A scale of degrees of independence of random variables (Q1265330)

The paper proposes a scale of independence degrees of random variables. The lowest point of the scale is the lack of correlation, and its highest point is the usual independence. The convolutional independence is situated between these points and is defined by the condition that the distribution function of the sum of random variables is the convolution of the distribution functions of the summands. The proposed scale is made up of two parts. The first part is a linearly ordered countable set with the lack of correlation as a minimal element. This part of the scale is well-defined for random variables having finite moments of any order. The second part of the scale is a partially ordered set of the power of continuum with the convolutional independence as a minimal element and the usual independence as the maximal one.