source: branches/2839_HiveProjectManagement/HeuristicLab.Algorithms.DataAnalysis/3.4/GaussianProcess/MeanFunctions/MeanProduct.cs @ 16057

Last change on this file since 16057 was 16057, checked in by jkarder, 15 months ago

#2839:

File size: 4.3 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2018 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
4 *
5 * This file is part of HeuristicLab.
6 *
7 * HeuristicLab is free software: you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation, either version 3 of the License, or
10 * (at your option) any later version.
11 *
12 * HeuristicLab is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
19 */
20#endregion
21
22using System.Collections.Generic;
23using System.Linq;
24using HeuristicLab.Common;
25using HeuristicLab.Core;
26using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
27
28namespace HeuristicLab.Algorithms.DataAnalysis {
29  [StorableClass]
30  [Item(Name = "MeanProduct", Description = "Product of mean functions for Gaussian processes.")]
31  public sealed class MeanProduct : Item, IMeanFunction {
32    [Storable]
33    private ItemList<IMeanFunction> factors;
34
35    [Storable]
36    private int numberOfVariables;
37
38    public ItemList<IMeanFunction> Factors {
39      get { return factors; }
40    }
41
42    [StorableConstructor]
43    private MeanProduct(bool deserializing)
44      : base(deserializing) {
45    }
46
47    private MeanProduct(MeanProduct original, Cloner cloner)
48      : base(original, cloner) {
49      this.factors = cloner.Clone(original.factors);
50      this.numberOfVariables = original.numberOfVariables;
51    }
52
53    public MeanProduct() {
54      this.factors = new ItemList<IMeanFunction>();
55    }
56    public override IDeepCloneable Clone(Cloner cloner) {
57      return new MeanProduct(this, cloner);
58    }
59
60    public int GetNumberOfParameters(int numberOfVariables) {
61      this.numberOfVariables = numberOfVariables;
62      return factors.Select(t => t.GetNumberOfParameters(numberOfVariables)).Sum();
63    }
64
65    public void SetParameter(double[] p) {
66      int offset = 0;
67      foreach (var t in factors) {
68        var numberOfParameters = t.GetNumberOfParameters(numberOfVariables);
69        t.SetParameter(p.Skip(offset).Take(numberOfParameters).ToArray());
70        offset += numberOfParameters;
71      }
72    }
73
74
75    public ParameterizedMeanFunction GetParameterizedMeanFunction(double[] p, int[] columnIndices) {
76      var factorMf = new List<ParameterizedMeanFunction>();
77      int totalNumberOfParameters = GetNumberOfParameters(numberOfVariables);
78      int[] factorIndexMap = new int[totalNumberOfParameters]; // maps k-th hyperparameter to the correct mean-term
79      int[] hyperParameterIndexMap = new int[totalNumberOfParameters]; // maps k-th hyperparameter to the l-th hyperparameter of the correct mean-term
80      int c = 0;
81      // get the parameterized mean function for each term
82      for (int factorIndex = 0; factorIndex < factors.Count; factorIndex++) {
83        var numberOfParameters = factors[factorIndex].GetNumberOfParameters(numberOfVariables);
84        factorMf.Add(factors[factorIndex].GetParameterizedMeanFunction(p.Take(numberOfParameters).ToArray(), columnIndices));
85        p = p.Skip(numberOfParameters).ToArray();
86
87        for (int hyperParameterIndex = 0; hyperParameterIndex < numberOfParameters; hyperParameterIndex++) {
88          factorIndexMap[c] = factorIndex;
89          hyperParameterIndexMap[c] = hyperParameterIndex;
90          c++;
91        }
92      }
93
94      var mf = new ParameterizedMeanFunction();
95      mf.Mean = (x, i) => factorMf.Select(t => t.Mean(x, i)).Aggregate((a, b) => a * b);
96      mf.Gradient = (x, i, k) => {
97        double result = 1.0;
98        int hyperParameterFactorIndex = factorIndexMap[k];
99        for (int factorIndex = 0; factorIndex < factors.Count; factorIndex++) {
100          if (factorIndex == hyperParameterFactorIndex) {
101            // multiply gradient
102            result *= factorMf[factorIndex].Gradient(x, i, hyperParameterIndexMap[k]);
103          } else {
104            // multiply mean
105            result *= factorMf[factorIndex].Mean(x, i);
106          }
107        }
108        return result;
109      };
110      return mf;
111    }
112  }
113}
Note: See TracBrowser for help on using the repository browser.