=q\), and furthermore by the Bogomolov-Miyaoka-Yau and Debarre inequalities \(\rho_g \leq 4\). Hence there are four types of irregular surfaces of general type with \(\chi=0\) according to the value of \(\rho_g\). The \(\rho_g=4\) case happens only for product of curves of genus 2 by the work of A. Beauville (as an appendix to the article of \textit{O. Debarre} [Bull. Soc. Math. Fr., 110, No 3, 319--346 (1982; Zbl 0543.14026)]). The \(\rho_g=3\) case was classified by \textit{F. Catanese, C. Ciliberto, M. Mendes Lopes} [Trans. Am. Math. Soc., 350, No. 1, 275--308 (1998; Zbl 0889.14019)], \textit{C. Hacon, R. Pardini} [Trans. Am. Math. Soc., 354, No. 7, 2631--2638, (2002 ; Zbl 1009.14004)] and \textit{G. Pirola} [Manuscr. Math., 108, No. 2, 163--170 (2002; Zbl 0997.14009)]. The reviewed paper concerns the next natural step, the classification of surfaces of general type with \(q=\rho_g=2\). In particular, it finishes the classification for those that either admit an isotrival fibration over a smooth curve or that are isogenous to a product of curves of mixed type. Despite the subject of the paper being algebraic geometry, the used methods are very group theoretical. It turns out by some initial considerations, partially based on the works of \textit{F. Catanese} [Am. J. Math., 122, No. 1, 1--44 (2000; Zbl 0983.14013)], \textit{F. Serrano} [Ann. Mat. Pura Appl., 171, 63--81 (1996; Zbl 0884.14016)] and \textit{F. Zucconi} [Can. J. Math., 55, No. 3, 649--672 (2003; Zbl 1053.14042)] that one really has to classify adequate group actions on pairs of curves, which is then equivalent to finding Galois covers of their quotients. However, the latter is really a topological question, finding quotients of the fundamental groups of the quotient curves with the branch points removed. Furthermore, the \(\rho_g=2\) and \(q=2\) assumption forces all the numerical invariants to be small, leaving with only finitely many possibilities. The paper analyzes thoroughly all the possible configurations of the numerical invariants that yield the desired quotients of fundamental groups and then consequently the desired group actions. Finally, the possible cases of quotients of fundamental groups can still yield more than one connected components of the moduli spaces. It turns out that this number can be computed by group theoretic considerations as well, which is also done in the paper.