Skip to content

Commit

Permalink
Improved examples
Browse files Browse the repository at this point in the history
  • Loading branch information
Ledmington committed Apr 3, 2024
1 parent 3b0a52c commit e2ab253
Show file tree
Hide file tree
Showing 6 changed files with 450 additions and 224 deletions.
16 changes: 14 additions & 2 deletions examples/src/main/java/com/ledmington/gal/examples/Diet.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -136,6 +136,7 @@ public boolean equals(final Object other) {
}

public Diet() {
final long beginning = System.nanoTime();
final RandomGenerator rng = RandomGeneratorFactory.getDefault().create(System.nanoTime());

final Supplier<GeneticAlgorithmConfig.GeneticAlgorithmConfigBuilder<Solution>> state =
Expand Down Expand Up @@ -228,13 +229,15 @@ public Diet() {
Executors.newFixedThreadPool(Runtime.getRuntime().availableProcessors());
final GeneticAlgorithm<Solution> ga = new ParallelGeneticAlgorithm<>(ex, rng);
Set<Solution> g = new HashSet<>();
final Set<Solution> allSolutions = new HashSet<>();

for (int it = 0; it < 10; it++) {
System.out.printf("Run n.%,d\n", it);
ga.setState(state.get().firstGeneration(g).build());
ga.run();

final Map<Solution, Double> scores = ga.getState().scores();
allSolutions.addAll(scores.keySet());
scores.entrySet().stream()
.sorted(Map.Entry.comparingByValue())
.limit(10)
Expand All @@ -253,10 +256,19 @@ public Diet() {
System.out.println();
}

while (!ex.isShutdown()) {
final long end = System.nanoTime();

System.out.printf("\n%,d solutions evaluated\n", allSolutions.size());
System.out.printf("Total search time: %.3f seconds\n", (double) (end - beginning) / 1_000_000_000.0);

if (!ex.isShutdown()) {
ex.shutdown();
}
while (!ex.isTerminated()) {
try {
ex.awaitTermination(1, TimeUnit.SECONDS);
if (ex.awaitTermination(1, TimeUnit.SECONDS)) {
break;
}
} catch (final InterruptedException ignored) {
}
}
Expand Down
271 changes: 194 additions & 77 deletions examples/src/main/java/com/ledmington/gal/examples/GeneticTsp.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -17,13 +17,23 @@
*/
package com.ledmington.gal.examples;

import java.util.Arrays;
import java.math.BigDecimal;
import java.math.BigInteger;
import java.util.HashSet;
import java.util.Map;
import java.util.Objects;
import java.util.Set;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.TimeUnit;
import java.util.function.Supplier;
import java.util.random.RandomGenerator;
import java.util.random.RandomGeneratorFactory;
import java.util.stream.Collectors;

import com.ledmington.gal.GeneticAlgorithm;
import com.ledmington.gal.GeneticAlgorithmConfig;
import com.ledmington.gal.SerialGeneticAlgorithm;
import com.ledmington.gal.ParallelGeneticAlgorithm;

public final class GeneticTsp {

Expand All @@ -42,100 +52,207 @@ private static void shuffle(int[] arr) {
}
}

private record Solution(int[] array) {}
private static final class Solution {

public GeneticTsp() {
private final int[] array;
private final int cachedHashCode;

public Solution(final int[] v) {
this.array = Objects.requireNonNull(v);

int h = 17;
for (final int j : v) {
h = 31 * h + j;
}
cachedHashCode = h;
}

public int[] array() {
return array;
}

public int get(final int i) {
return array[i];
}

public int hashCode() {
return cachedHashCode;
}

public boolean equals(final Object other) {
if (other == null) {
return false;
}
if (this == other) {
return true;
}
if (!this.getClass().equals(other.getClass())) {
return false;
}
final Solution s = (Solution) other;
if (s.array.length != this.array.length) {
return false;
}
for (int i = 0; i < array.length; i++) {
if (this.array[i] != s.array[i]) {
return false;
}
}
return true;
}

public String toString() {
final StringBuilder sb = new StringBuilder();
sb.append("[");
boolean firstElement = true;
for (final int j : array) {
if (firstElement) {
firstElement = false;
} else {
sb.append(", ");
}
sb.append(String.format("%2d", j));
}
sb.append("]");
return sb.toString();
}
}

private static BigInteger factorial(BigInteger n) {
if (n.compareTo(BigInteger.ZERO) == 0 || n.compareTo(BigInteger.ONE) == 0) {
return BigInteger.ONE;
}
BigInteger x = BigInteger.ONE;
while (n.compareTo(BigInteger.ZERO) > 0) {
x = x.multiply(n);
n = n.subtract(BigInteger.ONE);
}
return x;
}

final int nCities = 20;
public GeneticTsp() {
final long beginning = System.nanoTime();
final int nCities = 30;
final double[][] coordinates = new double[2][nCities];

System.out.println("Traveling Salesman Problem's data:");
System.out.printf("Number of cities : %,d\n", nCities);
System.out.printf(
"Number of unique paths : %.3e\n",
new BigDecimal(factorial(BigInteger.valueOf(nCities - 1)).divide(BigInteger.TWO)));

