Carlitz-Wan conjecture for permutation polynomials and Weil bound for curves over finite fields (Q1800713)

This is a very interesting article about the non-existence of permutation polynomials of degree \(n\) over \(\mathbb{F}_{q}\), for a large enough value \(q\) (\(q > C_n\)). Mullen's conjecture about this fact provided the bound \(C_n= n(n-2)\), but it was proven to be false. In this article, using results of algebraic geometry such as the Weil bound for estimating the number of rational points of algebraic curves, the bound \(C_n= n^2(n-2)^2\) is proposed. This improves the previous proposed bound of \(C_n= n^4\).{ }Section 1 of the article contains several facts about permutation polynomials, exceptional polynomials, and implications between them. A story about Carlitz conjecture on the non-existence of permutation polynomials over finite fields, and a generalization of it by Wan many years after (Carlitz-Wan conjecture) as described in the abstract. In addition, results on the maximal cardinality of the value set of a polynomial \(f\) over \(\mathbb{F}_{q}\) (\(|V_f|\)) are provided for subsequent results. { }In section 2, preliminary results about separable permutation polynomials, as well as maximal bound results on sums of products of positive integers and positive real numbers, are provided. Finally, a version of Weil bound that will be crucial to prove the main result is presented in this section.{ }The main result (Theorem 3.1), which is a improved version of theorem 1.2, is provided in Section 3. The arguments are similar of von zur Gathen's arguments that provided the bound \(C_n= n^4\). Using Weil bound results, the previous bound can be refined to \(C_n= n^2(n-2)^2\) by using a contradiction proof. The number of rational points in the components of algebraic curves, and Bezout's theorem applied to this curves lead to the contradiction that \(|V_f| < q\). Contradicting that \(f\) is a permutation.{ }Finally a remark is included presenting a slightly better (of complicated form) bound \(C_n\) for Theorem 3.1.