This work describes a new heuristic algorithm that estimates structural and geometric similarity of three-dimensional morphologies. It is an extension to previously developed measure of similarity [25] that was only able to consider the structure of 3D constructs. Morphologies are modeled as graphs with vertices as points in a 3D space, and edges connecting these vertices. This model is very general, therefore the proposed algorithm can be applied in (and across) a number of disciplines including artificial life, evolutionary design, engineering, robotics, biology and chemistry. The primary areas of application of this fast numerical similarity measure are artificial life and evolutionary design, where great numbers of morphologies result from simulated evolutionary processes, and both structural and geometric aspects are significant. Geometry of 3D constructs (i.e., locations of body parts in space) is as important as the structure (i.e., connections of body parts), because both determine behavior of creatures or designs and their fitness in a particular environment. In this work both morphological aspects are incorporated in a single, highly discriminative measure of similarity.
Abstract. In this paper a formal approach to construction of a similarity measure for complex creatures is presented. The simulation model is described, and a Framsticks agent is expressed in a formal way. This helps in defining a dissimilarity measure. Two main ideas are discussed with reference to biology, namely genotypic and phenotypic methods. The holistic phenotypic measure is then proposed, where a fast, heuristic algorithm is used. Examples of its application are shown, including mutation and crossing over analysis, and a clustering tree based on distances between pairs of seven artificial individuals.
Abstract. Memetic algorithms usually employ long running times, since local search is performed every time a new solution is generated. Acceleration of a memetic algorithm requires focusing on local search, the most time-consuming component. This paper describes the application of two acceleration techniques to local search in a memetic algorithm: caching of values of objective function for neighbours and forbidding moves which could increase distance between solutions. Computational experiments indicate that in the capacitated vehicle routing problem the usage of these techniques is not really profitable, because of cache management overhead and implementation issues.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.