Re: Genetic Algorithms

*To*: mathgroup at smc.vnet.net*Subject*: [mg32610] Re: [mg32575] Genetic Algorithms*From*: Daniel Lichtblau <danl at wolfram.com>*Date*: Thu, 31 Jan 2002 01:45:40 -0500 (EST)*References*: <200201300818.DAA29913@smc.vnet.net>*Sender*: owner-wri-mathgroup at wolfram.com

IA wrote: > > To all, > > I would like to know whether Mathematica includes any special modules for > Genetic Algorithms. > I took a look at wolfram.com however was not able to retrieve any GA extra > module. > > Sharing experience(s) with GA programming is mostly welcomed.Any further > advise is welcome aswell.... > > Cheers, > Paul Below is a minor alteration of a response I just sent to the related recent post on the Usenet news group sci.math.symbolic. There is no Genetic Algorithm code in the standard distribution of Mathematica. That said, at least some people have worked with GAs in Mathematica. One recent article on the topic is: Fowler, David (2000). Genetically seeking sparse rulers. Mathematica in Education and Research 9(2) 33-40. A related area, genetic programming, is discussed in: Nachbar, Robert (1995). "Genetic programming". The Mathematica Journal 5(3) 36-47. The text below covers alot of ground involving GAs in particular and evolutionary programming methods more generally. Jacob, Christian (2001). Illustrating Evolutionary Computation with Mathematica. Morgan Kaufmann. This is an update of an earlier German version titled Principia Evolvica. Chapter three deals specifically with genetic algorithms although later chapters on evolutionary programming techniques may also be of interest. Note that this book provides alot of actual code. A complete set of electronic notebooks that accompany this work may be found at the author's Evolvica web site. http://www.cpsc.ucalgary.ca/njacob/IEC/ At WRI we have implemented, for a future release of Mathematica, some related technology for doing optimization. The specific method we use is from: Price, Kenneth, and Storn, Rainer (1997). Differential evolution. Dr. Dobb's Journal, April 1997, 18-23 (C code presented p. 78). This is not quite a GA insofar as "chromosome" vector elements are reals rather than taking discrete values. It falls into the general category of Evolutionary Strategy (ES), as discussed in chapter four of Jacob's book. Some Mathematica notebooks discussing this differential evolution implementation may be found at the URL below (look for notebooks by Brett Champion and by myself). http://library.wolfram.com/conferences/devconf2001/#numerics They describe use rather than the code itself but at least this might give an idea of what one can do with an evolutionary method coded in Mathematica. Daniel Lichtblau Wolfram Research

**References**:**Genetic Algorithms***From:*"IA" <paul_stuart2002@hotmail.com>