001 /*
002 * Licensed to the Apache Software Foundation (ASF) under one or more
003 * contributor license agreements. See the NOTICE file distributed with
004 * this work for additional information regarding copyright ownership.
005 * The ASF licenses this file to You under the Apache License, Version 2.0
006 * (the "License"); you may not use this file except in compliance with
007 * the License. You may obtain a copy of the License at
008 *
009 * http://www.apache.org/licenses/LICENSE-2.0
010 *
011 * Unless required by applicable law or agreed to in writing, software
012 * distributed under the License is distributed on an "AS IS" BASIS,
013 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
014 * See the License for the specific language governing permissions and
015 * limitations under the License.
016 */
017 package org.apache.commons.math3.optim.nonlinear.vector;
018
019 import java.util.Collections;
020 import java.util.List;
021 import java.util.ArrayList;
022 import java.util.Comparator;
023 import org.apache.commons.math3.exception.NotStrictlyPositiveException;
024 import org.apache.commons.math3.exception.NullArgumentException;
025 import org.apache.commons.math3.linear.RealMatrix;
026 import org.apache.commons.math3.linear.RealVector;
027 import org.apache.commons.math3.linear.ArrayRealVector;
028 import org.apache.commons.math3.random.RandomVectorGenerator;
029 import org.apache.commons.math3.optim.BaseMultiStartMultivariateOptimizer;
030 import org.apache.commons.math3.optim.PointVectorValuePair;
031
032 /**
033 * Multi-start optimizer for a (vector) model function.
034 *
035 * This class wraps an optimizer in order to use it several times in
036 * turn with different starting points (trying to avoid being trapped
037 * in a local extremum when looking for a global one).
038 *
039 * @version $Id$
040 * @since 3.0
041 */
042 public class MultiStartMultivariateVectorOptimizer
043 extends BaseMultiStartMultivariateOptimizer<PointVectorValuePair> {
044 /** Underlying optimizer. */
045 private final MultivariateVectorOptimizer optimizer;
046 /** Found optima. */
047 private final List<PointVectorValuePair> optima = new ArrayList<PointVectorValuePair>();
048
049 /**
050 * Create a multi-start optimizer from a single-start optimizer.
051 *
052 * @param optimizer Single-start optimizer to wrap.
053 * @param starts Number of starts to perform.
054 * If {@code starts == 1}, the result will be same as if {@code optimizer}
055 * is called directly.
056 * @param generator Random vector generator to use for restarts.
057 * @throws NullArgumentException if {@code optimizer} or {@code generator}
058 * is {@code null}.
059 * @throws NotStrictlyPositiveException if {@code starts < 1}.
060 */
061 public MultiStartMultivariateVectorOptimizer(final MultivariateVectorOptimizer optimizer,
062 final int starts,
063 final RandomVectorGenerator generator)
064 throws NullArgumentException,
065 NotStrictlyPositiveException {
066 super(optimizer, starts, generator);
067 this.optimizer = optimizer;
068 }
069
070 /**
071 * {@inheritDoc}
072 */
073 @Override
074 public PointVectorValuePair[] getOptima() {
075 Collections.sort(optima, getPairComparator());
076 return optima.toArray(new PointVectorValuePair[0]);
077 }
078
079 /**
080 * {@inheritDoc}
081 */
082 @Override
083 protected void store(PointVectorValuePair optimum) {
084 optima.add(optimum);
085 }
086
087 /**
088 * {@inheritDoc}
089 */
090 @Override
091 protected void clear() {
092 optima.clear();
093 }
094
095 /**
096 * @return a comparator for sorting the optima.
097 */
098 private Comparator<PointVectorValuePair> getPairComparator() {
099 return new Comparator<PointVectorValuePair>() {
100 private final RealVector target = new ArrayRealVector(optimizer.getTarget(), false);
101 private final RealMatrix weight = optimizer.getWeight();
102
103 public int compare(final PointVectorValuePair o1,
104 final PointVectorValuePair o2) {
105 if (o1 == null) {
106 return (o2 == null) ? 0 : 1;
107 } else if (o2 == null) {
108 return -1;
109 }
110 return Double.compare(weightedResidual(o1),
111 weightedResidual(o2));
112 }
113
114 private double weightedResidual(final PointVectorValuePair pv) {
115 final RealVector v = new ArrayRealVector(pv.getValueRef(), false);
116 final RealVector r = target.subtract(v);
117 return r.dotProduct(weight.operate(r));
118 }
119 };
120 }
121 }