Wolfgang A. Schmid (CMLS, École polytechnique, Palaiseau)


On zero-sum problems related to counting functions of algebraic integers with
certain arithmetical properties

One classical way to extend the Prime Number Theorem and related results from the integers to the ring of
integers of an algebraic number field is replacing the notion "prime number" by the notion "prime ideal."

Yet, also problems of the following form received considerable attention (since the 1960s). Given some
number field K with ring of algebraic integers R and AR a subset defined via some arithmetical constraint
(fulfilling certain conditions), e.g., the set of all irreducibles in R, or the set of all elements having an essentially
unique factorization into irreducibles in R. Let A(x) denote the number of pairwise non-associated elements in
A of total norm at most x. Determine the asymptotics of A(x).

It is known, by the work of Rémond, Narkiewicz, Halter-Koch and others, that the order of magnitude of A(x)
is equal to (x/(Log x)B)(Log Log x)C where B and C depend on the ideal class group of R only, namely they
are determined by the solution to a certain zero-sum problem over the class group. However, for many types
of groups these zero-sum problems are open.

First, we summarize why zero-sum problems over the class group arise in this context, and we recall the precise
nature of these zero-sum problems (for certain sets A).

Then, we present recent results on these zero-sum problems, including results on the generalized Davenport
constants, thus determining or estimating the numerical value of these constants B and C in certain cases.