INFORMATION TRANSFER IN COMPLEX SYSTEMS, WITH APPLICATIONS TO REGULATION.
ILLINOIS UNIV URBANA BIOLOGICAL COMPUTER LAB
Pagination or Media Count:
The study is concerned with information theory and its relevance to the study of complex systems. When information about every detail of their activity is kept, many systems are too complex to be manageable and can only be dealt with by sacrificing detail. It is shown here that multivariable information theory is capable of eliminating much detail while preserving information about the interrelations between parts of a system, even when those interrelations are very complex. A procedure is described and exemplified, for example, which is helpful in the decomposition of hierarchical systems. It is shown, among other results, that when two variables are related in the set theoretic sense the transmission between them is maximized when their behaviors are isomorphic. This observation leads to an algorithm for the computation of channel capacity for arbitrary finite-state systems of a very general type. The importance of information in regulatory processes is discussed and quantified, and several basic regulatory schemes are discussed in terms of the information involved, showing in an exact way how information transfer and channel capacity limit the ability of any system to act as a successful regulator. Author