G-functions, motives, and unlikely intersections - old and new.
Yves André · 2025
Read G-functions, motives, and unlikely intersections - old and new. by Yves André (2025) 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-2501-09867.txt.
Introduction
G-functions are power series with deep ties to geometry, and Yves André, a solo author, surveys their role in arithmetic geometry, the field where number theory meets shapes. These functions solve the equations that govern how the periods of a family of shapes change as the family bends. Where two such functions meet in an algebraic relation, the meeting point often has extra hidden symmetry.
A principle from Bombieri caps how special such points can be, an idea that fed into a famous conjecture on unlikely meetings in geometry. André closes with the method's recent revival around the Zilber-Pink conjecture, a sweeping claim about when special points can appear. The survey links old results to the newest work.
Catalog record
| CREATOR | Yves André |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 21 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-09867
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
G-functions, motives, and unlikely intersections - old and new. Yves André Abstract . In this survey, we outline the role of G-functions in arithmetic geometry, notably their link with Picard-Fuchs differential equations and periods. We explain how polynomial relations between special values of G-functions arising from a pencil of algebraic varieties may occur at a parameter where the fiber has more “motivic” symmetries; and how Bombieri’s principle of global relations can be used to control the height of such parameters (which was also one of the origins of the André-Oort conjecture). At the end, we sketch the recent revival of the G-function method in the context of unlikely intersections and the Zilber-Pink conjecture. 1. G-functions and G-operators. (General references: [ 2 ] [ 22 ] [ 36 ] ]) 1.1. G-functions G-functions were introduced by C.-L. Siegel in his great memoir on diophantine approximation [ 37 ] . They are formal power series g(z)=\sum_{0}^{\infty}\,a_{n}z^{n} with coefficients in some number field K , which satisfy a linear differential equation with polynomial coefficients, and such that the conjugates of the a_{n} ’s as well as their denominators have at most an exponential growth in n . At the end of his memoir, Siegel stated without proof an irrationality property of some special values of G-functions. It took more than 50 years to get a proof of Siegel’s claim in full generality: this was achieved in E. Bombieri’s pioneering paper “On G-functions” [ 13 ] . 1.2. G-operators and geometry. Bombieri’s paper initiates the general study of linear differential operators satisfied by G-functions, building on p -adic techniques. Specifically, it focuses on differential operators \Lambda\in K\langle z,d/dz\rangle satisfying Bombieri’s condition: the product \prod_{p}R_{p}(\Lambda) of the p -adic radii of convergence at the generic disk is non-zero. Writing \Lambda in matrix form \frac{d}{dz}-A(z) , let T(z)\in K[z] be a common denominator for the entries of A , and \frac{1}{n!}(\frac{d}{dz})^{n}-A_{n}(z) stand for the unique left multiple of \Lambda of this form. It turns out that Bombieri’s condition is equivalent to the condition (introduced by A. Galochkin) that the common denominator of the coefficients of the entries of T(z)^{n}A_{n}(z) has an exponential growth in n . Such operators are called G-operators . They are fuchsian, with rational exponents ( cf. [ 27 ] ). An important theorem due to Chudnovsky asserts that for any G-function g(z) , a differential operator of minimal order which kills g(z) is always a G-operator 1 1 1 Bombieri’s proof of Siegel irrationality claim is conditional to the Bombieri’s condition, but Chudnovsky’s theorem allows to take it for granted . Conversely, any formal power series solution of a G-operator is a G-function 2 2 2 if 0 is a singularity, solutions involve powers of \log and fractional powers of z , but a similar statement also holds for the power series which “occur” in the solutions. . The significance of G-operators lies in their conjectural geometric origin (Bombieri-Dwork). Here, “ \Lambda is of geometric origin” can be understood as follows: \Lambda\in K\langle z,d/dz\rangle is a product of factors of Picard-Fuchs operators; equivalently, the associated differential module is an iterated extension of factors of differential modules of Gauss-Manin type. It was proven in [ 2 , V App.] that differential operators of geometric origin are G-operators . The converse remains very much open. The theory generalizes to several variables [ 8 ] [ 21 ] . 