Krein condition in probabilistic moment problems (Q1590233): Difference between revisions

This paper makes an extensive use of the well-known Krein and Lin conditions [which are integral tests on the density function, see e.g. \textit{G. D. Lin}, Stat. Probab. Lett. 35, No. 1, 85-90 (1997; Zbl 0904.62021)], to prove the M-determinacy or the M-indeterminacy of powers of the log-normal, normal, and inverse Gaussian distribution. As in previous papers [\textit{C. Berg}, Ann. Probab. 16, No. 2, 910-913 (1988; Zbl 0645.60018) and \textit{G. D. Lin} and \textit{J. S. Huang}, Aust. J. Stat. 39, No. 3, 247-252 (1997; Zbl 0898.62015)], the cube appears to be the frontier from which the powers of the latter distributions become M-indeterminate. There is also a conjecture concerning the classes of functions preserving or destroying the M-determinacy.