Computation of genus 2 Kleinian hyperelliptic functions via Richelot isogenies
Matvey Smirnov · 2026
Read Computation of genus 2 Kleinian hyperelliptic functions via Richelot isogenies by Matvey Smirnov (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-23188.txt.
Introduction
Smirnov proposes an algorithm to compute genus-two Kleinian hyperelliptic functions. The method builds a chain of curves linked by Richelot isogenies. A recurrent procedure then reduces the work to a degenerate curve. At each step, one of fifteen possible partner curves must be chosen. Smirnov gives a choice rule that keeps the curve equations convergent. The same rule keeps weight-two Kleinian functions convergent under iteration. The algorithm evaluates the functions numerically.
Catalog record
| CREATOR | Matvey Smirnov |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 113940 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.23188
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Computation of genus 2 Kleinian hyperelliptic functions via Richelot isogenies Matvey Smirnov Abstract . In this work we propose an algorithm that numerically evaluates Kleinian hyperelliptic functions associated with a complex curve of genus 2. This algorithm is based upon constructing a sequence of curves with Richelot isogenous Jacobians and a recurrent procedures that reduces the calculation to a degenerate curve. As a part of mentioned algorithm we propose a method of choosing a Richelot isogenous curve (among 15 possibilities) that guarantees convergence of the equations of the curves and associated Kleinian functions of weight 2 under iterations. Keywords. Kleinian hyperelliptic functions, Richelot isogeny, Numerical algorithms. 1991 Mathematics Subject Classification 32A08, 32A10 1. Introduction This work is the final part of the cycle of papers that are devoted to giving an algorithm that computes Kleinian hyperelliptic functions associated with complex curves of genus 2. We refer to the previous parts [ 13 ] and [ 12 ] and references therein for context and motivation. At first let us recall some of the results of [ 13 ] and [ 12 ] (the precise facts and formulas will be stated in Section 2 ). In [ 13 ] we have associated a quadruple of special functions \mathscr{S}_{f}=(S^{f}\mathchar 59\relax S_{22}^{f}\mathchar 59\relax S_{12}^{f}\mathchar 59\relax S_{11}^{f}) (which we called Kleinian functions of weight 2) with any complex polynomial of degree 5 or 6 without multiple roots. These functions are closely related with the corresponding to f curve of genus 2 (namely, the curve given by the equation y^{2}=f(x) ). In particular these functions can be used to embed the Kummer surface of the mentioned curve into \CC\PP(3) . Moreover, in [ 13 ] we presented the explicit formulas that express classical Kleinian hyperelliptic functions \wp_{jk}^{f} , \zeta_{j}^{f} , and \sigma^{f} through \mathscr{S}_{f} . The paper [ 12 ] deals with the relation between \mathscr{S}_{f} and \mathscr{S}_{\hat{f}} , where \hat{f} is the equation of the curve, which has Richelot isogenous Jacobi variety with the Jacobian variety of the initial curve. In fact, we found an explicit expression of \mathscr{S}_{f} through \mathscr{S}_{\hat{f}} . Now, using the foregoing results we return to the idea of the numerical algorithm. That is, given f to compute \mathscr{S}_{f} at first we find a sequence of polynomials f^{(0)}=f , f^{(1)} , \dots , f^{(n)} , where each f^{(k)} is obtained via Richelot’s construction from f^{(k-1)} . If the functions \mathscr{S}_{f^{(n)}} can be somehow approximated, then, by using the results of [ 12 ] , we can calculate recursively \mathscr{S}_{f^{k-1}} through \mathscr{S}^{f^{(k)}} and, as the result, obtain an approximation of \mathscr{S}_{f} . The remaining difficulty is the requirement that \mathscr{S}_{f^{(n)}} can be effectively approximated. Even for the similar algorithms in genus 1 (i.e. Landen’s or AGM methods; see, e.g. [ 3 ] and [ 11 ] ) it is known that the effectiveness (and even convergence itself) heavily depends on the choice of the next isogenous curve on each step. For the Richelot’s construction there are 15 distinct choices of the next curve (in the generic case), corresponding to the partitions of the roots of f (counting \infty if \deg f=5 ) into three pairs. Thus, the missing part of the desired algorithm is the method to choose such partition on each step, such that the resulting sequence of polynomials \{f^{(n)}\} converges and yields convergent sequence of functions \{\mathscr{S}_{f^{(n)}}\} with explicitly computable limit. In this work we propose the following idea. Let D_{1}\mathchar 59\relax D_{2}\mathchar 59\relax D_{3}\subset\CC\PP(3) be disjoint open discs. We say that f