Properties of Stirling polynomials and a disproof of Robertson's conjecture (Q805782)

During his long research period in function theory, M. S. Robertson enriched the field with a bunch of conjectures some of which turned out to have an important influence to the research progress, e.g. to the proof of the Bieberbach conjecture by \textit{L. de Branges} [Acta Math. 154, 137-152 (1985; Zbl 0573.30014)]. In the present paper one of Robertson's conjectures [\textit{M. S. Robertson}, Lect. Notes Pure Appl. Math. Vol. 36, 1-33 (1978; Zbl 0394.30007)] is disproved. As this disproof consists only of calculations, it is natural to obtain the same result by means of Computer Algebra which had been done by the reviewer [Complex Variables, Theory Appl. 16, 127-130 (1991)]. Similarly the conjecture that \(a_{kn}\leq 0\) iff \(k=4j+1\) (j\(\geq 3)\) or \(k=4j+2\) (j\(\geq 3)\) where \[ ([(\frac{1+z}{1- z})^ x- 1]/2xz)^{1/2}=:\sum^{\infty}_{n=0}(\sum^{n}_{k=0}a_{kn}x^ k)z^ n \] which is formulated in the present paper, can be verified by means of Computer Algebra for \(n\leq 50\). A further conjecture concerning the sign pattern of certain power series is posed. Strangly in the present paper no reference to Robertson's work is given.