org.cicirello.experiments.modifiedlam.BoundMaxExperiment Maven / Gradle / Ivy
Show all versions of modified-lam-experiments Show documentation
/*
* Experiments with the original version, and optimized version,
* of the Modified Lam annealing schedule.
* Copyright (C) 2020 Vincent A. Cicirello
*
* This program is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see Driver program for experiment comparing the runtime of
* the original Modified Lam annealing schedule to an optimized
* version of the Modified Lam annealing schedule, on the
* BoundMax problem. The BoundMax problem is a generalization of
* the OneMax problem to non-binary strings, specifically to general
* vectors of integers, from the interval [0,B] for some bound B.
* It was chosen for the experiment to have a simple optimization
* problem with an integer representation of solutions.
*
* Output of the program is a table consisting of the
* following columns:
* length cost1 cost2 cpu1 cpu2
* where the length is the number of simulated annealing evaluations,
* the cost1 is the best of run value of the cost function for
* the original Modified Lam (and cost2 for the optimized version),
* and cpu1 is the amount of cpu time (in nanoseconds) for the original
* Modified Lam schedule (cpu2 is the same but for the optimized version).
*
* The cost function values are included in the output to confirm that
* there is no effect on the cost function between the two versions, since
* the sequences of temperatures, and target acceptance rates should be equivalent
* between the two versions.
*
* @author Vincent A. Cicirello,
* https://www.cicirello.org/
*/
public class BoundMaxExperiment {
/**
* Runs the experiment.
* @param args There are no command line arguments.
*/
public static void main(String[] args) {
final int WARMUP_NUM_SAMPLES = 10;
final int NUM_SAMPLES = 100;
final int N = 650;
final int B = 127;
final int MIN_RUN_LENGTH = 10000;
final int MAX_RUN_LENGTH = 1000000;
BoundMax problem = new BoundMax(N, B);
final double P = 1.0/N;
final int K = 1;
// Warm up JVM prior to timing alternatives
// The warm up phase uses the longest run length.
for (int i = 0; i < WARMUP_NUM_SAMPLES; i++) {
SimulatedAnnealing sa1 = new SimulatedAnnealing(
problem,
new UndoableRandomValueChangeMutation(0, B, P, K),
(Initializer)problem.split(),
new ModifiedLamOriginal()
);
SimulatedAnnealing sa2 = new SimulatedAnnealing(
problem,
new UndoableRandomValueChangeMutation(0, B, P, K),
(Initializer)problem.split(),
new ModifiedLam()
);
sa1.optimize(MAX_RUN_LENGTH);
sa2.optimize(MAX_RUN_LENGTH);
}
// End warm up
ThreadMXBean bean = ManagementFactory.getThreadMXBean();
System.out.printf("%7s\t%5s\t%5s\t%12s\t%12s\n",
"length",
"cost1",
"cost2",
"cpu1",
"cpu2"
);
for (int runLength = MIN_RUN_LENGTH; runLength <= MAX_RUN_LENGTH; runLength *= 10) {
for (int i = 0; i < NUM_SAMPLES; i++) {
SimulatedAnnealing sa1 = new SimulatedAnnealing(
problem,
new UndoableRandomValueChangeMutation(0, B, P, K),
(Initializer)problem.split(),
new ModifiedLamOriginal()
);
SimulatedAnnealing sa2 = new SimulatedAnnealing(
problem,
new UndoableRandomValueChangeMutation(0, B, P, K),
(Initializer)problem.split(),
new ModifiedLam()
);
long start = bean.getCurrentThreadCpuTime();
sa1.optimize(runLength);
long mid = bean.getCurrentThreadCpuTime();
sa2.optimize(runLength);
long end = bean.getCurrentThreadCpuTime();
System.out.printf("%7d\t%5d\t%5d\t%12d\t%12d\n",
runLength,
sa1.getProgressTracker().getCost(),
sa2.getProgressTracker().getCost(),
mid-start,
end-mid
);
}
}
}
}