Tutorial Details: Parallel Strategies With Genetic Algorithms
|Date:||Thursday, March 29, 2007, 01:00 pm - 03:00 pm|
|Instructor(s):||Benjamin J. Lynch, MSI|
Many scientific problems require the optimization of multidimensional functions that have dozens or even thousands of local minima and maxima. Local optimization methods such as the Newton-Raphson method and its many variants can find nearby local extrema but have little or no chance of finding a global minimum or maximum. Using a genetic algorithm (GA) is often a very efficient means to find a global minimum or maximum for complex functions.
This hands-on tutorial will show how genetic algorithms work and will discuss general optimization strategies with GAs. It will also work through a few GA problems using parallel GA optimization routines.
|Prerequisites:||Basic skills in MPI and Fortran or C|