scientific article; zbMATH DE number 4027581

From MaRDI portal
Publication:3768983





[For the entire collection see Zbl 0621.00008.] These two papers contain a significant new proof of some of Baker's theorems on linear forms in logarithms. Let \(\alpha_ 1,...,\alpha_ n\) be non-zero algebraic numbers of heights at most \(A_ 1\geq 4,...,A_ n\geq 4\) respectively, and let \(\beta_ 0,\beta_ 1,...,\beta_ n\) be algebraic numbers of heights at most \(B\geq 4\). Suppose \(\Lambda =\beta_ 0+\beta_ 1 \log \alpha_ 1+... +\beta_ n \log \alpha_ n\neq 0\) for some determination of the logarithms. Then we have the lower bound \[ (*)\quad | \Lambda | \geq (B\Omega)^{-c\Omega \log \Omega}, \] where \(\Omega =\log A_ 1... \log A_ n\) and \(c>0\) depends only on the choice of logarithms and the degree of the field generated over the rationals by \(\alpha_ 1,...,\alpha_ n,\beta_ 0,\beta_ 1,...,\beta_ n.\) This was proved by \textit{A. Baker} [e.g., Transcend. Theory Adv. Appl. 1976, 1-27 (1977; Zbl 0361.10028)]; even with log \(\Omega\) replaced by log(\(\Omega\) /log A), where \(A=\max (A_ 1,...,A_ n)\). An essential ingredient of the proof is the use of Kummer theory. The present author establishes (*) using, instead of Kummer theory, some multiplicity estimates in the style of his ground-breaking (though still unpublished) paper ``Multiplicity estimates on group varieties. However, the latter has first to be generalized to multiprojective space, in order to estimate the multidegrees of the ``obstruction subgroup. This is carried out for the linear group \({\mathbb{G}}_ a\times ({\mathbb{G}}_ m)^ n\) in Part I of the papers under review. The result is a little technical to be quoted here (compare also recent work of \textit{P. Philippon} [Bull. Soc. Math. Fr. 114, 355-383 (1986; Zbl 0617.14001)]). In Part II comes another idea; the author cleverly uses the geometry of numbers to estimate the equations defining the obstruction subgroup in terms of its multidegrees. This leads to a proof of (*) by induction on n. The author mentions that his methods will also prove Baker's refinement of (*) when \(\beta_ 1,...,\beta_ n\) are rational integers and \(\beta_ 0=0\); then the \(B\Omega\) term can be replaced by B. Recently there has been a great deal of extra activity in this field; see the forthcoming 1986 Durham Conference Proceedings ``New advances in transcendence theory, A. Baker (ed.), Cambridge University Press.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3768983)