Measures of simultaneous approximation for quasi-periods of Abelian varieties (Q1609540)

A measure of simultaneous approximation for a family of complex numbers \(\theta_1,\dots,\theta_n\) is a lower bound for the distance from \((\theta_1,\dots,\theta_n)\) to an arbitrary point \((\alpha_1,\dots,\alpha_n)\in \overline{{\mathbb Q}}^n\). More precisely, it is defined as a function \(\phi \colon {\mathbb N}\times{\mathbb R}_+ \to {\mathbb R}_+\) such that, for an appropriate choice of constant \(C>0\), we have \[ \log\max |\theta_i-\alpha_i|\geq -C\phi(D,h) \] for any point \((\alpha_1,\dots,\alpha_n)\in \overline{{\mathbb Q}}^n\), any upper bound \(D\geq C\) on the degree of \({\mathbb Q}(\alpha_1,\dots,\alpha_n)\) over \({\mathbb Q}\), and any upper bound \(h\geq C\) on \(h(\alpha_1), \dots, h(\alpha_n)\), the absolute logarithmic heights of the \(\alpha_i\)'s. Comparing such a measure with results of simultaneous approximation may lead, under certain conditions, to a lower bound for the transcendence degree of the field \({\mathbb Q}(\theta_1, \dots,\theta_n)\) over \({\mathbb Q}\) [see for example \textit{D. Roy} and \textit{M. Waldschmidt}, Ramanujan J. 1, 379-430 (1997; Zbl 0916.11042)]. In the present paper, the author establishes measures of simultaneous approximation for quasi-periods of Abelian varieties from which he deduces results of algebraic independence generalizing work of \textit{G. V. Chudnovsky} about elliptic curves [see Chap. 7 of Contributions to the theory of transcendental numbers, Am. Math. Soc., Providence, RI (1984; Zbl 0594.10024)]. Let \(A\) be an abelian variety of dimension \(g\) defined over a subfield \(K\) of \({\mathbb C}\), let \(T_0A\) be its tangent space at the origin, let \(\exp_A\colon T_0A({\mathbb C})\to A({\mathbb C})\) be its exponential map, let \(\omega_1,\dots, \omega_{2g}\) denote a basis for the space \(H^1_{\text{ DR}}(A,K)\) of \(K\)-rational classes of differentials of the second kind on \(A\), and let \(u_1,\dots,u_r\) be \({\mathbb Q}\)-linearly independent elements of \(T_0A({\mathbb C})\) such that \(\exp_A(u_i)\in A(K)\) for \(i=1,\dots,r\). As a consequence of its main result, the author shows: (1) If \(K\subset\overline{{\mathbb Q}}\), the family of integrals \(\int_0^{u_j} \omega_i\) with \(1\leq i\leq 2g\) and \(1\leq j\leq r\) admits the approximation measure \[ D^{(3/2)+(g/r)}(\log D)^{-1/2}(D^{2g/r}+h). \] (2) If \(u_1,\dots,u_r\) are periods of \(A\), the same set of integrals together with a generating system of \(K\) over \({\mathbb Q}\) admits the approximation measure \[ [D(h+\log D)]^{3g/r}. \] In the case (1), he deduces that the \(2gr\) integrals generate a field of transcendence degree at least \(2\) when \(r\geq 2g\). For \(g=1\), this means that if \(A\) is an elliptic curve with algebraic invariants \(g_2,g_3\) and Weierstrass functions \(\wp, \zeta\), and if \(u,u'\) are linearly independent complex numbers such that \(\wp(u),\wp(u')\in\overline{{\mathbb Q}}\) then at least two of the numbers \(u,u',\zeta(u),\zeta(u')\) are algebraically independent over \({\mathbb Q}\). In the case (2), assuming \(K\subset \overline{{\mathbb Q}}\), the more precise main result (Prop.\ 6.1) of the author implies the same conclusion provided simply that \(r>g\). However, for \(g=1\) and \(r=2\), the above measure implies that among the invariants \(g_2,g_3\), the periods and the quasi-periods of the elliptic curve \(A\), there are at least two algebraically independent numbers over \({\mathbb Q}\). The proof of the main theorem involves looking at the extension of \(A\) by \({\mathbf G}_a^{2g}\) associated with a basis of \(H^1_{\text{ DR}}(A,K)\). The author describes a projective embedding of such an extension in terms of ``algebraic'' theta functions. The formulas that he derives for addition, multiplication and differentiation are of independent interest. The author also proves some \(G\)-function property for the coordinates functions, which allow him to generalize what he calls the Baker-Coates-Anderson-Chudnovsky trick. The last section of this paper concludes with several related open problems and partial answers. In particular, Remark 7.1 of the paper proposes a measure of approximation which implies the full statement of Corollary 1.1 (2). The latter also follows from Proposition 7.3 (with \(q=2g\)), a result which is proven in complete details as Theorem 3.11.1 of the Ph.D. thesis of the author.