Design

A key idea in metaheuristics is that specific solutions to optimization problems can represented in a more general form, called an encoding. Often, this form is an array of boolean, integer, or floating-point values. By using an encoding, the solution can be operated on using general operators and algorithms, since they do not require knowledge of the problem space to be effective. In order for this arrangement to work, a suitable method is required to evaluate the encoded solution and provide that feedback to the algorithm. This concept is illustrated below:

problem_space_abstraction.png

Adopting this approach leads to the following software design:

sw_design.png

Last edited Oct 20, 2011 at 11:03 PM by jamestunnell, version 9

Comments

No comments yet.