Multiplicity estimates and degeneration (Q2436629)

While sharp zero estimates and multiplicity estimates are available for commutative linear algebraic groups, the known estimates for noncommutative groups \textit{M. Nakamaye} [Bull. Soc. Math. Fr. 123, No. 2, 155--188 (1995; Zbl 0841.11037)] are too weak to be useful in transcendence proofs, so far. In his paper [Acta Math. 175, No. 1, 49--73 (1995; Zbl 0833.11031)], \textit{D. Roy} devised a new transcendence approach to the problem of algebraic independence of logarithms of algebraic numbers, where he studied the points on varieties whose coordinates are logarithms of algebraic numbers. He constructed a new type of auxiliary function related with the action of \({\mathrm{GL}}_m({\mathbb{Z}})\) on \(\bigwedge^k{\mathbb{C}}^m\), the missing tool for obtaining new transcendence results being a zero estimate. In the present paper, the authors prove a corresponding multiplicity estimate related with the action of \({\mathrm{SL}}_2({\mathbb{Z}})\) on \({\mathbb{G}}_m^2\). The proof is based on the fact that a multiplicity estimate can be derived from a lower bound for Seshadri's constant. The authors prove that Seshadri's constant can only decrease under specialization. The new degeneration technique which is introduced in this paper may lead to further interesting applications in transcendence theory; a drawback of this argument is that the result involves a possible finite set of exceptions.