Quaternions, polarizations and class numbers (Q4422798)
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 1969346
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Quaternions, polarizations and class numbers |
scientific article; zbMATH DE number 1969346 |
Statements
Quaternions, polarizations and class numbers (English)
0 references
24 August 2003
0 references
abelian variety
0 references
principal polarisation
0 references
quaternion algebra
0 references
class number
0 references
0.835187554359436
0 references
0.8182681798934937
0 references
0.8074117302894592
0 references
0.807081401348114
0 references
One motivation for this paper is the question of how many distinct principal polarisations an abelian variety may have. In general an abelian variety has, of course, no principal polarisation at all, and those that do mostly have only one. But it has been known since Humbert that there may be more, even for simple abelian surfaces. \textit{H. Lange} [Duke Math. J. 55, 617--628 (1987; Zbl 0657.14023)] went further in this direction, but also showed a uniform bound for the number of principal polarisations on an abelian variety with real multiplication. In contrast, a consequence of the results of this paper is that the number of principal polarisations on an abelian variety \(A\) may be arbitrarily large if the dimension of \(A\) is even. On the other hand, it is shown (by methods similar to Lange's) that if the dimension is odd and square-free, the number of principal polarisations cannot exceed \(2^{\dim A -1}\). To achieve this unbounded behaviour the author looks at abelian varieties with quaternionic multiplication. For such varieties, the main theorem here gives an exact formula for the number of principal polarisations in terms of class numbers. It is not immediately evident that this number may be arbitrarily large, but the author is able to prove this by an application of the Brauer-Siegel theorem, adapting an argument of \textit{K. Horie} and \textit{M. Horie} [Acta Arith. 55, No. 2, 157--170 (1990; Zbl 0659.12011)] who estimated relative class numbers for CM fields. The case of nonsquarefree odd dimension is left open. The author conjectures that the number of principal polarisations should in this case be possibly arbitrarily large. The suggestion is that in this case one should consider abelian varieties with endomorphism ring an order in a noncommutative division algebra over a CM field. In the case of dimension 2, it follows from the Torelli theorem and the results of this paper that there must exists arbitrarily large sets of nonisomorphic genus 2 curves having isomorphic unpolarised Jacobians. (Subsequent work of E. Howe provides infinitely many pairs of such curves, and also in genus 3, by means of real multiplication.) Another striking consequence of this paper, mentioned but not highlighted by the author, is that there exists an abelian variety \(A\) which is isomorphic (as unpolarised varieties) to its dual but which nevertheless does not possess a principal polarisation. To prove the results, the author considers an abelian variety \(A\) with quaternionic multiplication, that is, \({\text{End}} A\) is a hereditary order \(\mathcal O\) in totally indefinite quaternion division algebra \(B\) over a totally real number field \(F\) of degree \({{1}\over{2}}\dim A\). For such \(A\), he gives a description of the Néron-Severi group in terms of the arithmetic of \(B\). This leads to the notion of Pollack conjugation over \({\mathcal O}\), whereby two elements \(\mu_1,\mu_2\in B\) are equivalent if \(\mu_1=\bar\alpha\mu_2\alpha\) for some \(\alpha\in {\mathcal O}^*\). Isomorphism classes of line bundles, on the other hand, are related to certain ideals of \(F\) and to optimal embeddings in the sense of Eichler. To complete the counting, however, it is necessary to compare Pollack conjugation with conjugation in the sense of Eichler, and the technical core of the paper is a theorem, perhaps of more general interest, which does this in a slightly more general algebraic context.
0 references