On a sequence of rational functions (Q908706)

Let \((R_ n)_{n\geq 2}\) be the sequence of rational functions defined on \([2,+\infty [\) by: \(R_ 2(x)=1\), \(R_{2n}(x)+R_{2n+1}(x)=x\) (n\(\geq 1)\) and \(R_{2n}(x)R_{2n-1}(x)=R_ n(x)\) (n\(\geq 2)\). This sequence was introduced in various contexts by \textit{D. Bessis} and \textit{P. Moussa} [Commun. Math. Phys. 88, 503-523 (1983; Zbl 0523.30019)], \textit{D. Bessis}, \textit{M. L. Mehta} and \textit{P. Moussa} [Lett. Math. Phys. 6, 123- 140 (1982; Zbl 0483.33006)]. They have proved the following inequalities \(0<R_{2n}(x)<R_ 2(x)<R_ 3(x)<R_{2n+1}(x)<x\) and conjectured that \(R_{2^ kn+a}\) lies between \(R_ a\) and \(R_{2^ k+a}\). The author disproves this conjecture, namely in a right neighbourhood of 2 the value \(R_{2^ kn+a}(x)\) (for \(a\neq 0,1)\) is never between \(R_{2^ k+a}(x)\) and \(R_ a(x)\). The proof is elementary and by the way, the following amusing result is established: \(R_ n(0)=(-1)^{R'n(2)}\) and the sequence \((R_ n(0))_ n\) is recognizable by a 2-automaton. Connection with a paper-folding sequence is noticed.