Cycle types of complete mappings of finite fields (Q2664512)

There is a much wider interest among algebraists to study complete mappings as a tool to construct orthogonal Latin squares. The existence of such mappings is also a fascinating area of study for many mathematicians. Let \((G, +)\) be an additive (though not necessarily abelian) group. A function \(f : G \longrightarrow G\) is called a \textit{complete mapping} of \(G\) if both \(f\) and \(f +id : g \longrightarrow f(g) +g\) are permutations of \(G\). In case \(G\) is the underlying additive group of a field \(K\), a complete mapping of \(G\) is also called a complete mapping of \(K\). For a divisor \(d\) of \(q - 1\), an \textit{index \(d\) first-order cyclotomic mapping} of \(\mathbb{F}_{q}\) is a function \(f : \mathbb{F}_{q} \longrightarrow \mathbb{F}_{q}\) such that \[ f(x) = \begin{cases} 0 & \quad\text{ if } x = 0 \\ a_{i}x & \quad\text{ if } x \in C_{i} \text{ for some } i = 0, 1, . . . , d - 1, \end{cases}. \] where \(a_{0}, a_{1}, . . . , a_{d-1} \in \mathbb{F}_{q}\) are fixed and \(C_{i} = \omega ^{i}C\) where \(\omega\) is a fixed primitive root of \(\mathbb{F}_{q}\) and \(C\) denotes the index \(d\) subgroup of \(\mathbb{F}^{*}_{q}\) . The three main theorems in the article are as follows: Theorem 1. Let \(d\) and \(n\) be positive integers, and let \( \varepsilon > 0\). There is a \(q_{0} = q_{0}(d, n, \epsilon) > 0\) such that for all prime powers \(q \geq q_{0}\) with \(q \equiv 1 \mod d\) and all \(c_{1}, c_{2}, \ldots, c_{n} \in \mathbb{F}_{q}\), every cycle type of a first-order cyclotomic permutation of \(\mathbb{F}_{q}\) of index \(d\) all of whose cycles on \(\mathbb{ F}_{q}^{*}\) are of length at least \( \varepsilon q\) can be realized as the cycle type of a first-order cyclotomic permutation \(f\) of \(\mathbb{F}_{q}\) such that \(f +c_{1} id, f +c_{2} id, \ldots, f +c_{n} id\) are permutations of \(\mathbb{F}_{q}\). Theorem 2. Let \(d\) and \(n\) be positive integers. There is a \(q_{1} = q_{1}(d, n) > 0\) such that for all prime powers \(q \geq q_{1}\) with \(q \equiv 1 (mod \ d)\), the following holds: Let \(c_{1}, c_{2}, \ldots, c_{n} \in\mathbb{F}_{q}\) be pairwise distinct, and let \(s_{1}, s_{2}, \ldots, s_{n}\) be functions \(M \longrightarrow M\) where \(M = \mathbb{F}_{q}^{*} /C\) is the set of all cosets in \(\mathbb{F}_{q}^{*}\) of the index \(d\) subgroup \(C\) of \(\mathbb{F}_{q}^{*}\). Then there is a first-order cyclotomic mapping \(f\) of \(\mathbb{F}_{q}\) of index \(d\) such that for all \(j = 1, 2, \ldots, n \) and all cosets \(\mathcal{C}\) of \(C\) in \(\mathbb{F}_{q}^{*}\) , one has \((f + c_{j} id)(\mathcal{C}) = s_{j}(\mathcal{C})\). Theorem 3. Let \(d > 1\) be an integer. There is a constant \(C = C(d) > 0\) such that if \(q\) is a prime power with \(q \equiv 1 \pmod d\) and \(q \geq C\), then the finite field \(\mathbb{F}_{q}\) admits a complete mapping \(f\) with \(f(0) = 0\) such that both \(f\) and \(f +id\) permute \(\mathbb{F}_{q}^{*}\) cyclically and \(f\) is not an additive function \(\mathbb{F}_{q} \longrightarrow \mathbb{F}_{q}\). The article spread over 34 pages has covered several technical concepts, theorems and lemmas from 36 references. Careful reading is required to understand the proof of the main theorems. Also, the authors have put forward three challenging open questions related to theorem 3 for further research.