| Abstract: | A GENETIC ALGORITHM IS A PROBLEM-SOLVING METHOD IN WHICH POINTS IN THE SEARCH SPACE ARE ENCODED AS A POPULATION OF BITSTRINGS CALLED CHROMOSOMES. LIKE EVOLUTION IN GENETICS, FROM WHICH THE ALGORITHM TAKES ITS NAME, THE ALGORITHM PRODUCES SUCCESSIVE GENERATIONS OF CHROMOSOMES BY APPLYING A SERIES OF REPRODUCTIVE OPERATORS. CHROMOSOMES THAT ARE MORE `FIT' ARE ALLOWED TO REPORDUCE WITH A HIGHER PROBABILITY, WHERE `FITNESS' IS A RELA- TIVE MEASURE OF WHICH CHROMOSOMES ARE CLOSER TO SOLUTION POINTS IN THE SEARCH SPACE. THE PRIMARY REPRODUCTIVE OPERATORS, CROSSOVER AND MUTATION, ARE THEN APPLIED TO THE POPULATION OF CHROMOSOMES TO PRODUCE A WHOLE NEW POPULATION SET OF THE SAME SIZE. THE ALGORITHM STOPS WHEN THE SEARCH NO LONGER SEEMS TO BE ADVANCING; THE ASSUMPTION BEING THAT A GLOBAL MAXIMUM HAS BEEN REACHED. SINCE GENETIC ALGORITHMS ARE SIMILAR TO EVOLUTION IN NATURE, THEY ARE UTILIZED NOT JUST AS A SEARCH TECHNIQUE IN COMPUTER SCIENCE, BUT AS BIOLO- GICAL MODELS OF EVOLUTION WITHIN VARIOUS SUBDISCIPLINES OF BIOLOGY: GENETIC
ALGORITHMS SERVE A DUAL PURPOSE IN SCIENCE. WE PROPOSE A VARIANT OF THE GENETIC ALGORITHM WHICH OFFERS AN IMPROVE- MENT IN SEARCH PERFORMANCE, AS WELL AS BEING MORE HIGHLY PARALLELIZABLE, FOR THE NEEDS OF COMPUTER SCIENCE, WITH THE ADDED BENEFIT FOR BIOLOGY OF BEING A MORE ACCURATE MODEL OF NATURAL EVOLUTION. |
