rinde.sim.pdptw.vanlon14.VanLon14 Maven / Gradle / Ivy
The newest version!
package rinde.sim.pdptw.vanlon14;
import static com.google.common.base.Preconditions.checkState;
import static com.google.common.collect.Lists.newArrayList;
import java.io.File;
import java.io.IOException;
import java.util.List;
import javax.annotation.Nullable;
import org.apache.commons.math3.random.MersenneTwister;
import org.apache.commons.math3.random.RandomGenerator;
import org.joda.time.format.DateTimeFormatter;
import org.joda.time.format.ISODateTimeFormat;
import rinde.sim.core.graph.Point;
import rinde.sim.core.model.pdp.PDPScenarioEvent;
import rinde.sim.pdptw.common.DynamicPDPTWScenario.ProblemClass;
import rinde.sim.pdptw.common.ScenarioIO;
import rinde.sim.pdptw.generator.Analysis;
import rinde.sim.pdptw.generator.LoadRequirement;
import rinde.sim.pdptw.generator.Metrics;
import rinde.sim.pdptw.generator.ScenarioGenerator;
import rinde.sim.pdptw.generator.ScenarioGenerator.ScenarioFactory;
import rinde.sim.scenario.Scenario;
import rinde.sim.scenario.TimedEvent;
import rinde.sim.util.TimeWindow;
import com.google.common.base.Charsets;
import com.google.common.base.Function;
import com.google.common.base.Joiner;
import com.google.common.collect.ImmutableList;
import com.google.common.collect.ImmutableMap;
import com.google.common.collect.ImmutableMultiset;
import com.google.common.collect.Lists;
import com.google.common.collect.Multiset;
import com.google.common.io.Files;
public class VanLon14 {
public enum Dynamism {
LOW(1d / 3d),
MEDIUM(2d / 3d),
HIGH(1.0);
private final double value;
Dynamism(double val) {
value = val;
}
public double getValue() {
return value;
}
}
public enum Scale {
SMALL(10d),
MEDIUM(20d),
LARGE(30d);
private final double value;
Scale(double val) {
value = val;
}
public double getValue() {
return value;
}
}
public enum ExperimentClass implements ProblemClass {
LOW_SMALL(Dynamism.LOW, Scale.SMALL),
LOW_MEDIUM(Dynamism.LOW, Scale.MEDIUM),
LOW_LARGE(Dynamism.LOW, Scale.LARGE),
MEDIUM_SMALL(Dynamism.MEDIUM, Scale.SMALL),
MEDIUM_MEDIUM(Dynamism.MEDIUM, Scale.MEDIUM),
MEDIUM_LARGE(Dynamism.MEDIUM, Scale.LARGE),
HIGH_SMALL(Dynamism.HIGH, Scale.SMALL),
HIGH_MEDIUM(Dynamism.HIGH, Scale.MEDIUM),
HIGH_LARGE(Dynamism.HIGH, Scale.LARGE);
private final Dynamism dynamism;
private final Scale scale;
ExperimentClass(Dynamism d, Scale s) {
dynamism = d;
scale = s;
}
public Dynamism getDynamism() {
return dynamism;
}
public Scale getScale() {
return scale;
}
@Override
public String getId() {
return dynamism.name() + "-" + scale.name();
}
}
static ImmutableList generateMeanLoadGraph(
ScenarioGenerator generator, RandomGenerator rng, int numSamples) {
final List> allLoads = newArrayList();
for (int i = 0; i < numSamples; i++) {
final Scenario s = generator.generate(rng);
Metrics.checkTimeWindowStrictness(s);
final List loads = Metrics.measureRelativeLoad(s);
allLoads.add(loads);
}
final ImmutableList meanLoadGraph = ImmutableList.copyOf(Lists
.transform(
mean(allLoads),
new Function() {
@Override
@Nullable
public Double apply(@Nullable Double input) {
if (input == null || input.equals(0d)) {
return Double.MIN_VALUE;
}
return input;
}
}));
return meanLoadGraph;
}
public static void generateDataSet(File parentDir) {
final File dataSetDir = new File(parentDir, "vanlon14/");
try {
Files.createParentDirs(dataSetDir);
} catch (final IOException e) {
throw new IllegalStateException(e);
}
checkState(dataSetDir.exists() || dataSetDir.mkdir());
final DateTimeFormatter formatter = ISODateTimeFormat
.dateHourMinuteSecondMillis();
final RandomGenerator rng = new MersenneTwister(123);
// 480 minutes or 8 hours
final long dayLength = 480L;
// 0.1 vehicle per km2
final double vehicleDensity = .1d;
// number of orders per km2 per hour
final double orderIntensity = .20d;
// number of samples drawn from each problem class for generating mean load
// graph
final int numSamples = 50;
// number of scenarios to generate for each problem class
final int numScenarios = 5;
// deviation of the load graph
final double maxMeanDeviation = .2;
final double maxMaxDeviation = .5;
final List> allLoads = newArrayList();
for (final ExperimentClass ec : ExperimentClass.values()) {
final ScenarioGenerator generator =
ScenarioGenerator
.builder(new VanLon14ScenarioFactory(
ec))
