Add new images and update README with network modifications; improve ConsoleTable formatting

This commit is contained in:
2025-03-22 09:39:23 +01:00
parent e48bddf94d
commit 3844a46379
9 changed files with 1132 additions and 1044 deletions

View File

@@ -1,11 +1,13 @@
package net.berack.upo.valpre;
import java.util.concurrent.ExecutionException;
import java.util.List;
import java.util.function.BiFunction;
import net.berack.upo.valpre.rand.Distribution;
import net.berack.upo.valpre.sim.Net;
import net.berack.upo.valpre.sim.ServerNode;
import net.berack.upo.valpre.sim.SimulationMultiple;
import net.berack.upo.valpre.sim.stats.CsvResult;
import net.berack.upo.valpre.sim.stats.Result;
/**
@@ -16,30 +18,79 @@ import net.berack.upo.valpre.sim.stats.Result;
public final class NetExamples {
/**
* Main method to test the example networks.
* It runs the fist network and prints the results.
* The network will have the distribution changed but the mean will be the same.
* Main method to test the networks.
* The first network will have the distribution changed but the mean will be the
* same. The second network will have the distribution changed but the mean will
* be the same. The results will be saved to a csv file.
*
* @param args not needed
* @throws ExecutionException if the execution fails
* @throws InterruptedException if the execution is interrupted
* @throws Exception if the simulation fails or the file is not saved
*/
public static void main(String[] args) throws InterruptedException, ExecutionException {
var avg1 = 3.2;
var seed = 0l;
public static void main(String[] args) throws Exception {
var seed = 123456789L;
runNet(seed, 3.2, 1, "net1.csv", (spawn, dist) -> {
var name = dist.getClass().getSimpleName() + "_" + spawn;
return NetExamples.getNet1(spawn, name, dist);
});
runNet(seed, 1 / 3.5, 2, "net2.csv", (spawn, dist) -> {
var name = dist.getClass().getSimpleName() + "_" + spawn;
return NetExamples.getNet2(spawn, name, dist);
});
}
var nets = new Net[] {
getNet1("Normal", new Distribution.NormalBoxMuller(avg1, 0.6)),
getNet1("Exponential", new Distribution.Exponential(1 / avg1)),
getNet1("Erlang", new Distribution.Erlang(5, 5 / avg1)),
getNet1("Uniform", new Distribution.Uniform(avg1 - 1, avg1 + 1))
};
/**
* Method to test whatever network you input.
* The network will have the distribution changed but the mean will be the same.
* The bifunction requested is to get the network you want to test passing the
* spawn and the distribution with the same mean.
* The network will be tested with spawn totals of 1, 2, 5, 7, 10, 25, 50, 75,
* 100, 250, 500, 750, 1000, 1500, 2000.
* The results will be saved to a csv passed as argument.
*
* @param seed the seed for the simulation
* @param avg the mean of the distribution
* @param nodeToWatch the node to watch
* @param csv the file to save the results
* @param getNet the bifunction to get the network
* @throws Exception if the simulation fails or the file is not saved
*/
public static void runNet(long seed, double avg, int nodeToWatch, String csv,
BiFunction<Integer, Distribution, Net> getNet) throws Exception {
var build = new Result.Builder().seed(seed);
var spawnTotals = new int[] { 1, 2, 5, 7, 10, 25, 50, 75, 100, 250, 500, 750, 1000, 1500, 2000 };
for (var net : nets) {
var summary = new SimulationMultiple(net).runParallel(seed, 1000);
var table = Result.getResultString(summary.getNodes(), summary.getStats());
System.out.println(table);
var normal = new Distribution.NormalBoxMuller(avg, 0.6);
var exponential = new Distribution.Exponential(1 / avg);
var erlang = new Distribution.Erlang(5, 5 / avg);
var uniform = new Distribution.Uniform(avg - (avg * 0.1), avg + (avg * 0.1));
var hyper = new Distribution.HyperExponential(
new double[] { 1 / (avg * 0.5), 1 / (avg * 1.5) },
