Parallel rewrite systems in the form of stringbased L-systems are used for modeling and visualizing growth processes of artificial plants. A package implementing context-sensitive (m,n)-L-systems is presented which takes full advantage of Mathematicas expr
entate a drawing tool, known as a turtle, in three-dimensional space (rotate,yaw, pitch), thus translating a one-dimensional string into a 3D-object resemb–ling a plant.
2Genetic L-system programming
Genetic Programming (GP) has been introduced as a method to automaticallydevelop populations of computer programs through simulated evolution [3,4].Considering L-systems as rulebased development programs it is easy to defineprogram evolution. Each program encoded as a symbolic expression has to beinterpreted and is assigned a fitness value dependent on the optimization taskto be solved. On the basis of these fitnesses the individual programs strugglefor survival of the fittest“ and for the chance to become members of the nextgeneration. In order to introduce variations into the program encoding struc-tures genetic operators like mutation or crossover – gleaned from nature´smutating and recombining operators on cell genomes – are applied. The evo-lution process develops new populations of programs from generation togeneration, the interplay of modifying operators and selection hopefully lead–ing to ever better programs.
2.1Encoding L-systems
In order to use expression evolution for L-systems a proper encoding schemehas to be defined which will be done for the more general case of IL-systems(with I referring to the number of context symbols). In context-sensitive IL-systems the rewriting of a letter depends on m of its left and n of its rightneighbors, where m and n are fixed integers. These systems are denoted as(m,n)L-systems which resemble context-sensitive Chomsky grammars, but –as L-system rewriting is parallel in nature – every symbol is rewritten in eachderivation step; this is especially important whenever there is an overlap ofcontext strings. The following L-system representations are used in our IL-system package which is part of MathEvolvica.
Each IL-system rule has the form l < p > r → s with l, p, r and s denoting theleft context, predecessor, right context and successor, respectively. The sym-bols "<" and ">" separate context and predecessor strings. Thus each rule canbe represented by an expression of the form
LRule[ LEFT[ l ], PRED[ p ], RIGHT[ r ], SUCC[ s ] ]. Accordingly, an L-system with its axiom and rule set is encoded by an expres-sion of the form
LSystem[ _AXIOM, LRULES[ __LRule] ]
using Mathematica pattern notation.
IMS´95, First Int. Mathematica Symposium, Southampton, UK, 1995, Computational Mechanics Pub.
Evolving Teamwork and Coordination with Genetic Programming Some problems can ...First, there are questions of team diversity and of breeding strateg......
Publications System Evolving using Ant Colony Optimization Algorithm Nada M.A....? Automatic Programming is the area in which Artificial Intelligent and ...
Available online at www.sciencedirect.com Journal of Power Sources 180 (2008) 468–475 Frequency control in micro-grid power system combined with ...
Available online at www.sciencedirect.com Journal of Power Sources 180 (2008) 468–475 Frequency control in micro-grid power system combined with ...
热门文档