A GENETIC ALGORITHM TECHNIQUE FOR APPROXIMATING FUNCTIONS OF MULTIPLE INDEPENDENT VARIABLES
Aravind Gurumurthy · OhioLink ETD Center (Ohio Library and Information Network) · 2003
This thesis addresses approximation of functions of multiple independent variables.Engineers generally use look up tables, Taylor series or other series representations for interpolating data, with varying degrees of accuracy.In this research we consider another method.We extend a genetic algorithm technique for approximating functions of one variable with a set of polynomials, with integer coefficients, to functions of more than one variable.The goal is to minimize the sum of squared errors over a range of experimentally gathered or sampled data.This research is particularly applicable to semi-custom hardware designs and embedded programming applications.