Unconstrained GA
Working on a GA to solve a simple geomtry problem - make a polygon that approximates a circle - without ever refering to the analytical formula for a circle.
Selection was based on number of line intersections and the even-ness of the length from the centroid. Had to fiddle around with the rates of mutation and population size to stop it converging too quickly on a local maximum.
Selection was based on number of line intersections and the even-ness of the length from the centroid. Had to fiddle around with the rates of mutation and population size to stop it converging too quickly on a local maximum.
Comments