System.out.println();
System.out.println("Cities coordinates:");
for (int i = 0; i < nCities; i++) {
coordinates[0][i] = rng.nextDouble(-10.0, 10.0);
coordinates[1][i] = rng.nextDouble(-10.0, 10.0);
System.out.printf("%2d: (%+.3f; %+.3f)\n", i, coordinates[0][i], coordinates[1][i]);
}
System.out.println();

final double[][] distances = new double[nCities][nCities];

for (int i = 0; i < nCities; i++) {
for (int j = 0; j < nCities; j++) {
// euclidean distance
final double dist =
Math.hypot(coordinates[0][i] - coordinates[0][j], coordinates[1][i] - coordinates[1][j]);
distances[i][j] = dist;
distances[j][i] = dist;
}
}

final GeneticAlgorithmConfig<Solution> config = GeneticAlgorithmConfig.<Solution>builder()
.populationSize(100)
.survivalRate(0.2)
.mutationRate(0.2)
.crossoverRate(0.1)
.creation(() -> {
final int[] v = new int[nCities];
for (int i = 0; i < nCities; i++) {
v[i] = i;
}
shuffle(v);
return new Solution(v);
})
.crossover((a, b) -> {
final int[] result = new int[nCities];

for (int i = 0; i < nCities; i++) {
final double choice = rng.nextDouble(0.0, 1.0);
if (choice < 0.45) {
result[i] = a.array()[i];
} else if (choice < 0.9) {
result[i] = b.array()[i];
} else {
result[i] = rng.nextInt(0, nCities);
}
}

int mistakes;
// we create a new input randomly mixing the two parents, until we get one that is valid
do {
final int i = rng.nextInt(0, nCities);
final double choice = rng.nextDouble(0.0, 1.0);
if (choice < 0.45) {
result[i] = a.array()[i];
} else if (choice < 0.9) {
result[i] = b.array()[i];
} else {
result[i] = rng.nextInt(0, nCities);
}

mistakes = (int)
(nCities - Arrays.stream(result).distinct().count());

} while (mistakes > 0);
return new Solution(result);
})
.mutation(x -> {
final int[] y = new int[nCities];
System.arraycopy(x.array(), 0, y, 0, nCities);
final int i = rng.nextInt(0, nCities);
int j;
do {
j = rng.nextInt(0, nCities);
} while (i == j);
int tmp = y[i];
y[i] = y[j];
y[j] = tmp;
return new Solution(y);
})
.minimize(x -> {
double s = 0.0;
for (int i = 0; i < nCities; i++) {
s += distances[x.array()[i]][x.array()[(i + 1) % nCities]];
}
// we return it inverted because it is a maximization algorithm for a minimization problem
return s;
})
.build();

final GeneticAlgorithm<Solution> ga = new SerialGeneticAlgorithm<>();
ga.setState(config);
ga.run();
final Supplier<GeneticAlgorithmConfig.GeneticAlgorithmConfigBuilder<Solution>> state =
() -> GeneticAlgorithmConfig.<Solution>builder()
.populationSize(1_000)
.maxGenerations(100)
.survivalRate(0.2)
.crossoverRate(0.7)
.mutationRate(0.2)
.creation(() -> {
final int[] v = new int[nCities];
for (int i = 0; i < nCities; i++) {
v[i] = i;
}
shuffle(v);
return new Solution(v);
})
.crossover((a, b) -> {
final int[] result = new int[nCities];

for (int i = 0; i < nCities; i++) {
result[i] = rng.nextBoolean() ? a.get(i) : b.get(i);
}

return new Solution(result);
})
.mutation(x -> {
final int[] y = new int[nCities];
System.arraycopy(x.array(), 0, y, 0, nCities);
final int i = rng.nextInt(0, nCities);
y[i] = rng.nextInt(0, nCities);
return new Solution(y);
})
.minimize(x -> {
final boolean[] visited = new boolean[nCities];
int visitedCities = 0;
double s = 0.0;
for (int i = 0; i < nCities; i++) {
if (!visited[x.get(i)]) {
visitedCities++;
}
visited[x.get(i)] = true;
s += distances[x.array()[i]][x.array()[(i + 1) % nCities]];
}
if (visitedCities < nCities) {
// This solution has not visited all cities, so this solution is not valid: we
// return a really high value (not infinity for now)
// FIXME: allow usage of infinities
return 1_000_000.0;
}
return s;
});

final ExecutorService ex =
Executors.newFixedThreadPool(Runtime.getRuntime().availableProcessors());
final GeneticAlgorithm<Solution> ga = new ParallelGeneticAlgorithm<>(ex, rng);
Set<Solution> g = new HashSet<>();
final Set<Solution> allSolutions = new HashSet<>();

for (int it = 0; it < 10; it++) {
System.out.printf("Run n.%,d\n", it);
ga.setState(state.get().firstGeneration(g).build());
ga.run();

final Map<Solution, Double> scores = ga.getState().scores();
allSolutions.addAll(scores.keySet());
scores.entrySet().stream()
.sorted(Map.Entry.comparingByValue())
.limit(10)
.forEach(e -> System.out.printf("%s -> %f\n", e.getKey(), e.getValue()));
g = scores.entrySet().stream()
.sorted(Map.Entry.comparingByValue())
.limit(10)
.map(Map.Entry::getKey)
.collect(Collectors.toSet());
System.out.println();
}

final long end = System.nanoTime();

System.out.printf("\n%,d solutions evaluated\n", allSolutions.size());
System.out.printf("Total search time: %.3f seconds\n", (double) (end - beginning) / 1_000_000_000.0);

if (!ex.isShutdown()) {
ex.shutdown();
}
while (!ex.isTerminated()) {
try {
if (ex.awaitTermination(1, TimeUnit.SECONDS)) {
break;
}
} catch (final InterruptedException ignored) {
}
}
}
}
Loading

0 comments on commit e2ab253

Please sign in to comment.