Foundations of synergetics. I: Distributed active systems

DOI10.1007/978-3-642-78556-6MaRDI QIDQ1188763zbMATH OpenOpenAlexFDO

Publication date 17 September 1992

Published in Springer Series in Synergetics (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1007/978-3-642-78556-6

Research exposition (monographs, survey articles) pertaining to biology (92-02) Neural networks for/in biological studies, artificial life and related topics (92B20) Developmental biology, pattern formation (92C15) Neural biology (92C20) Ecology (92D40) Research exposition (monographs, survey articles) pertaining to systems and control theory (93-02)

Review text

Synergetics is defined as an interdisciplinary field of research concerned with the cooperation of individual parts of a system that produces macroscopic spatial, temporal, or functional structures. The author's book provides an introduction to the mathematical theory of cooperative behavior in active systems of various origins, both natural and artificial. It is the first volume of a two-volume set based on a lecture course presented at the University of Moscow. Synergetics deals with deterministic as well as stochastic processes; this volume concerns the former. It treats the problems of pattern formation and the properties of self-organized regular patterns in distributed active systems. It also contains a discussion of distributed analog information processing which is based on the cooperative dynamics of active systems. Stochastic aspects of self-organization are left to the second volume of the set. In the preface, the author states that he has tried to avoid delving into particular applications. In fact, a number of examples of the general mathematical principles of synergetics are included, and the book is better for their presence. A few of the applications are mentioned in the chapter-by-chapter summary which follows. Chapter 1. Introduction. A distributed active system is defined as a set of interacting active elements which are either distinct units or represent sufficiently small pieces of a continuous medium. General equations describing distributed active systems are introduced. Chapter 2. Bistable media. Each element in a bistable medium has two stationary states that are stable under small perturbations. Large perturbations can trigger transitions between the two steady states. Trigger waves, which represent waves of transition from one state to another, can propagate in bistable media. Examples include the growth of bacteria in an environment containing food and the toppling of a row of dominoes. Chapter 3. Excitable media. An excitable element has a single stationary state that is stable under small perturbations. However, if the perturbation exceeds a certain threshold, the element performs a specified sequence of transitions before it returns to the initial state. The elements may be represented by differential equations or as finite- state cellular automata. Excitation pulses or waves can propagate along a chain of such elements. Action potentials in nerve and muscle tissue are examples of such pulses. Concentric patterns and spiral waves, as in the Belousov-Zhabotinskij chemical reaction, appear in two-dimensional arrays of excitable elements. Patterns which are still more complex arise in three-dimensional arrays. Chapter 4. Oscillatory media. Oscillatory media represent the continuous limit of a large population of elements that perform stable limit-cycle oscillations. Interactions between elements are weak and are principally manifested in changes of the oscillation phases. Plane and spiral waves can be produced in homogeneous oscillatory media. Inhomogeneities in the medium function as sources of periodic concentric waves. When two such sources operate in the same medium, the higher-frequency source acts as a pacemaker which suppresses the lower-frequency source. Chapter 5. Active media with long-range inhibition. A special class of two-component media is considered in which one of the components functions as an activator and the other as an inhibitor. An isolated element can be bistable, excitable, or oscillatory. If the inhibitor acts at long range and quickly adjusts to the momentary distribution of the activator, the formation of stable, stationary, dissipative patterns is favored. There are two basic kinds of these patterns, spikes and broad strata. A spike is a small region of increased activator density immersed in a much larger region of increased inhibitor density. Broad strata are large domains of relatively constant activator density separated by interfaces in which the activator density changes sharply. Chapter 6. Neural networks. Neural networks are distributed active systems that consist of discrete two-state elements linked by long-range activatory (excitatory) and inhibitory connections. Since the connections often project to distant elements and differ in their properties, the network is highly nonuniform and has a complicated topology. General networks of this kind are difficult to analyze, but much can be learned by examining neural networks that are designed to realize various analog distributed information-processing functions. The McCulloch-Pitts neural network is presented, and its formal relation to the Ising model of magnetic spins is explored. It is shown that neural nets can display associative memory and that they can learn to respond to a chosen set of input patterns if the weights attached to the connections between their elements are appropriately modified. Prominent among the many neural networks which are discussed are the perceptron and the Boltzmann learning machines. Chapter 7. Reproductive networks. Multicomponent active systems of reproductive agents are considered. The agents interact, directly or through mediators, to mutually influence their own reproduction rates. Reproductive networks comprise a large number of models in various fields ranging from nuclear chain reactions to evolution, ecological systems, and market economics. These networks can be used for analog information processing. It can be demonstrated, for example, that evolution is closely related to learning. When all agents in a reproductive network compete for a common renewable resource and no directed interactions are present, competition leads to survival of the single fittest agent. If each agent has, in addition, its own specialized resources, steady coexistence of various agents is possible. With inhibition, the system becomes multistable, i.e. it has many stable states that correspond to the survival of different agents. In networks with directed interactions, reproduction waves can propagate.

Recommendations

Cited in

(66)

This page was built for publication: Foundations of synergetics. I: Distributed active systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1188763)