"A Fast Approximation-Guided Evolutionary Multi-Objective Algorithm" Markus Wagner and Frank Neumann Genetic and Evolutionary Computation Conference 2013 http://cs.adelaide.edu.au/~markus/pub/2013gecco-age2.pdf Creator: Markus Wagner (wagner@acrocon.com) Feel free to contact me. There are many gems hidden in this code. /** * Usage example * * VIA COMMANDLINE * =============== * * /home/mwagner/scratch/age/age.sh ZDT1 ZDT1.pf 100 100000 BinaryTournament 0.01 zdtmu100 * * --> * * function: ZDT1 * Pareto front file: ZDT1.pf * population size mu: 100 * evaluations: 100000 * parent selection: BinaryTournament * epsilon grid for the approximating archive: 0.01 (note: try out different orders of magnitude) * output folder for the final population: zdtmu100 * * VIA METHOD CALLS * ================ * * you need to go via AGE.java: * - either execute a mini-study via AGE_main.java (see examples below) * - or execute a single run (see examples at the very bottom of AGE.java) * * DTLZ 1/2/3/4 * ============ * - Several DTLZ fronts are available in originalParetoFronts, due to space * reasons (several GBs), we do not provide fronts with >1000 points. However, * these can be easily generated using the uniform sampler in /ParetoFrontGenerator.java * For our publications, we sample the front 1.000.000 times. * - DTLZ 1 has the fronts DTLZ1.*.pf * - DTLZ 2/3/4 have the same fronts DTLZ2.*.pf * * Note: * the seventh parameter to AGE_main.main is for a powerful printer: * - yes: deactivates the regular print of the hypervolume + epsilon approximation of the current population * - no: regular prints * Important: the regular prints invoke costly computations, and thus should not be run for time-restricted studies! * */