The probability of rectangular unimodular matrices over \(\mathbb F_q[x]\) (Q1938585)

Let \(R\) be a ring and let denote by \(M_{k\times n}(R)\) the set of all \(k\times n\) matrices over \(R\). A matrix \(A\in M_{k\times n}(R)\) is called unimodular if it can be extended to an \(n\times n\) nonsingular matrix. When \(R= \mathbb{Z}\), the probability that a random \(k\times n\) matrix is unimodular is given by \(\prod_{j=n-k+1}^{n} \zeta (j)^{-1},\) where \(\zeta (z)\) is the Riemann's zeta function. This result has been proved by \textit{G. Maze, J. Rosenthal} and \textit{U. Wagner} [Linear Algebra Appl. 434, No. 5, 1319--1324 (2011; Zbl 1211.15044)]. It constitutes an extension to matrices of the well-known Dirichlet density theorem for the probability of \(n\) integers are co-prime, see [\textit{G. L. Dirichlet}, ``Über die Bestimmung der mittleren Werte in der Zahlentheorie'', Abhand. König. Preuss. Akad. Wiss. 69--83 (1849); G. Lejeune Dirichlet's Werke. Erster und zweiter Band. Bronx, N. Y.: Chelsea Publishing Company (1969; Zbl 0212.00801)]. The aim of the paper under review is to extend the above results to the polynomial ring \(R= \mathbb{F}_{q}[x]\), where \(\mathbb{F}_{q}\) is a finite field consisting of \(q\) elements, for any prime power \(q\). Let denote by \(\mathcal{M}_{N}(R)\) the subset of \(M_{k\times n}(R)\) consisting of all matrices with entries in \(\{f_{0}(x),\dots, f_{N}(x) \}\). For any \(S\subseteq M_{k\times n} (R),\) the natural density of \(S\) in \(M_{k\times n}(R)\) is defined as \(D(S)= \lim_{N\rightarrow \infty} \frac {|S \cap \mathcal{M}_{N}(R)|}{ | \mathcal{M}_{N}(R)|}\) if the above limit exists. Notice that \textit{K. E. Morrison} [Random polynomials over finite fields (1999), available from {\url{http://www.calpoly.edu/~kmorriso/Research/RPFF.pdf}}], has introduced the concept of natural density of a subset of \(\mathbb{F}_{q}[x]\) in \(\mathbb{F}_{q}[x]\) and its value has been done. Indeed, let \(E\) be the set of all \(k\times n\) unimodular matrices over \(\mathbb{F}_{q}[x]\). Then the authors prove that if \(k<n\), the natural density of \(E\) is \(\prod _{j=n-k+1}^{n} \zeta_{q}(j)^{-1}\). Here \(\zeta_{q}(j)= \prod_{m=1}^{\infty} (1- q^{-jm})^{-\varphi_{m}}\) is a \(q\)-zeta function and \(\varphi_{m}\) is the number of monic irreducible polynomials in \(\mathbb{F}_{q}[x]\) of degree \(m\). Furthermore, they deduce that if \(k=n\), then the natural density of \(n\times n\) unimodular matrices over \(\mathbb{F}_{q}[x]\) is zero. Notice that recently several authors, e.g. \textit{A. T. Benjamin} and \textit{C. D. Bennett} [Math. Mag. 80, No. 3, 196--202 (2007; Zbl 1204.11059)] have also considered the question of the probability of \(n\) polynomials to be co-prime in \(\mathbb{F}_{q}[x]\). Their methods are based on the Euclidean algorithm.