On uniqueness of the problem of moments in the class of -distributions (Q4522792)

The paper is devoted to solving the general problem of finding appropriate conditions under which the distribution of a random vector \(x= (x_1,\dots, x_n)\), taking values from the set \(\{(q^{r_1}_1,\dots, q^{r_n}_n)\mid r_i= 0,1,\dots; i= 1,\dots, n\}\), where \(q= (q_1,\dots, q_n)\), \(q_i> 1\), \(i= 1,\dots\), is uniquely determined by the sequence of its mixed moments in the class of \(q\)-distributions, i.e. distributions whose support values belong to the above-mentioned set. The author defines \(q\)-binomial invariants as a suitable tool to investigate \(q\)-distributions, finds relations between \(q\)-binomial invariants and the mixed moments of the \(x\) random vector distributions, gives sufficient conditions under which the distribution of \(x\) is uniquely determined by the sequence of its mixed moments, and provides explicit expressions for the probabilties \(\text{P}(x_1= q^{r_1}_1,\dots, x_n= q^{r_n}_n)\), \(r_1,\dots, r_n= 0,1,\dots\), and estimates of these probabilities, which for \(n= 1\) can be considered as \(q\)-analogues of the well-known Bonferroni inequalities. A result on the convergence of a sequence of \(q\)-distributions is proved under the condition that the corresponding \(q\)-binomial invariants converge, too. These results are used to study the limit distribution of the number of solutions of a class of systems of linear equations over a finite field. An example is also presented, which allows to show that if the conditions guaranteeing the uniqueness of the reconstruction of a \(q\)-distribution on the basis of the sequence of its mixed moments are not fulfilled, then more than one \(q\)-distribution with the same moments can exist.