On inverted Kloosterman sums over finite fields

Xin Lin, Daqing WanarXiv 2023 · DOI · arXiv

TL;DR

This paper provides the first comprehensive complex and $p$-adic estimates for "inverted" $n$-variable Kloosterman sums, answering a 1995 challenge posed by Nicholas Katz. The authors demonstrate that these inverted sums are significantly more structurally complex than their classical counterparts, featuring larger cohomological dimensions and distinct $p$-adic slope sequences.

What Problem Does This Paper Address?

In number theory, Kloosterman sums are a fundamental type of exponential sum over finite fields, appearing frequently in the study of modular forms and analytic number theory. While "classical" $n$-variable Kloosterman sums are well-understood thanks to the work of Deligne (complex estimates) and Sperber ($p$-adic estimates), the "inverted" version remained largely mysterious for $n > 1$.

An inverted Kloosterman sum essentially places the standard Laurent polynomial used in the sum into the denominator of the additive character. Nicholas Katz raised the question of how to estimate these sums in 1995. The challenge lies in the fact that inverting the polynomial creates a rational function that is no longer a Laurent polynomial, making standard "toric" methods difficult to apply directly.

Methods

The authors employ a multi-pronged approach to tackle the problem:

  1. Elementary Method: They use the orthogonality of characters and properties of Gauss sums to derive an initial complex estimate.
  2. Cohomological Reduction: To get deeper results, they reduce the inverted sum to a "twisted toric exponential sum." This allows them to use powerful machinery from Adolphson-Sperber, Denef-Loeser, and Fu, provided the characteristic $p$ of the field does not divide $n+1$.
  3. Dwork’s $p$-adic Theory: To understand the $p$-adic valuations (the "slopes") of the associated L-function, the authors use Dwork’s trace formula and facial decomposition of Newton polyhedra.
  4. Newton and Hodge Polygons: They compare the Newton polygon (which tracks $p$-adic valuations) with the Hodge polygon (a lower bound derived from geometry) to determine exactly when the L-function is "ordinary."

Datasets

Not discussed in the paper. As a theoretical mathematics paper, it relies on formal proofs and algebraic structures rather than empirical datasets or benchmarks.

Key Insights and Ideas

  • Increased Complexity: The authors discover that the "primitive middle cohomology" associated with these sums has dimension $2n$, whereas the classical version has dimension $n+1$. Furthermore, the Hodge numbers are mostly 2 instead of 1. This geometric "bulkiness" explains why these sums are harder to bound.
  • The Main Term: Unlike classical sums where one expects pure "square-root cancellation," the inverted sums (when characters are trivial) possess a non-trivial main term $-q^{n-1}\chi_1(b)$. This means the sum is dominated by a term of order $q^{n-1}$ rather than $q^{n/2}$.
  • The $p \equiv 1 \pmod{n+1}$ Condition: The authors show that the Newton polygon coincides with the Hodge polygon (meaning the L-function is "ordinary") if and only if $p \equiv 1 \pmod{n+1}$. This provides a complete description of the $p$-adic slopes in this specific case.
  • Non-Degeneracy: They prove that the Laurent polynomial related to the inverted sum is non-degenerate if and only if $p \nmid (n+1)$, which is the key requirement for applying deep cohomological results.

Key Results

  • Complex Bounds: The authors establish a sharp upper bound for the sum $S_n(\chi, b)$. Specifically, they show that after accounting for the main term, the error term is bounded by $(2n+1)q^{n/2}$ (Theorem 1.2).
  • L-function Rationality: They prove that the L-function associated with these sums is a rational function and determine its precise form (Theorem 1.3).
  • $p$-adic Slope Sequence: For the case $p \equiv 1 \pmod{n+1}$, they determine the exact sequence of $p$-adic valuations for the roots of the L-function: $\{0, 1, 1, 2, 2, \dots, n-1, n-1, n\}$.

Limitations and Future Work

  • The Singular Case: The results largely assume $p \nmid (n+1)$. If $p$ divides $n+1$, the underlying polynomial becomes "degenerate," and the standard toric methods fail. The authors note that a different approach is needed for this singular case.
  • General $p$-adic Slopes: While the $p$-adic slopes are fully determined for $p \equiv 1 \pmod{n+1}$, they remain unknown for other primes. In those cases, the Newton polygon sits strictly above the Hodge polygon, making the valuations harder to calculate.
  • Optimal Constants: While they provide square-root cancellation, the authors suggest that the constants in their bounds (like $2n+1$) might be further improvable with different techniques.

Summary updated

Abstract

The classical $n$-variable Kloosterman sums over finite fields are well understood by Deligne's theorem from complex point of view and by Sperber's theorem from $p$-adic point of view. In this paper, we study the complex and $p$-adic estimates of inverted $n$-variable Kloosterman sums, addressing a question of N. Katz (1995). We shall give two complex estimates. The first one is elementary based on Gauss sums. The second estimate is deeper, depending on the cohomological results of Adolphson-Sperber, Denef-Loeser and Fu for twisted toric exponential sums. This deeper result assumes that the characteristic $p$ does not divide $n+1$. Combining with Dwork's $p$-adic theory, we also determine the exact $p$-adic valuations for zeros and poles of the L-function associated to inverted $n$-variable Kloosterman sums in the case $p \equiv 1 \mod (n+1)$. As we shall see, the inverted $n$-variable Kloosterman sum is more complicated than the classical $n$-variable Kloosterman sum in all aspects in the sense that our understanding is less complete, partly because the Hodge numbers are now mostly $2$ instead of $1$.
BibTeX
@misc{lin2023inverted,
  author = {Xin Lin and Daqing Wan},
  title = {{On inverted Kloosterman sums over finite fields}},
  year = {2023},
  eprint = {2301.04287},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2301.04287},
}
Figure