abstract

A fashion-based cellular automata is one whose updating rule follows the form of an ecological competition model. The rule for the automata is specified by a square matrix with entries quantifying the influence each state has on each other when they both occur within a neighborhood. Because they preserve areas containing a single cell state, these rules are well able to specify automata that rapidly transform a random initial condition into a map appearing as a collection of caverns. Because the automata acts in a purely local fashion, it is valuable for generating collections of maps with similar look-and-feel, but different details, enabling automatic content generation and replayability in video games. This study extends an earlier study, examining new fitness functions and studying reusability, scalability, and the impact of parameter tuning for this type of cellular automata for automatically designing level maps. A representation for evolutionary computation is morphable if convex combinations of instances of the representation are instances of the representation. The fashion-based rules, being specified by real values matrices, are morphable. The ability to produce new, more complex maps by exploiting morphability is also explored.

abstract

We study the parallel version of the Ising model, introduced as a model for opinion formation. We first recall some results about the statistical analysis of the serial and fully parallel version. We introduce the dilution (or asyncronism) of the updating rule and show that the chequerboard patterns that appear in the fully parallel version are unstable with respect to dilution, but exhibit finite-size effects and long-lasting metastable states.

abstract

We investigate some phase transitions of a nonlinear, parallel version of the Ising model, characterized by an antiferromagnetic linear coupling and a ferromagnetic nonlinear one. This model arises in problems of opinion formation. The mean-field approximation shows chaotic oscillations by changing the linear coupling or the connectivity. The spatial model exhibits bifurcations in the average magnetization, similar to what is seen in the mean-field approximation induced by the change of the topology, after rewiring short-range to long-range connections as predicted by the small-world effect. These coherent periodic and chaotic oscillations of the magnetization reflect a certain degree of synchronization of the spins, induced by long-range couplings. Similar bifurcations may be induced in the randomly connected model by changing the coupling or the connectivity and the synchronism of the updating (dilution of the rule).

abstract

In a recent paper [arXiv:1506.06649 [nlin.CG]], we presented an example of a 3-state cellular automaton which exhibits behaviour analogous to degenerate hyperbolicity often observed in finite-dimensional dynamical systems. We also calculated densities of 0, 1 and 2 after \(n\) iterations of this rule, using finite state machines to conjecture patterns present in preimage sets. Here, we re-derive the same formulae in a rigorous way, without resorting to any semi-empirical methods. This is done by analysing the behaviour of continuous clusters of symbols and by considering their interactions.

abstract

Considered is a 2D cellular automaton with moving agents. Each cell contains a particle with a value \(=\) (color, marker), which can be changed by an agent. The objective is to form a specific target pattern belonging to a predefined pattern class. Areas of applications are the alignment of spins, particles, or fibers. The target patterns shall consist of preferably long narrow paths with the same color, they are called “path patterns”. The quality of a path pattern is measured by the degree of order that is computed by counting matching \(3 \times 3\) patterns (templates). The markers act like artificial “pheromones” that improve the solution’s quality (effectiveness) and the efficiency of the task. The agents’ behavior is controlled by a finite state machine (FSM). The agents used can perform 32 actions, combinations of moving, turning and value setting. They react on the own particle’s value, the value in front, and blocking situations. For a given set of \(n \times n\) fields near optimal, FSMs were evolved by a genetic algorithm. The evolved agents are capable of forming path patterns with a very high degree of order (90–100%). The whole multi-agent system was described by cellular automata. The CA-w model (cellular automata with write access) was used for the implementation of the system in order to reduce the implementation effort and speed up the simulation.

abstract

A model of simulated cognitive agents (naïve creatures) learning to safely cross a cellular automaton-based highway is described. These creatures are minimal, equipped with the ability to “perceive”, “reason”, “judge”, and “respond” in order to learn from each other by evaluating if a creature in the past was successful in crossing the highway for their current situation. A large amount of data files are generated from this simulation model under different configurations of the simulation parameters’ values (such as the traffic density and the nature of these creatures in terms of fear and desire). These simulation parameters heavily influence the learning outcomes examined through the collected simulation metrics. We study how these parameters influence these metrics using regression trees.

abstract

In the microscopic modeling of swarm of robots, individual robots may be identified as cognitive agents. We describe a model of a population of simple cognitive agents, naïve creatures, experiencing fear and/or desire when learning to cross a highway. The creatures use an observational social learning mechanism in their decision to cross the highway or not. We study in various traffic environments characterized by vehicle traffic density, how fear and/or desire affects creatures’ performance measured by the number of killed, queued and successful creatures. We investigate how this performance is affected when the creatures’ knowledge accumulated in one traffic environment is transferred to a different traffic environment for them to continue their learning. We consider the case when creatures are not allowed to change their crossing point and when they are allowed to change it.

abstract

Improving the personalization of recommendation methods is a hot topic with wide application in real on-line commercial systems. One major concern is that an algorithm that focuses too strongly on diversity is putting recommendation accuracy at risk. Based on the method described in [Proc. Natl. Acad. Sci. USA 107, 4511 (2010)], we propose a more personalized algorithm in which each user is assigned with a parameter for the initial configuration setting and a parameter for the hybridization. We find that each user has his/her optimal parameters which are very different from user to user. We finally design a simple method to estimate users’ personalized parameters and the recommendation accuracy can be improved accordingly.

abstract

The local interaction of multi-species populations can be described by a discrete in space lattice differential equation, where the microscopic local rules of interaction are given in terms of functions that describe the incentive of individuals to move from their current site into a neighboring site. By refining the discrete microscopic description, we derive a macroscopic continuous model with cross diffusion. We present an a priori criterion that allows to verify whether the model preserves non-negativity of populations, which is an important property in biological applications.

abstract

The paper presents recent trends in the area of Cellular Automata (CA) originated microscopic models of crowd. There are many interesting applications of CA-based models, which incorporate different definitions of floor fields, grouping behavior, building complex scenarios, parallelisation of computing processes, as well as the development of validation and verification tests. Thus, due to the observed rapid development of discrete crowd simulation models, they become a real alternative to force-based ones. It is especially visible in a new kind of crowd-model application, where simulation is required to cooperate with different sensors in real-time regime.