📄 dmi_aboutfitnessalgorithms.java
字号:
/* * DMI_AboutFitnessAlgorithms.java * * Created on March 11, 2003, 9:19 PM */package com.adaptiveview.ospso.dmi;/** Documentation: About <b>PSO Fitness Algorithms</b>. *<p>If you've taken a look at the FitnessAlgorithm classes in the .test package * you may think that what we are calling Fitness Algorithms would be more * appropriately named Problem Definitions. From the perspective of the particles, * though, they don't know -- and they don't need to know -- anything about the * problem you're using PSO to find solutions for. All they need to know is how * their location compares in terms of "fitness" with other locations. How you * use the coordinates of their location to determine a fitness value is up to * you so long as the comparison of fitness values for different locations * accurately indicates which location has the better fitness value. * *<p>In general, it's up to you to define how fitness values are ranked. However, * the implementation used in these programs assumes that a <b>higher</b> fitness value * (closer to positive infinity) <b>is better</b> than a lower one and, in practice, we * tend to use zero (0) as the target fitness value and negative fitness values * for all sub-optimal solutions. Therefore, for a function like <code>x/y = Math.PI</code>, * for example, we calculate the fitness using: * <code>fitness = 0.0 - Math.abs(Math.PI - (x/y));</code>. * *<p>It's important to keep in mind that no matter how close two particles get to * each other in the solution space if one of the particle's coordinates yield a * better solution then that particle's location must get a higher fitness. The * Move Algorithm is desinged to <i>favor</i> heading in the direction of locations with * higher fitness values. If you create a fitness function that gives <i>every</i> wrong * solution the same fitness value (e.g., -10.0) the Move Algorithm won't be able to tell if * one location is better than another (unless a particle happens to land on the * coordinates of a solution). * * @author AdaptiveView.com */public interface DMI_AboutFitnessAlgorithms extends DMI {}
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -