Package org.uma.evolver.example.base

Class NSGAIIBiObjectiveWithObserversTSP

java.lang.Object
org.uma.evolver.example.base.NSGAIIBiObjectiveWithObserversTSP
public class NSGAIIBiObjectiveWithObserversTSP extends Object
This class demonstrates the configuration and execution of NSGA-II (Non-dominated Sorting Genetic Algorithm II) for solving bi-objective Traveling Salesman Problem (TSP) instances, with additional runtime visualization. It extends the basic NSGA-II implementation by incorporating multiple observers for monitoring the algorithm's progress in real-time.

Key features of this implementation:

  • Solves the KroAB100 TSP instance with 100 cities
  • Uses a reference front from 'resources/referenceFrontsTSP/KroAB100TSP.csv' for comparison
  • Provides three types of runtime visualization:
    • Run-time chart showing the evolution of solutions in the objective space
    • Epsilon indicator plot for convergence analysis
    • Hypervolume indicator plot for diversity and convergence analysis

The algorithm is configured with the following components by default:

  • Population size: 100 solutions
  • Maximum evaluations: 1,000,000
  • Crossover: Cycle Crossover (CX) with probability 0.6848
  • Mutation: Swap mutation with probability 0.0973
  • Selection: Binary tournament selection

The results are output to two files:

  • VAR.csv: Contains the variable values (permutation of cities)
  • FUN.csv: Contains the objective function values
Author:
Antonio J. Nebro (ajnebro@uma.es)
See Also: