The mean values of cubic \(L\)-functions over function fields (Q2166207)

This paper aims to obtain an asymptotic formula for the mean value of \(L\)-functions associated with cubic characters over \(\mathbb{F}_q[T]\) when \(q \neq 0 \mod 3\). It is good to note that the literature about moments of cubic Dirichlet twists is somehow scarce even over number fields. The present paper considers the case over a function field in one indeterminate over a finite field. Even though the problem the authors are solving here was considered before by \textit{M. Rosen} in [CMS Conf. Proc. 15, 307--323 (1995; Zbl 0839.11059)] in a different fashion. The proofs of the main results rely on careful manipulation of cubic Gauss sums founded on the work of Kubota. This constitutes one of the novel features of this work. The main results consist in finding an asymptotic formula for the mean value of Dirichlet \(L\)-functions \(L_q (1/2, \chi)\) as \(\chi\) varies over the primitive cubic Dirichlet characters of \(\mathbb F_q[T]\). The authors achieve this task by finding an asymptotic formula for the first moment of cubic \(L\)-functions, that is the sum of \(L_q (1/2, \chi)\) when \(\chi\) varies through the primitive cubic character whose genus is \(g\) for a fixed \(g\). In the Kummer case though, this sum is only computed over characters whose restriction on \(\mathbb{F}_q^*\) are \(\chi_3\) discarding the other character to simplify the computations. The authors note that similar restrictions were made when such computations were performed in the literature under the quadratic case. The first moment of a cubic \(L\)-function in the non-Kummer setting (\(q \equiv 2 \mod 3\)) is solved in Theorem 1.1. The authors use the approximate functional equation to write \(L_q( 1/2, \chi)\) as the sum of the principal sum and the dual sum. The present work reveals some interesting cancellations between the main term and the dual term coming from the approximate functional equation of the \(L\)-functions. The formula obtained is the analog to the smoothed first moment obtained by \textit{S. Baier} and \textit{M. P. Young} in [J. Number Theory 130, No. 4, 879--903 (2010; Zbl 1204.11135)] for the number field case. Both papers use the approximate functional equation for the \(L\)-functions to obtain the main term and a dual term. In addition, both works reveal similar asymmetry of the sums of the cubic characters which are naturally a sum over the field to which we have adjoined the first root of unity and the truncated Dirichlet series of the \(L\)-function. As for the important differences, the present authors find an asymptotic formula for the first moment while Baier and Young chose to only bound the first moment in the number case. This delicate consideration builds on the work of \textit{J. Hoffstein} in [Invent. Math. 107, No. 1, 61--86 (1992; Zbl 0761.11024)] and \textit{S. J. Patterson} in [Glasg. Math. J. 49, No. 2, 243--255 (2007; Zbl 1160.11020)]. Those considerations are the one permitting them to exhibit some explicit and very interesting cancellation between the main terms and the dual terms. The author got inspired by the work of \textit{A. Florea} in [Forum Math. 29, No. 4, 873--892 (2017; Zbl 1429.11152)] to obtain this interesting cancellation. Even though it is worth noting that handling the cubic Gauss sums is significantly more difficult that the quadratic case considered by Florea. The cancellation is shown once the authors have obtained an exact formula for the residues of the generating series improving the work of \textit{G. Chinta} [Acta Arith. 132, No. 4, 377--391 (2008; Zbl 1165.11048)] by computing explicitly a certain constant. Moreover, whilst the authors of this paper use the Riemann hypothesis to bound the error term in the number field case such a tool was not necessary and even, in certain instances provided a weaker bound than the one found here. The first moment of a cubic \(L\)-function in the Kummer setting (\(q \equiv 1 \mod 3\)), Theorem 1.2, is then obtained in section 5. Theorem 1.2 is the first result when one considers all the primitive cubic characters. Surprisingly, it appears that the Kummer case is more difficult to obtain than the non-Kummer case and the error term is not as good as in Theorem 1.1. The size of the family of cubic twists in the Kummer case seems to build an obstruction to obtaining the cancellation as in the non-Kummer case and lead to a bigger error term. Another reason could be the lack of literature considering such a case. The authors prove that the Lindelöf hypothesis is valid for \(L\)-functions over function fields (Lemma 2.5) permitting them to bound the second moment. The authors also suggest some ways forward to improve their results and some explanations for the obtained results also serving as a check that the results are coherent with what is known in the literature. This paper definitely opens many doors to an interesting research path.