Skip to content

Latest commit

 

History

History
62 lines (34 loc) · 2.93 KB

Documentation.md

File metadata and controls

62 lines (34 loc) · 2.93 KB
Raw

This namespace contains the following:

IMetaHeuristics

This is an interface. Every meta-heuristic algorithm class in this project implements this interface. It contains methods such as

//accepts functions that are neccessary in most meta-heuristic algorithms
void create(Configuration<SolutionType> config);

//performs a single iteration step
SolutionType singleIteration();

//performs a specified number of iterations
SolutionType fullIteration();

List<double> getIterationSequence();

LAHC

This means Late Acceptance Hill Climbing, a meta-heuristic proposed by Yuri Bykov in August 2008. This class has not been implemented as a stand-alone meta-heuristic.

Rather, a stand-alone meta-heuristic such as Artificial Bee Colony can inherit from it, as its Bee class does to create Bee-Lahc, a Bee specie that uses the LAHC algorithm implicitly.

Such a class would gain late acceptance hill climbing powers, which would give it more exploratory power. Hmm, cool idea for a super power, huh? ExploraBot, the auto bot that goes adventuring in deep space ....

Note: This should probably be simply Late Acceptance rather than Late Acceptance Hill Climbing because it is not the full LAHC Algorithm. I'll rename this to LateAcceptance instead

Search

The Search class is to contain Enums and other common resources to be used throughout this namespace.

Selection

The Selection class contains static tested selection methods that you can pass into meta-heuristic implementations instead of having to create yours again. You can use selection methods such as:

  • Roullete Wheel Selection

This is similar to the russian game with the same name. Spin a wheel containing multiple options, when it stops, the chosen option wins. The probability for an option to be chosen depends on the angle size of its wheel section.

  • Rank Based Selection

Here, Solutions are ranked based on their fitness and the best n solutions are selected.

  • Stochastic Universal Sampling Selection

Umm, check it out on the Wiki i made a contribution to.

  • Tournament Selection

Here, a few individuals are selected and made to run a tournament to determine the individuals to be selected. Check out its Wiki.

HillClimb

Hill Climbing is a mathematical optimization technique which attempts to find a better solution by incrementally chaging a single element of an existing solution. It updates the existing solution if and only if the change leads to a better solution, repeating this sequence until no further improvements can be made.

Check out its wiki here.

Configuration

A Meta-Heuristic configuration object type ... Check it out here