is subordinate to the triple of discs (D_{1}\mathord{\mathchar 59\relax}D_{2}\mathord{\mathchar 59\relax}D_{3}) , if each of these discs contains exactly two roots of f . Clearly, the described situation naturally splits the roots of f into three pairs. We prove that the Richelot’s construction applied to f (with respect to the foregoing partition into three pairs) yields a polynomials that again does not have multiple roots and is subordinate to (D_{1}\mathord{\mathchar 59\relax}D_{2}\mathord{\mathchar 59\relax}D_{3}) . Moreover, the iterations of the described procedure form a sequence of polynomials f^{(n)} that quadratically converges to a polynomial with three double roots. Finally, we prove that the sequence \{\mathscr{S}_{f^{(n)}}\} of associated Kleinian functions of weight 2 converges (uniformly on compact sets), and we find an expression for the limit. Based on these results we propose an algorithm to numerically compute Kleinian functions associated with curves of genus 2. The paper is organized as follows. In Section 2 we collect all necessary statements and formulas from [ 13 ] and [ 12 ] . Section 3 contains the analysis of the Richelot’s construction (and its iterations) given that a polynomial is subordinate to a triple of disjoint discs. Then, in Section 4 , we prove that the corresponding Kleinian functions of weight 2 converge and find the expression for the limit. Finally, in Section 5 we formulate the algorithms. 2. Preliminaries 2.1. Curves of genus 2 and Kleinian functions We shall denote the space of complex polynomials of degree \leq k by \mathfrak{P}_{k} . Given a polynomial f we denote by f_{j} the coefficient of x^{j} in f . Moreover, we call a polynomial f\in\mathfrak{P}_{6} admissible if it does not have multiple roots and \deg f is equal to either 5 , or 6 . A polynomial f of degree 5 with leading coefficient f_{5}=4 we call a polynomial in Weierstrass form . With an admissible polynomial we associate a genus 2 curve \mathcal{X}_{f} (whose affine part is defined by the equation y^{2}=f(x) ). On this curve we fix the 1 -forms \omega_{1}^{f}=\frac{dx}{y}\mathord{\mathchar 59\relax}\;\;\omega_{2}^{f}=\frac{xdx}{y}\mathord{\mathchar 59\relax}\;\;r_{1}^{f}=\frac{f_{3}x+2f_{4}x^{2}+3f_{5}x^{3}+4f_{6}x^{4}}{4y}dx\mathord{\mathchar 59\relax}\;\;r_{2}^{f}=\frac{f_{5}x^{2}+2f_{6}x^{3}}{4y}dx. The forms \omega_{1}^{f}\mathchar 59\relax\omega_{2}^{f} constitute a basis of the space of holomorphic 1 -forms on \mathcal{X}_{f} . We denote the lattice of periods of forms \omega_{1}^{f}\mathchar 59\relax\omega_{2}^{f} by \Per_{f}\subset\CC^{2} . This lattice is canonically equipped with the intersection pairing \langle\cdot\mathchar 59\relax\cdot\rangle_{f} , induced from H_{1}(\mathcal{X}_{f}\mathchar 59\relax\ZZ) . Moreover, given a period w\in\Per_{f} we define \eta^{f}(w)\in\CC^{2} by the rule \eta^{f}(w)=-\int_{\gamma}\begin{pmatrix}r_{1}^{f}\\ r_{2}^{f}\end{pmatrix}\mathord{\mathchar 59\relax}\text{ where }w=\int_{\gamma}\begin{pmatrix}\omega_{1}^{f}\\ \omega_{2}^{f}\end{pmatrix}. The periods \eta^{f} are well-defined, since \rho_{1}^{f} and \rho_{2}^{f} are of the second kind. We recall [ 13 ] the definition of the space \mathfrak{S}_{f} of Kleinian hyperelliptic functions of weight 2 . It consists of all holomorphic functions \phi on \mathbb{C}^{2} that satisfy the property (2.1) \addcontentsline{lla}{section}{\numberline qSpaceDef}\phi(z+w)=\exp\left[2\eta^{f}(w)^{T}\left(z+\frac{w}{2}\right)\right]\phi(z) for all z\in\CC^{2} and w\in\Per_{f} . It is proved (see [ 13 , Proposition 3.1 (iv)] ) that this space is four-dimensional and an element \phi\in\mathfrak{S}_{f} is uniquely determined by its order 2 Taylor expansion at zero. Therefore, \mathfrak{S}_{f} is spanned by the four functions S^{f} , S^{f}_{11} , S^{f}_{12} , S^{f}_{22} , where S^{f} is the unique element of \mathfrak{S}_{f} that satisfies S^{f}(z)=z_{1}^{2}+\bar{o}(z^{2}) , and the other basis elements have the expansions (2.2) \addcontentsline{lla}{section}{\numberline qTaylorExpansions}S_{22}^{f}(z)=2z_{1}z_{2}+\bar{o}(z^{2})\mathord{\mathchar 59\relax}\;\;S_{12}^{f}(z)=-z_{2}^{2}+\bar{o}(z^{2})\mathord{\mathchar 59\relax}\;\;S_{11}^{f}(z)=1+\bar{o}(z^{2}). Note, that S_{jk}^{f}=S^{f}\wp_{jk}^{f} for all j\mathord{\mathchar 59\relax}k , where \wp_{jk}^{f} are the Kleinian \wp -functions associated with the curve \mathcal{X}_{f} (in fact, in the paper [ 13 ] the functions S_{jk}^{f} were defined using direct formulas involving \wp -functions, while the expansions ( 2.2 ) were derived from this definition, see [ 13 , Theorem 4.1] ; here we are going to use only the expansions ( 2.2 )). Finally, we need the relation between Kleinian functions of weight 2 and theta functions of weight 2 (see [