Complex socio-ecological systems/Genetic algorithms

From Wikiversity
Jump to navigation Jump to search

Discussants: Ricardo and Bob

Holland, John H. 1992. Genetic Algorithms. Scientific American, July 1992, pp 66-72.

Holland's classical paper on Genetic Algorithms is a description of the application of natural selection and biological evolution in computer science to solve complex problems in fields as design of jet turbines or to anticipate effects of choices in agent-based models to explain human behavior. In the 1960's, building on previous research on the applications of genetic evolution to devise computer programs, Holland developed a program technique which he called genetic algorithm that emulates both natural selection and mating or sexual reproduction evolutionary processes existent in living organisms. The program consists of strings with binary combination of zeros and ones, forming classifier systems that anticipates outcomes encoded as conditions and actions, encoded in the strings. To evolve classifier rules that solve a particular problem, one simply starts with a population of random strings of 1's and 0's and rates each string according to the quality of its result. According to the desired outcome or question, high-quality strings mate, whereas low-quality ones perish. The strength of simple genetic algorithms is that they are at their best when exploring complex landscapes to locate regions of enhanced opportunity. However, according to the author, if a partial solution can be improved further by making small changes in a few variables, it is better to augment the genetic algorithm with other standard methods.

Axelrod, Robert. 1997. Evolving New Strategies. Chapter 1 in Axelrod, Robert, 1997, The Complexity of Cooperation: Agent-based models of competition and collaboration. Princeton University Press.
Axelrod, the known dad of “Prisoner Dilemma” and one of the exponents on the theory of cooperation, search in this Book collect a sequence of studies that, using computational agent based models, explain how people interact and make decisions in competitive and collaborative environments. The nonlinear non rational decision models outputs are used to explain emergent properties of the system. The first chapter, used for discussion in this section, presents the use of Holland “Genetic Algorithm” to simulate multiple rounds of the Prisoner Dilemma Game, impossible to be simulated in tournaments, like tested by Axelrod in his previous works. The chapter presents Genetic Algorithm and the adequacy of use this technique to multiple rounds of the game. The model results show the emergence of a rule behavior of tit for tat after several generations, evolving from a less collaborative strategy at the beginning. As conclusion he express the efficiency of genetic algorithms to simulate strategies in a complex environment

Gell-mann, Murray. 1994. The Jaguar and the Quark. pp: 16-29.

Gell-Mann, M., 1994. The quark and the jaguar : adventures in the simple and the complex, New York: Owl Books.
Gell-Mann explores the notion that underlying complexity are fundamentally simple principles. His metaphor woven throughout the book explores the interconnectivity (logical, conceptual, and practical) between the simplicity of the quark and the mystery of the jaguar. The volume is a collection of essays in which Gell-Mann surveys current (as of 1994) scientific thought and the philosophy of science. Much of the text is autobiographical in nature. The content touches upon quantum physics, sociobiology, ecology and computer engineering. The chapters are written in a sweeping fashion and reflect a quasi-omniscience on the part of the author. "Scientific Enterprise" is surveys science at large. The largest section "Quantum Universe," explores physic and many of the developments throughout the 40s and 70s in which Gell-Mann was a part. "Diversity" explores the suite of theories using a complex systems (plectics) framework. The author spends substantial time on complex adaptive systems, in contrast to discussions of both chaos and algorithmic complexity, and attempts to identify how emergent properties might be understood.Tegeticula (Sam)