new double[] { 0.5f, 0.5f });
System.out.println("Normal: " + normal.mean);
System.out.println("Uniform: " + uniform.min + " - " + uniform.max);
for (var spawn : spawnTotals) {
System.out.println("Spawn: " + spawn);
var nets = new Net[] {
getNet.apply(spawn, normal),
getNet.apply(spawn, exponential),
getNet.apply(spawn, erlang),
getNet.apply(spawn, uniform),
getNet.apply(spawn, hyper),
};
for (var net : nets) {
var summary = new SimulationMultiple(net).runParallel(build.seed, 1000);
var name = net.getNode(nodeToWatch).name;
var stat = summary.getSummaryOf(name).average;
build.addNode(name, stat);
}
}
var result = build.build();
new CsvResult(csv).saveResults(List.of(result));
System.out.println("Results saved to " + csv);
}
/**
@@ -53,20 +104,21 @@ public final class NetExamples {
*/
public static Net getNet1() {
var norm3_2 = new Distribution.NormalBoxMuller(3.2, 0.6);
return getNet1("Queue", norm3_2);
return getNet1(10000, "Queue", norm3_2);
}
/**
* Return the first example network.
* The net is composed of a terminal node and a queue node.
* The terminal node is connected to the queue node.
* The terminal node generates 10000 jobs with an exponential distribution 4.5.
* The terminal node generates N jobs with an exponential distribution 4.5.
*
* @param spawn the number of jobs to generate
* @param name the name of the queue node
* @param queue the distribution of the queue node
* @return the first example network
*/
public static Net getNet1(String name, Distribution queue) {
var spawn = 10000;
public static Net getNet1(int spawn, String name, Distribution queue) {
var source = new Distribution.Exponential(1.0 / 4.5);
var net1 = new Net();
@@ -90,13 +142,8 @@ public final class NetExamples {
* @return the second example network
*/
public static Net getNet2() {
var exp1_5 = new Distribution.Exponential(1.5);
var exp2 = new Distribution.Exponential(2.0);
var exp3_5 = new Distribution.Exponential(3.5);
var exp10 = new Distribution.Exponential(10.0);
var unExp = new Distribution.UnavailableTime(0.1, exp10);
var spawn = 10000;
return getNet2(spawn, exp1_5, exp2, exp3_5, unExp);
return getNet2(10000, "Service2", exp3_5);
}
/**
@@ -106,18 +153,20 @@ public final class NetExamples {
* The first queue node is connected to the second queue node.
*
* @param spawn the number of jobs to generate
* @param source the distribution of the source node
* @param service1 the distribution of the first queue node
* @param name the name of the second queue node
* @param service2 the distribution of the second queue node
* @param unExp the distribution of the unavailable time
* @return the second example network
*/
public static Net getNet2(int spawn, Distribution source, Distribution service1, Distribution service2,
Distribution unExp) {
public static Net getNet2(int spawn, String name, Distribution service2) {
var exp1_5 = new Distribution.Exponential(1.5);
var exp2 = new Distribution.Exponential(2.0);
var exp10 = new Distribution.Exponential(10.0);
var unExp = new Distribution.UnavailableTime(0.1, exp10);
var net3 = new Net();
net3.addNode(ServerNode.Builder.terminal("Source", spawn, source));
net3.addNode(ServerNode.Builder.queue("Service1", 1, service1));
net3.addNode(ServerNode.Builder.queue("Service2", 1, service2, unExp));
net3.addNode(ServerNode.Builder.terminal("Source", spawn, exp1_5));
net3.addNode(ServerNode.Builder.queue("Service", 1, exp2));
net3.addNode(ServerNode.Builder.queue(name, 1, service2, unExp));
net3.addConnection(0, 1, 1.0);
net3.addConnection(1, 2, 1.0);
return net3;

View File

@@ -49,7 +49,7 @@ public class ConsoleTable {
builder.append('║');
builder.append(" ".repeat(first));
builder.append(val);
builder.append(" ".repeat(diff - first));
builder.append(" ".repeat(Math.max(diff - first, 0)));
}
builder.append("\n");