Some remarks on the moment problem and its relation to the theory of extrapolation of spaces (Q2435915)

It is known that the coincidence of \(n\)-th power moments, \(n\in \mathbb N\), of two nonnegative random variables does not imply the coincidence of their distributions. Moreover, as the author shows, in this case the ratio of half-integer moments may tend to infinity arbitrarily fast. On the other hand, if \(\xi_1\) and \(\xi_2\) are nonnegative random variables, such that \(\mathbb E\xi_1^{p_n}=\mathbb E\xi_2^{p_n}\) for some sequence \(\{ p_n\}_{n=1}^\infty\) of different numbers from a finite interval \([a,b]\), and \(\mathbb E\xi_j^{b+\varepsilon}<\infty\), \(j=1,2\), for some \(\varepsilon >0\), then the distributions of \(\xi_1\) and \(\xi_2\) coincide. It is impossible here to replace the coincidence of all the moments by a two-sided inequality between them. It can happen that in this situation \[ \limsup \limits_{\tau \to \infty}\frac{\mathbb P\{ \omega: \;\xi_1>\tau \}}{\mathbb P\{ \omega: \;\xi_2>\tau \}}=+\infty . \] Relations with the theory of extrapolation of function spaces are discussed.