Values of hypergeometric functions (Q1105634)

The original goal of this paper was to prove transcendence of values of the Gaussian hypergeometric function \(F(a,b,c;z)\) \((a,b,c\in\mathbb Q)\) at algebraic points \(z\neq 0\). This could be realized by using a recent deep theorem of G. Wüstholz on the linear independence over \(\overline{\mathbb Q}\) of periods of Abelian varieties over \(\overline{\mathbb Q}\). For almost all choices of \(a, b, c\) this approach works and gives the desired transcendence results. However, as a beautiful bonus, the author discovered a certain set of triples \((a,b,c)\) for which \(F(a,b,c;z)\) assumes algebraic values at a set of algebraic points lying dense in \(\mathbb C\). I think the significance of these results justifies a short sketch of ideas in this review. Letting \(N\) be the common denominator of \(a, b, c\), the number \(F(a,b,c;z)\) can be written as a quotient \(\oint \omega_ z/\oint \omega_ 0\) where \(\omega_ z\) is a suitable \(l\)-form on the significant factor \(T_ z\) of the Jacobian of the curve \(y^N=x^A(x-1)^B(x-z)^C\). Here \(A, B, C\) depend on \(a, b, c\). Wüstholz's theorem states that if \(T_ z\) and \(T_ 0\) have no factors in common and if \(z\in \overline{\mathbb Q}\), then \(F(a,b,c;z)\) is transcendental over \(\mathbb Q\). The surprise is that there exist \(a, b, c\) such that \(T_ z\) has a factor in common with \(T_ 0\) for infinitely many algebraic \(z\). Families of such \(T_ z\) correspond to certain one-parameter Shimura families of Abelian varieties. In the present paper all these considerations are worked out quite explicitly and in a very clear style.