Complex Pisot numbers of small modulus. (Q1408132)

Recall that a PV-number (named after Pisot and Vijayaraghavan) is an algebraic integer \(\alpha>1\) whose remaining conjugates (if any) lie in the open unit disc. Similarly, a complex PV-number is a non-real algebraic integer \(\alpha,\) where \(| \alpha| >1,\) whose remaining conjugates other than \(\overline{\alpha}\) lie in the open unit disc. It is known that the positive zero \(\theta_0=1.324\dots\) of \(x^3-x-1\) is the smallest PV-number (Siegel) and that \((1+\sqrt{5})/2\) is the smallest limit point of the set of all PV-numbers (Dufresnoy and Pisot). In contrast, \textit{C. Chamfy} [Ann. Inst. Fourier 8, 211--261 (1958; Zbl 0087.07603)] has shown that the smallest (by modulus) complex PV-number has modulus \(\sqrt {\theta_0}=1.150\dots\). (For instance, so is a complex root of \(x^3-x^2+1\).) However, a corresponding result concerning the smallest limit point which claims that \(\sqrt{(1+\sqrt{5})/2}=1.272\dots\) is the smallest limit point of the moduli of all complex PV-numbers is only a conjecture. In this paper, the author makes a step towards this result. He proves that the interval \((1,1.17)\) contains no limit points, so only the interval \([1.17, 1.272\dots)\) remains to be handled. In the proof he uses an algorithm of Chamfy. In fact, in his thesis he finds a list of 72 complex PV-numbers less than \(\sqrt{(1+\sqrt{5})/2}\) in modulus, but it is not clear whether this list is complete.