ON COMBINATORIAL ENTROPY OF SOURCES OF BINARY INFORMATION

Abstract

The paper considers the question of converting the entropy of a probabilistic information source generating binary sequences into combinatorial forms of its two constituent sources. One of them is basic, and the second has a conditional entropy with respect to the first. It is shown that together such sources make it possible to optimally encode information using numbering at the efficiency level of known methods of optimal coding. In terms of their functions, they are universal and can solve all the problems that are solved by the methods of coding probabilistic sources. However, unlike them, they do not require statistical tests before optimal coding, or they are carried out to a much lesser extent. They also simplify the calculation of the entropy of a probabilistic source of information, which is useful in solving many problems where it is necessary to know the value of the entropy of the source, for example, in error-correcting coding.