Volume 18, Issue 2

The role of a distributed index from the perspective of an individual actor (node) is to minimize its separation from all other actors (nodes). From a systemwide perspective, an optimal distributed index is one that minimizes the diameter of the index graph. We tackle this optimization problem in an evolutionary fashion by performing a series of topology crossovers and fitness-based selections. A set of constraints regulate the fitness function. Different classes of topologies such as star, circle, and skip lists emerge as diameter-optimal structures under different constraints. Knowledge of the optimal topology class in a given context provides strategic information for distributed agents to (re)construct a global index structure based on local information. We also investigate a deterministic approach called polygon embedding, to build topologies with similar properties to that of the evolved topologies.

Ramón Alonso-Sanz (1,2), Larry Bull (1)
Elementary Cellular Automata with Minimal Memory and Random Number Generation

The effect of endowing cells with memory of their last two state values in elementary one-dimensional cellular automata is analyzed in this paper. The potential value of such elementary cellular automata with minimal memory embedded in cells as random number generators is assessed.

Patrick Bahls, Mark McClure, Josh Knox
Decremental Tag Systems and Random Trees

We introduce a variation of Post's tag systems that leads to a finite state machine. Our system is simpler than those considered by Post, in that there are only finitely many states. It is more complicated, in that any given state can evolve in multiple directions. Most importantly, we are able to analyze the system fairly completely and use it to investigate the properties of certain types of randomly generated trees.

Fiona C. Maclachlan
Random Division and the Size Distribution of Business Firms

A computational model of business firm size based on random division is presented. Simulations generate size distributions that are positively skewed with Pareto (power-law) upper tails. Furthermore, the simulated distributions are shown to deviate from the lognormal in ways consistent with some recent empirical findings.

Hasan Akin
The Entropy of Linear Cellular Automata with Respect to Any Bernoulli Measure

This paper deals with the measure-theoretical entropy of a linear cellular automaton (LCA) , generated by a bipermutative local rule (m≥2 and l, ), with respect to the Bernoulli measure on defined by a probability vector . We prove that the measure entropy of the one-dimensional LCA with respect to any Bernoulli measure is equal to .

Bolesław Kopocinski
Modeling Rope Reliability

In this paper we attempt to model the reliability of a steel rope that is susceptible to breakage during its operation. Each individual strand of the rope is modeled using a counting process for breakages and the associated process describing new breakages. The decision to replace the rope is based on the maximum number of breakages in the strands along a given length. Using a stochastic simulation, significant simplifications in the model of the associated process are introduced. The decision to replace the rope is based on the convolution of the associated processes acting on the strands. This is approximated by a Gaussian process. Deterministic modeling of the strands indicates the possibility of estimating the parameters of the processes that characterize the rope.

Dominik Strzałka, Franciszek Grabowski
Influence of Excess 1/f Noise on Channel Capacity

This paper deals with new analytical and experimental aspects of a channel's capacity in the presence of excess nonwhite Gaussian noise with long-range dependencies described by the Hurst parameter H. Shannon's theory of information based on the assumptions given by the Boltzmann-Gibbs extensive thermodynamical basis, does not allow description of many different phenomena directly connected with the ideas of a complex systems approach. This theory is also a basis of many considerations in communication, but a new approach to transmission channels is needed. The transmission channels are no longer simple systems built with only one wire connection, but consist of many different transmission media. For each type of partial connection in such channels there are many various interferences that influence some parts of the channel in different ways. We suggest that in many cases the real capacity of the whole channel can no longer be determined by Shannon's equation without taking into account the problem of excess 1/f noise, which appears as an intrinsic feature of dynamically packet switched networks. The ideas presented in this paper show how the complex system approach can provide a good perspective for analyzing the whole transmission channel.