Thomae type formulae for singular \(Z_N\) curves (Q2457222)

Thomae's formulas are a classical expression for functions of the branchpoints of a hyperelliptic curve in terms of the thetanulls. The formulas have been recently applied to many important areas of mathematical physics, and a wide spectrum of references is given in the paper under review. In the late 1980s, \textit{M. Berdshasky} and \textit{A. Radul} [``Conformal field theories with additional \(Z_N\) symmetry'', Int. J. Mod. Phys. A 2, No. 1, 165--178 (1987); Commun. Math. Phys. 116, No. 4, 689--700 (1988; Zbl 0657.14019)] gave the first modern generalization of Thomae's formulas, for cyclic curves. There have been others, and again references can be found in the paper under review, but not these three significant advances in the algebro-geometric interpretation and generalization of the formulas: (1) \textit{A. Eisenmann} and \textit{H. M. Farkas}, ``An elementary proof of Thomae's formulae'', Online J. Anal. Comb. No. 3, Art. 2 (2008), (2) \textit{Y. Kopeliovich}, Int. J. Number Theory 4, No. 5, 725--733 (2008; Zbl 1160.14019), and ``Theta constants identities for Jacobians of cyclic 3-sheeted covers of the sphere and representations of the symmetric group'', \url{arXiv:0704.1032}; (3) \textit{N. I. Shepherd-Barron}, ``Thomae's formulae for non-hyperelliptic curves and spinorial square roots of theta-constants on the moduli space of curves'', \url{arXiv:0802.3014}, the latest one of which gives the formulas in principle for any curve. In the paper under review, the authors derive an explicit proof for the Bershadsky-Radul version in the case when the \(Z_N\)-curve \[ C_{N,m}=\{(\lambda,y)\text{ is singular}: y^N=(\prod_{k=1}^m(\lambda-\lambda_{2k}))^{N-1}\prod_{k=0}^m(\lambda -\lambda_{2k+1})\}. \] The authors had given an explicit solution for the Riemann-Hilbert problem on this curve [\textit{V. Z. Enolski} and \textit{T. Grava}, Int. Math. Res. Not. 2004, No. 32, 1619--1683 (2004; Zbl 1104.34061)], for a monodromy consisting of quasi-permutation matrices around the \(\lambda\)'s, in terms of Szegö kernels and \(\vartheta\) functions on \(C_{N,m}\). The technique used in this paper is the same as Thomae's, as opposed to the algebraic technique of Eisenmann-Farkas, \textit{loc. cit.}, that controls linear series on the curve. To compare derivatives in moduli with the ones in abelian coordinates, the authors use, besides the heat equation, the Rauch variation formula: \[ {\partial\over\partial\lambda_k}\Pi_{ij}= 2\pi\imath\text{Res}_{| \lambda =\lambda_k}\left( {1\over (d\lambda_k (P))^2}\sum_{s=1}^Ndv_i(P^s)dv_j(P^s)\right), \] where \(\Pi\) is the period matrix, \(dv_i\) are normalized holomorphic differentials, and the points \(P^s\) sweep a neighborhood of a branchpoint on the \(s\)-th sheet. The authors give Thomae-type formulas for specific non-special divisors on this singular curve, classified by \textit{G. González Díez} [Proc. Lond. Math. Soc., III. Ser. 62, No. 3, 469--489 (1991; Zbl 0679.14010)]; other divisors would require more complicated calculations of the same type. In the most general theorem of this paper they give an expression for \(\vartheta[e_m]\) (\(m\) denotes roughly speaking a suitable partition of \(N\)) in terms of the \(\lambda\)'s and of the \(a\)-periods of a basis of non-normalized holomorphic differentials \(du_i\).