Stark's conjecture over complex cubic number fields (Q4813595)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 2090991
Language Label Description Also known as
default for all languages
No label defined
    English
    Stark's conjecture over complex cubic number fields
    scientific article; zbMATH DE number 2090991

      Statements

      Stark's conjecture over complex cubic number fields (English)
      0 references
      0 references
      0 references
      0 references
      13 August 2004
      0 references
      regulators
      0 references
      units
      0 references
      L-functions
      0 references
      This article is concerned with Stark's conjecture (rank-one abelian version) in an interesting and so far not very well-studied setting. More precisely, one considers an abelian \(G\)-extension \(K/k\) where the base field is complex cubic, that is, a (non-normal) cubic extensions of the rationals with exactly one real place and one complex place up to conjugation; it is further assumed that the real place of \(k\) is not totally split in \(K\), which forces \(G\) to have even order. Here the prediction made by Stark's conjecture in terms of L-values is precise enough to pin down the hypothetical unit \(\varepsilon\in K\) up to a root of unity, but in contrast with the situation where \(k\) is the rational field and \(K\) is totally real, it is much harder to obtain \(\varepsilon\) numerically since we only have the absolute values of its conjugates, which are in general truly complex numbers, so how should we choose the arguments of these complex numbers? This ``ambiguity'' does not contradict the fact that \(\varepsilon\) is essentially unique; the reason is that making wrong choices for these arguments would, most of the time, not even lead to an element of \(K\), let alone a unit. The rather clever approach of the authors is, roughly, the following: they use the quadratic subfield \(M\) of \(K\) over \(k\), and for this the Stark unit is known. This gives bounds on the coefficients of the minimal polynomial of \(\varepsilon\) over \(M\), hence a finite (but still impressive) number of candidates; it then happens that only for one among them, the conjecture comes out approximately right, and for this one, the conjecture is shown true to high numerical accuracy (28 digits). Fortunately, the relevant L-values have to be calculated just once for each example \(K/k\), and this is done by the method of Lavrik and Friedman. By this method, the authors succeed in verifying, with proved accuracy of at least 28 decimal digits, Stark's conjecture for all extensions \(K/k\) of the following type: \(k\) is complex cubic of discriminant up to \(-4300\) with class number 3, \(K/k\) is cyclic of degree 6, being the composite of the Hilbert class field \(H_k\) and the quadratic extension \(k(\sqrt{-\eta})\), where \(\eta\) is a fundamental unit of \(k\), taken to be positive at the real place. There are 58 such fields \(k\). (A single case, similar to the first case among those 58, had been done previously by Dasgupta, by less efficient methods. Otherwise there do not seem to be any previous calculations in this framework.) The resulting data are given in a table, and the calculation is explained with precision for one of the examples, the method being always the same. The paper in its entirety is highly readable.
      0 references

      Identifiers