Dedekind Sums, Shintani Generating Functions and Shintani Cocycles

(References in brackets are to my papers and preprints. Click for more details)

In 1976, Takuro Shintani gave explicit formulae for the values at negative integers of the partial ray-class zeta-functions modulo a conductor of a totally real field k. These can be interpreted in terms of generating functions associated to a `cone decomposition' of a fundamental domain for the action of a unit group on a lattice in R^n.

The initial aim of [8] was to make the cone decomposition more canonical for k real quadratic by considering the sequence of conductors fp^nO_k (for fixed f and prime p) which corresponds to a Z_p tower by class field theory. By examining the behaviour of the associated (2-variable) generating functions under `juxtaposition of cones', I proved that, as n increases, they tend to a limit in an appropriate p-adic normed space. The limiting function enjoys special properties and generates partial zeta-values at all levels in the tower. Article [9] uses the same juxtaposition property of generating functions (now associated to abstract plane lattices and cones) to give an algebraic construction of certain 1-cocycles on PGL_2(Q) which I called `Shintani Cocycles'. This leads to new algebraic proofs of the Petersson-Knopp identities and a reciprocity law for highly generalised Dedekind sums originally proved analytically by Halbritter. Article [10] establishes the non-triviality of these cocycles and examines some cohomologous variants. One is identical with the cocycles constructed by G. Stevens using an extension of modular symbols to the Borel-Serre compactification of the upper half-plane. Another variant is shown to be $p$-adically interpolable (after `smoothing') and so has applications to the construction of $p$-adic partial zeta-functions. previously, R. Sczech had also used also analytic methods to construct higher dimensional cocycles which, in dimension 2, are closely related to Stevens'. The joint paper [11] with S. Hu attempts to generalise the construction in dimension n>2 to produce Shintani (n-1)-cocycles on PGL_n(Q). We succeed for n=3 by dualising the parametrisation of the cones. However we obtain only a partial result for n>3 because of certain degenerate configurations. Subsequently R. Chapman produced reciprocity laws for higher-dimensional Dedekind using techniques inspired by our method. More recently R. Hill developed it considerably and succeeded in constructing higher-dimensional Shintani cocycles by using `algebraic infinitesimals' to cope with the degenerations (unpublished).

Back to my home page