Regularity properties of measurable functions satisfying a multiplicative type functional equation almost everywhere (Q2346354): Difference between revisions

According to a classical result, if two independent random variables are given such that their sum and difference are also independent, then all have normal distributions. The problem itself can be interpreted as a functional equation involving density functions and which is satisfied only almost everywhere. This phenomenon, when measurable solutions of a functional equation almost everywhere are considered, appears as typically characterization problems of probability distributions. In such situations, usually one have to prove that the functions satisfying the equation almost everywhere are almost equal to smooth functions satisfying the equation everywhere. This paper gives an excellent demonstration of this fact, presenting a general and effective method. The approach is based on the deep and clever combination of the techniques of classical analysis and measure theory.