2. G-values and periods. (General references: [ 2 ] [ 7 ] [ 24 ] [ 23 ] ) 2.1. Periods Integrals of regular 1 -forms on an algebraic curve depend on the integration path, and the ambiguities become periods of abelian functions through Abel-Jacobi inversion. In spite of the lack of such an interpretation in higher dimensions, the term “period” happened to be used, more generally, for any integral \,\int_{\Delta}\omega\, of an algebraic n -form \omega taken on a domain \Delta bounded by algebraic equations. In a tradition which goes back to Euler, Legendre and Gauss, two special cases are prominent: - the arithmetic case (formalized in [ 29 ] ): (arithmetic) periods are complex numbers whose real and imaginary parts are integrals \int_{\Delta}\omega of an algebraic differential form \omega on an algebraic variety X defined over \mathbb{Q} , taken over a domain \Delta\subset X(\mathbb{R}) defined by polynomial inequalities with coefficients in \mathbb{Q} ; - the functional case: (functional) periods are integrals of differential forms \omega on semialgebraic domains \Delta , depending algebraically on a parameter z (the constant field being \mathbb{C} ). These multivalued functions satisfy fuchsian linear differential equations (Picard-Fuchs). They have been revisited by J. Ayoub, who proved that after performing an algebraic change of variable, \Delta can be taken to be the unit polydisk . Relations between periods have sparked speculations since Leibniz’s time, cf. [ 39 ] . In the functional case, the problem has been completely settled by Ayoub [ 9 ] [ 10 ] ( cf. also [ 11 ] ). In the arithmetic case, it remains very much open (conjectures of Grothendieck and Kontsevich-Zagier, cf. [ 2 , IX] [ 26 , XIII] [ 29 ] ). 2.2. G-functions and functional periods Let us consider the situation where the functional periods are integrals on the fiber of a smooth pencil X\to S defined a number field K\subset\mathbb{C} , z being a local coordinate on S . Then the Picard-Fuchs differential operator which controls the variation of the periods is a G-operator, whence a relation between functional periods and G-functions. Namely, writing the operator in matrix form d/dz-A(z) , and assuming for simplicity that z=0 is not a singularity, there is a fundamental matrix G(z) of solutions with G(0)=I with G-function entries G_{ij}(z) ; on the other hand, the period matrix P(z) is a fundamental matrix of solutions, so that G(z)=P(z)P(0)^{-1} (hence functional periods are linear combinations of G-functions with coefficients in the ring of arithmetic periods). In particular, for any specialization z\leadsto\zeta\in\bar{K} where G converges, the G-values G_{ij}(\zeta) are quotients of (arithmetic) periods. One can be more precise, using Ayoub’s reduction of the domain to a polydisk: any period is a G-value (Fresán, unpublished). The question of whether the converse holds is closely related to the Bombieri-Dwork conjecture. 2.3. Relations between periods, and relations between G-values 2.3.1. If for some point s\in S(\bar{K}) , the fiber X_{s} has more “symmetries” than the general fiber e.g. if it has more endomorphisms, in the usual or in the motivic sense - more precisely, if the motivic Galois of X_{s} is smaller than the motivic Galois of the general fiber - , then this yields polynomial relations between the periods P_{ij}(\zeta) of X_{s} , which do not come from polynomial relations between the functions P_{ij}(z) (and according to Grothendieck’s period conjecture, all such relations should come in this way). If moreover there are sufficiently many such relations, elimination of the entries of P(0) may yield polynomial relations between the G-values G_{ij}(\zeta) , which do not come from relations between the G-functions G_{ij}(z) . 2.3.2. A nice such setting arises when, instead of taking for 0 a non-singularity as above, one considers certain columns of G which are related to the notch 0 of the monodromy-weight filtration ( i.e. if one