Motivic height zeta functions (Q2800420)

Results involving counting \(k\)-rational points of a variety \(V\) for a finite field \(k\) can often be generalized to infinite fields \(k\) and turned into geometric statements by using, instead of the cardinality \(\#V(k)\), the image of \(V\) in the Grothendieck ring \(\text{Var}_k\) of varieties over \(k\). In particular, this is the fundamental idea behind motivic integration: Computing a Lebesgue integral over, say, the \(p\)-adics involves counting over the residue field \(\mathbb{F}_p\); by working with images in \(\text{Var}_k\) instead, one obtains an integration theory for the power series fields \(k((t))\) for fields \(k\) of characteristic \(0\).NEWLINENEWLINEIn the present paper, this idea is applied to height zeta functions. Given a variety \(X\), a well-studied problem from diophantine geometry consists in determining the number of integral points \(X(\mathbb{Z})\) of bounded height. The height zeta function encodes how this number depends on the height bound.NEWLINENEWLINEBy applying the above idea to this, one obtains a height zeta function ``counting'' the points of \(X(k[t])\) of bounded height, where \(k\) is algebraically closed and of characteristic \(0\), and where counting now means taking the image in \(\text{Var}_k\). In this setting, the height of a point in \(X(k[t])\) is defined as some kind of degree. (All this is done in a somewhat more general setting, where \(k[t]\) is replaced by the regular functions on a smooth, quasi-projective curve.)NEWLINENEWLINEIt is not too difficult to verify that the set \(M_n \subset X(k[t])\) of points of height bounded by \(n\) can be considered as a constructible set over \(k\), so that one can define a Laurent series NEWLINE\[NEWLINE Z(T) := \sum_n [M_n]T^n \in \text{Var}_k[[T]]. NEWLINE\]NEWLINE The main result of the paper is a very precise description of this Laurent series for a specific class of varieties \(X\) (which in particular contain \(\mathbb{G}_a^n\) as a dense subvariety). In those cases, \(Z(T)\) is a rational function in \(T\), and the possible denominators are also specified. (They are the same as those of motivic Poincaré series.)