Elliptic (p,q)-difference modules

From MaRDI portal
Publication:2045161



Abstract: We study finite dimensional vector spaces over fields of elliptic functions equipped with two commuting aotomorphisms sigma and au induced by isogenies of relatively prime orders. We give a structure theorem for such objects, that reveals a connection to the classification of vector bundles on elliptic curves by Atiyah. As an application we prove an elliptic analogue of a conjecture of Loxton and van der Poorten which has been recently proved by Adamczewski and Bell.


Let \(p\) and \(q\) be multiplicatively independent positive integers and consider the so-called Mahler operators \(x\mapsto x^p\) and \(x\mapsto x^q\). A power series \(F(x)\in Kx\) satisfies a \(p\)-Mahler fuctional equation if the vector space generated by \(F(x), F(x^p), F(x^{p^2}), \ldots\) has finite dimension. \textit{B. Adamczewski} and \textit{J. P. Bell} [Ann. Sc. Norm. Super. Pisa, Cl. Sci. (5) 17, No. 4, 1301--1355 (2017; Zbl 1432.11086)] proved that a power series \(F(x)\) satisfies both a \(p\)- and a \(q\)-Mahler functional equation if and only if it is a rational function. \textit{J.-P. Bézivin} and \textit{A. Boutabaa} [Collect. Math. 43, No. 2, 125--140 (1992; Zbl 0778.39009)] gave a similar result for dilations \(x\mapsto px\) and \(x\mapsto qx\). This paper considers similar objects called elliptic \((p,q)\)-difference modules for an elliptic curve \(\mathbb{C}/\Lambda\), where \(\Lambda\) is a lattice, and the field of \(\Lambda\)-elliptic functions. These are finite-dimensional vector spaces with commuting automorphisms given by \(\sigma f(z)=f(z/p), \tau f(z)=f(z/q)\) on the field of \(\Lambda\)-elliptic functions. The main theorem is a structure theorem for elliptic \((p,q)\)-difference modules in the case when \(p\) and \(q\) are relatively prime. As a consequence, it is shown that a power series which satisfies simultaneously a \(p\)-difference equation and a \(q\)-difference euqation with coefficients Laurent expansions of \(\Lambda\)-elliptic functions lies in the ring generated over the field of \(\Lambda'\)-elliptic functions by \(z^{\pm1}\) and \(\zeta(z,\Lambda')\) for some lattice \(\Lambda'\subset \Lambda\). Here, \(\zeta(z,\Lambda)\) is the Weierstrass zeta function for the elliptic curve. Conversely, every function from this ring satisfies a \(p\)-difference equation and a \(q\)-difference equation with elliptic functions as coefficients. The derivation explains the appearance of \(\zeta(z,\Lambda)\) in the elliptic case, unlike the cases for the simpler additive and multiplicative groups. In the 1930s, Mahler built a method for studying the transcendence of functions satisfying functional equations in the multiplicative case \(x\mapsto x^p\) and its higher-dimensional generalisations. These ideas found applications in computability because the generating functions of sets recognised by finite automata are related to Mahler functions. In the 1960s, Cobham proved that if a set is recognised by finite automata in base \(p\) and in base \(q\) then it is a union of a finite set and a finite number of arithmetic progressions. However, the precise link between finite automata and Mahler functions was elusive. Difference algebra, as further developed in this paper, has delivered the recent progress in this area.











This page was built for publication: Elliptic \((p,q)\)-difference modules

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2045161)