QuickXplainStrategy

/*******************************************************************************
 * SAT4J: a SATisfiability library for Java Copyright (C) 2004-2008 Daniel Le Berre
 *
 * All rights reserved. This program and the accompanying materials
 * are made available under the terms of the Eclipse Public License v1.0
 * which accompanies this distribution, and is available at
 * http://www.eclipse.org/legal/epl-v10.html
 *
 * Alternatively, the contents of this file may be used under the terms of
 * either the GNU Lesser General Public License Version 2.1 or later (the
 * "LGPL"), in which case the provisions of the LGPL are applicable instead
 * of those above. If you wish to allow use of your version of this file only
 * under the terms of the LGPL, and not to allow others to use your version of
 * this file under the terms of the EPL, indicate your decision by deleting
 * the provisions above and replace them with the notice and other provisions
 * required by the LGPL. If you do not delete the provisions above, a recipient
 * may use your version of this file under the terms of the EPL or the LGPL.
 *
 * Based on the original MiniSat specification from:
 *
 * An extensible SAT solver. Niklas Een and Niklas Sorensson. Proceedings of the
 * Sixth International Conference on Theory and Applications of Satisfiability
 * Testing, LNCS 2919, pp 502-518, 2003.
 *
 * See www.minisat.se for the original solver in C++.
 *
 *******************************************************************************/
package org.sat4j.tools.xplain;

import java.util.Map;
import java.util.Set;

import org.sat4j.core.VecInt;
import org.sat4j.specs.ISolver;
import org.sat4j.specs.IVecInt;
import org.sat4j.specs.IteratorInt;
import org.sat4j.specs.TimeoutException;

/**
 * An implementation of the QuickXplain algorithm as explained by Ulrich Junker
 * in the following paper:
 *
 * @inproceedings{DBLP:conf/aaai/Junker02, author = {Ulrich Junker}, title =
 *                                         {Preference-Based Search and
 *                                         Multi-Criteria Optimization},
 *                                         booktitle = {AAAI/IAAI}, year =
 *                                         {2002}, pages = {34-40}, bibsource =
 *                                         {DBLP, http://dblp.uni-trier.de} }
 *
 *
 *                                         The algorithm has been adapted to
 *                                         work properly in a context where we
 *                                         can afford to add a selector variable
 *                                         to each clause to enable or disable
 *                                         each constraint.
 *
 *                                         Note that for the moment, QuickXplain
 *                                         does not work properly in an
 *                                         optimization setting.
 *
 *
 * @since 2.1
 */
public class QuickXplainStrategy implements MinimizationStrategy {

	private boolean computationCanceled;

	public void cancelExplanationComputation() {
		computationCanceled = true;
	}

	public IVecInt explain(ISolver solver, Map<Integer, ?> constrs,
			IVecInt assumps) throws TimeoutException {
		computationCanceled = false;
		IVecInt encodingAssumptions = new VecInt(constrs.size()
				+ assumps.size());
		assumps.copyTo(encodingAssumptions);
		IVecInt firstExplanation = solver.unsatExplanation();
		IVecInt results = new VecInt(firstExplanation.size());
		if (firstExplanation.size() == 1) {
			results.push(-firstExplanation.get(0));
			return results;
		}
		if (solver.isVerbose()) {
			System.out.print(solver.getLogPrefix() + "initial unsat core ");
			firstExplanation.sort();
			for (IteratorInt it = firstExplanation.iterator(); it.hasNext();) {
				System.out.print(constrs.get(-it.next()));
				System.out.print(" ");
			}
			System.out.println();
		}
		for (int i = 0; i < firstExplanation.size();) {
			if (assumps.contains(firstExplanation.get(i))) {
				firstExplanation.delete(i);
			} else {
				i++;
			}
		}
		Set<Integer> constraintsVariables = constrs.keySet();
		IVecInt remainingVariables = new VecInt(constraintsVariables.size());
		for (Integer v : constraintsVariables) {
			remainingVariables.push(v);
		}
		int p;
		for (IteratorInt it = firstExplanation.iterator(); it.hasNext();) {
			p = it.next();
			if (p < 0) {
				p = -p;
			}
			remainingVariables.remove(p);
			encodingAssumptions.push(p);
		}
		int unsatcorelimit = encodingAssumptions.size() - 1;

		remainingVariables.copyTo(encodingAssumptions);
		computeExplanation(solver, constrs, encodingAssumptions,
				assumps.size(), unsatcorelimit, results);
		return results;
	}

	private void computeExplanation(ISolver solver, Map<Integer, ?> constrs,
			IVecInt encodingAssumptions, int start, int end, IVecInt result)
			throws TimeoutException {
		if (solver.isVerbose()) {
			System.out.println(solver.getLogPrefix() + "qxplain " + start + "/"
					+ end);
		}
		if (!solver.isSatisfiable(encodingAssumptions)) {
			return;
		}
		if (start == end) {
			result.push(encodingAssumptions.get(start));
			encodingAssumptions.set(start, -encodingAssumptions.get(start));
			if (solver.isVerbose()) {
				System.out.println(solver.getLogPrefix()
						+ constrs.get(-encodingAssumptions.get(start))
						+ " is mandatory ");
			}
			return;
		}
		int split = (end + start) / 2;
		if (split < end) {
			for (int j = start; j <= split; j++) {
				encodingAssumptions.set(j, -encodingAssumptions.get(j));
			}
			computeExplanation(solver, constrs, encodingAssumptions, split + 1,
					end, result);
		}
		if (start <= split) {
			for (int j = start; j <= split; j++) {
				encodingAssumptions.set(j, -encodingAssumptions.get(j));
			}
			computeExplanation(solver, constrs, encodingAssumptions, start,
					split, result);
		}
		if (computationCanceled) {
			throw new TimeoutException();
		}
	}

	@Override
	public String toString() {
		return "QuickXplain (AAAI 2004 version) minimization strategy";
	}

}