FitnessEvaluatorVirtualThread.java

package net.bmahe.genetics4j.core.evaluation;

import java.util.ArrayList;

import java.util.List;

import java.util.Objects;

import java.util.concurrent.CompletableFuture;

import java.util.concurrent.ExecutorService;

import java.util.concurrent.Executors;

import org.apache.commons.lang3.Validate;

import org.apache.logging.log4j.LogManager;

import org.apache.logging.log4j.Logger;

import net.bmahe.genetics4j.core.Fitness;

import net.bmahe.genetics4j.core.Genotype;

import net.bmahe.genetics4j.core.spec.EAConfiguration;

import net.bmahe.genetics4j.core.spec.EAExecutionContext;

/**

 * Virtual thread-based fitness evaluator that creates one virtual thread per individual evaluation.

 *

 * <p>This evaluator leverages Java 21+ virtual threads to provide massive parallelism for fitness evaluation without

 * the overhead of traditional platform threads. Each genotype gets its own virtual thread, making this approach

 * particularly suitable for:

 *

 * <ul>

 * <li><strong>I/O-bound fitness functions</strong>: Database queries, web service calls, file operations</li>

 * <li><strong>Large populations</strong>: Thousands of individuals without thread pool limitations</li>

 * <li><strong>Complex simulations</strong>: Long-running evaluations that benefit from massive parallelism</li>

 * <li><strong>External service integration</strong>: Fitness evaluations requiring network calls</li>

 * </ul>

 *

 * <p>Unlike traditional thread pool-based evaluators, this implementation:

 * <ul>

 * <li>Creates one virtual thread per genotype (no partitioning)</li>

 * <li>Leverages virtual thread's lightweight nature for maximum concurrency</li>

 * <li>Automatically manages virtual thread lifecycle</li>

 * <li>Provides optimal resource utilization for I/O-intensive workloads</li>

 * </ul>

 *

 * <p>Performance characteristics:

 * <ul>

 * <li><strong>Memory overhead</strong>: ~200 bytes per virtual thread vs ~2MB per platform thread</li>

 * <li><strong>Creation cost</strong>: Nearly zero compared to platform threads</li>

 * <li><strong>Blocking behavior</strong>: Virtual threads don't block carrier threads during I/O</li>

 * <li><strong>Scalability</strong>: Can handle millions of concurrent virtual threads</li>

 * </ul>

 *

 * <p>This evaluator automatically creates and manages a virtual thread executor, eliminating the need for explicit

 * thread pool configuration and management.

 *

 * @param <T> the type of fitness values produced, must be comparable for selection operations

 * @see FitnessEvaluator

 * @see FitnessEvaluatorSync

 * @see net.bmahe.genetics4j.core.Fitness

 */

public class FitnessEvaluatorVirtualThread<T extends Comparable<T>> implements FitnessEvaluator<T> {

	private final EAExecutionContext<T> eaExecutionContext;

	private final EAConfiguration<T> eaConfiguration;

	private final ExecutorService virtualThreadExecutor;

	public FitnessEvaluatorVirtualThread(final EAExecutionContext<T> _eaExecutionContext,

	@Override

	public void preEvaluation() {

	@Override

	public void postEvaluation() {

	@Override

	public List<T> evaluate(final long generation, final List<Genotype> population) {

		}

		// Create one virtual thread task per genotype - no partitioning needed

			}, virtualThreadExecutor);

		// Wait for all evaluations to complete and collect results in order

	}

}

Mutations

1.1
Location : <init>
Killed by : none Removed assignment to member variable eaExecutionContext → SURVIVED Covering tests

1.1
Location : <init>
Killed by : none Removed assignment to member variable eaConfiguration → SURVIVED Covering tests

1.1
Location : <init>
Killed by : none Removed assignment to member variable virtualThreadExecutor → SURVIVED Covering tests

2.2
Location : <init>
Killed by : none removed call to java/util/concurrent/Executors::newVirtualThreadPerTaskExecutor → SURVIVED Covering tests