An extension of the (strong) primitive normal basis theorem (Q2254530): Difference between revisions

Let \(p\) be a prime, \(q\) a power of \(p\), and \(m\) a positive integer. Let \(\mathbb{F}_q\) denote the finite field of order \(q\) and \(\overline{\mathbb{F}}_q\) its algebraic closure. A generator of the cyclic group \(\mathbb{F}_{q^m}^*\) is called a primitive element of \(\mathbb{F}_{q^m}\). Let \(x\in \mathbb{F}_{q^m}\). A basis of \(\mathbb{F}_{q^m}\) over \(\mathbb{F}_{q}\) of the form \(\{x, x^q, \ldots, x^{q^{m-1}}\}\) is called a \textit{normal basis} of \(\mathbb{F}_{q^m}\) over \(\mathbb{F}_{q}\), and such an element is called \textit{free} over \(\mathbb{F}_{q}\). In [J. Reine Angew. Math. 103, 230--237 (1888; JFM 20.0073.03)], \textit{K. Hensel} proved that there exists a normal basis of \(\mathbb{F}_{q^m}\) over \(\mathbb{F}_{q}\). In [Math. Comput. 48, No. 177, 217--231 (1987; Zbl 0615.12023)], \textit{H. W. Lenstra Jr.} and \textit{R. J. Schoof} proved \textit{The Primitive Normal Basis Theorem} which says that there is a primitive normal basis of \(\mathbb{F}_{q^m}\) over \(\mathbb{F}_{q}\), i.e. there exists some \(x\in \mathbb{F}_{q^m}\) that is simulataneously primitive and free over \(\mathbb{F}_q\). Recently, \textit{S. D. Cohen} and \textit{S. Huczynska} provided a computer-free proof of \textit{The Primitive Normal Basis Theorem} in [J. Lond. Math. Soc. 67, No. 1, 41--56 (2003; Zbl 1045.11087)]. \textit{S. D. Cohen} and \textit{S. Huczynska} proved \textit{The Strong Primitive Normal Basis Theorem} in [Acta Arith. 143, No. 4, 299--332 (2010; Zbl 1219.11182)] which says that there exists some \(x\in \mathbb{F}_{q^m}\) such that \(x\) and \(x^{-1}\) are both simulataneously primitive and free over \(\mathbb{F}_q\), unless the pair \((q, m)\) is one of \((2, 3), (2, 4), (3, 4), (4, 3)\) or \((5, 4)\). Let \(A = \begin{pmatrix} a & b\\ c & d \end{pmatrix} \in \text{GL}_2(\mathbb{F}_q)\) and \(\mathbb{I}_n:= \{F\in \mathbb{F}_q[X]\,:\, F\,\,\text{irreducible of degree}\,n\,\}\). Let \(F\in \mathbb{I}_n,\, n\geq 2\), and define \[ F^A(X):= (cX+d)^n\,F\Big(\displaystyle\frac{aX+b}{cX+d}\Big). \] Then \(x\in \overline{\mathbb{F}}_q\) is a root of \(F\) if and only if \((-dx+b)/(cx-a)\) is a root of \(F^A\). In the article under review, an extension to \textit{The Primitive Normal Basis Theorem} and \textit{The Strong Primitive Normal Basis Theorem} is considered, which is stated as the following problem. Does there exist some \(x\in \mathbb{F}_{q^m}\) such that both \(x\) and \((-dx+b)/(cx-a)\) are simultaneously primitive and free over \(\mathbb{F}_{q}\)? The author proves that the aforementioned problem can be solved for most \(A \in \text{GL}_2(\mathbb{F}_q)\) when \(q\geq 23\) and \(m\geq 17\). Hence, the main result of the reviewed article is the following theorem. \vskip 0.2in \noindent Theorem. Let \(q\geq 23\), \(m\geq 17\), and \(A = \begin{pmatrix} a & b\\ c & d \end{pmatrix} \in \text{GL}_2(\mathbb{F}_q)\), such that if \(A\) has exactly two non-zero entries and \(q\) is odd, then the quotient of these entries is a square in \(\mathbb{F}_{q^m}\) (thus \(A\) may have two, three or four non-zero entries). There exists some \(x\in \mathbb{F}_{q^m}\) such that both \(x\) and \((-dx+b)/(cx-a)\) are simultaneously primitive and free over \(\mathbb{F}_{q}\).