The maximum number of systoles for genus two Riemann surfaces with abelian differentials (Q2421728)

``The \textit{systolic length} of a length space \((X,d)\) is the infimum of the lengths of non-contractible loops in \(X\) '', so the beginning of this long paper. They consider such loops -- closed geodesics, called \textit{systoles} -- on compact Riemann surfaces, mainly of genus \(2\), not equipped with a hyperbolic metric but with the flat metric induced by a holomorphic 1-form \(\omega\). Systoles for hyperbolic metrics have been treated previously by \textit{F. Jenni} [Comment. Math. Helv. 59, 193--203 (1984; Zbl 0541.30034)] or in many papers by \textit{P. Schmutz Schaller} [Bull. Am. Math. Soc., New Ser. 35, No. 3, 193--214 (1998; Zbl 1012.30029)]. Here, the focus lies on the question of how many non-homothetic systoles can exist on such translation surfaces \((X,\omega)\), mainly for genus \(2\). The main theorem says that in genus \(2\) the maximal number of systoles is \(10\), and that this number occurs in precisely one case up to isometry. As Riemann surface this \(X\) is in fact the Bolza surface and has \(12\) systoles in hyperbolic geometry, there also the maximal possible number. This example maximizes the number of systoles but not at all the systolic length (for fixed volume of \(X\)), as shown by an example (Fig. 2, Conj. 1.2). For translation surfaces the proof needs a natural case-by-case analysis: if \(\omega\) has one double zero, the proof is not too hard using the \textit{Delauney decomposition}. The idea generalizes to higher genera \(g \ge 2\): if \(\omega\) has exactly one zero, \((X,\omega)\) has at most \(6g-3\) systoles. For hyperelliptic \(X\) this number can even strengthend to \(6g-5\). This number is attained if and only if there are two pairs of Delauney cells interchanged by the hyperelliptic involution. For \(g=2\) and if \(\omega\) has two different zeros, the proof becomes far more complicated because many different topologically and geometrically possible configurations of systoles, Weierstrass points and zeros of \(\omega\) on \(X\) have to be distinguished. The typical questions considered in this well written paper are the following. Do systoles always join Weierstrass points? How many systoles can cross a given Weierstrass point? Do systoles cut the surface in two connected components? Do systoles cross one or both zeros of \(\omega\)? The paper answers all these questions and reduces the proof to sophisticated, but relatively elementary arguments using straight lines and angles in flat parts of \(X\) -- after decomposition of \(X\) in suitable such regions. Remarkable: outside this 10-systole example on Bolza's surface the maximal number of systoles is 8, not 9.