.setOrderIntensityAndDynamism(orderIntensity,
ec.getDynamism().getValue())
.setScale(vehicleDensity, ec.getScale().getValue())
.setScenarioLength(dayLength)
.build();
final List meanLoads = generateMeanLoadGraph(generator, rng,
numSamples);
Analysis.writeLoads(meanLoads, new File(dataSetDir, "mean"
+ ec.name().toLowerCase() + ".load"));
allLoads.add(meanLoads);
}
final ImmutableList meanLoadGraph = ImmutableList.copyOf(Lists
.transform(
mean(allLoads),
new Function() {
@Override
@Nullable
public Double apply(@Nullable Double input) {
if (input == null || input.equals(0d)) {
return Double.MIN_VALUE;
}
return input;
}
}));
Analysis.writeLoads(meanLoadGraph, new File(dataSetDir, "global.load"));
// generation of scenarios starts here
for (final ExperimentClass ec : ExperimentClass.values()) {
final double scaledOrderIntensity = orderIntensity;
// if (ec.getScale() == Scale.MEDIUM) {
// scaledOrderIntensity = .22d;
// }
// else if (ec.getScale() == Scale.LARGE) {
// scaledOrderIntensity = .4d;
// }
// System.out.println(ec);
// System.out.println("order intensity: " + scaledOrderIntensity
// );
final ScenarioGenerator generator =
ScenarioGenerator
.builder(new VanLon14ScenarioFactory(
ec))
.setOrderIntensityAndDynamism(
scaledOrderIntensity,
ec.getDynamism().getValue())
.setScale(vehicleDensity,
ec.getScale().getValue())
.setScenarioLength(dayLength)
.addRequirement(
new LoadRequirement(meanLoadGraph, maxMeanDeviation,
maxMaxDeviation, true))
.build();
final List scenarios = generator.generate(rng,
numScenarios).scenarios;
final File dir = new File("files/dataset/vanlon14/"
+ ec.name().toLowerCase() + "/");
try {
Files.createParentDirs(dir);
checkState(dir.exists() || dir.mkdir(), "Could not create dir %s.", dir);
for (int i = 0; i < scenarios.size(); i++) {
final VanLon14Scenario s = scenarios.get(i);
final String scenarioName = ec
.name()
.toLowerCase() + "-" + i;
Files.write(ScenarioIO.write(s),
new File(dir, scenarioName + ".json"), Charsets.UTF_8);
Analysis.writeLoads(Metrics.measureRelativeLoad(s), new File(dir,
scenarioName + ".load"));
Analysis.writeLocationList(Metrics.getServicePoints(s), new File(dir,
scenarioName + ".points"));
Analysis.writeTimes(s.getTimeWindow().end,
Metrics.getArrivalTimes(s), new File(dir,
scenarioName + ".times"));
Metrics.checkTimeWindowStrictness(s);
final ImmutableMap.Builder properties = ImmutableMap
. builder()
.put("generation_date",
formatter.print(System.currentTimeMillis()))
// FIXME
// .put("dynamism", Metrics.measureDynamismOld(s))
.put("vehicle_speed_kmh", Metrics.getVehicleSpeed(s));
final ImmutableMultiset> eventTypes = Metrics
.getEventTypeCounts(s);
for (final Multiset.Entry> en : eventTypes.entrySet()) {
properties.put(en.getElement().name(), en.getCount());
}
final double parcelToVehicleRatio = (double) eventTypes
.count(PDPScenarioEvent.ADD_PARCEL)
/ eventTypes.count(PDPScenarioEvent.ADD_VEHICLE);
properties.put("parcel_to_vehicle_ratio", parcelToVehicleRatio);
// System.out.println(parcelToVehicleRatio + " " + eventTypes
// .count(PDPScenarioEvent.ADD_PARCEL) + " "
// + eventTypes.count(PDPScenarioEvent.ADD_VEHICLE));
Files.write(
Joiner.on("\n").withKeyValueSeparator(" = ")
.join(properties.build()),
new File(dir, scenarioName + ".properties"), Charsets.UTF_8);
}
} catch (final IOException e) {
throw new IllegalStateException(e);
}
}
}
public static class VanLon14ScenarioFactory implements
ScenarioFactory {
private final ProblemClass pc;
public VanLon14ScenarioFactory(
ProblemClass problemClass) {
pc = problemClass;
}
@Override
public VanLon14Scenario create(List events,
ScenarioGenerator generator, int instanceNumber) {
final TimeWindow tw = new TimeWindow(0, generator.getScenarioLength());
final Point min = generator.getMinPoint();
final Point max = generator.getMaxPoint();
final long tickSize = generator.getTickSize();
return new VanLon14Scenario(events, tw, min, max, tickSize, pc,
instanceNumber);
}
}
// use DoubleMath.mean(..) instead
@Deprecated
static List mean(List> lists) {
final ImmutableList.Builder builder = ImmutableList.builder();
boolean running = true;
int i = 0;
while (running) {
running = false;
double sum = 0d;
for (final List list : lists) {
if (i < list.size()) {
running = true;
sum += list.get(i);
}
}
sum /= lists.size();
builder.add(sum);
i++;
}
return builder.build();
}
}