Abstract:We study the Gauss and Jacobi sums from a viewpoint of motives. We exhibit isomorphisms between Chow motives arising from the Artin-Schreier curve and the Fermat varieties over a finite field, that can be regarded as (and yield a new proof of) classically known relations among Gauss and Jacobi sums such as Davenport-Hasse's multiplication formula. As a key step, we define motivic analogues of the Gauss and Jacobi sums as algebraic correspondences, and show that they represent the Frobenius endomorphisms of such motives. This generalizes Coleman's result for curves. These results are applied to investigate the group of invertible Chow motives with coefficients in a cyclotomic field.
What problem does this paper attempt to address?
This paper attempts to study Gauss sums and Jacobi sums from the perspective of motives. Specifically, the paper shows the isomorphic relationships constructed between Artin - Schreier curves and Fermat varieties over finite fields. These relationships can be regarded as new proofs of the classically known relationships between Gauss sums and Jacobi sums, such as the Davenport - Hasse multiplication formula, etc. By defining the motivic analogues of Gauss sums and Jacobi sums as algebraic counterparts and showing that they represent the Frobenius endomorphisms of these motives, the authors generalize Coleman's results on curves. These results are applied to study the group of invertible Chow motives with coefficients in cyclotomic fields.
In short, the main objectives of the paper are:
1. **Upgrading classical relationships to the motivic level**: By constructing appropriate isomorphisms between motives and connecting the Frobenius endomorphisms with the motivic Gauss sums and Jacobi sums, the classically known relationships between Gauss sums and Jacobi sums are promoted to the motivic level.
2. **Applying motivic relationships**: Using these motivic relationships, study the properties of the group of invertible Chow motives with coefficients in cyclotomic fields, especially the generators and relationships of these motive groups.
### Specific problem analysis
#### Definitions of Gauss sums and Jacobi sums
- **Gauss sums**:
\[
g(\psi, \chi)=-\sum_{m\in\kappa^{*}}\psi(m)\chi(m^{q - 1}/d)\in\mathbb{Q}(\zeta_{pd})
\]
where \(\psi\) is a non - trivial additive character of \(\kappa\) and \(\chi\) is a multiplicative character of \(\mu_d\).
- **Jacobi sums**:
\[
j(\chi_1,\ldots,\chi_n)=(- 1)^{n - 1}\sum_{m_i\in\kappa^{*},\sum_{i = 1}^n m_i=1}\chi_1(m_1^{q - 1}/d)\cdots\chi_n(m_n^{q - 1}/d)\in\mathbb{Q}(\zeta_d)
\]
where \(\chi_1,\ldots,\chi_n\) are multiplicative characters of \(\mu_d\).
#### Main theorems
- **Theorem 1.1**:
- If \(\psi\) and \(\chi_1,\ldots,\chi_n,\prod_{i = 1}^n\chi_i\) are all non - trivial, then there exists the following isomorphism:
\[
\bigotimes_{i = 1}^n h(A_d)(\psi,\chi_i)\cong h(A_d)(\psi,\prod_{i = 1}^n\chi_i)\otimes h(F_d^n)(\chi_1,\ldots,\chi_n)
\]
- Suppose \(n\) divides \(d\), and \(\alpha\) is a character of \(\mu_d\) such that \(\alpha^n\neq1\), then there exists the following isomorphism:
\[
h(F_d^n\langle n\rangle)(\alpha,\ldots,\alpha)\cong\bigotimes_{\chi^n = 1,\chi\neq1}h(F_d^2)(\alpha,\chi)
\]
- **Theorem 1.2**:
- If \(\chi_1,\ldots,\chi_n,\prod_{i = 1}^n\chi_i\)