Changeset 8982 for trunk/sources/HeuristicLab.Algorithms.DataAnalysis/3.4/GaussianProcess/MeanFunctions/MeanLinear.cs

Timestamp:

12/01/12 19:02:47 (11 years ago)

Author:

gkronber

Message:

#1902: removed class HyperParameter and changed implementations of covariance and mean functions to remove the parameter value caching and event handlers for parameter caching. Instead it is now possible to create the actual covariance and mean functions as Func from templates and specified parameter values. The instances of mean and covariance functions configured in the GUI are actually templates where the structure and fixed parameters can be specified.

File:

• trunk/sources/HeuristicLab.Algorithms.DataAnalysis/3.4/GaussianProcess/MeanFunctions/MeanLinear.cs (modified) (4 diffs)

trunk/sources/HeuristicLab.Algorithms.DataAnalysis/3.4/GaussianProcess/MeanFunctions/MeanLinear.cs

@@ -21 +21 @@
 
 using System;
+using System.Collections.Generic;
 using System.Linq;
 using HeuristicLab.Common;
 using HeuristicLab.Core;
 using HeuristicLab.Data;
+using HeuristicLab.Parameters;
 using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
 
@@ -31 +33 @@
   [Item(Name = "MeanLinear", Description = "Linear mean function for Gaussian processes.")]
   public sealed class MeanLinear : ParameterizedNamedItem, IMeanFunction {
-    [Storable]
-    private double[] weights;
-    [Storable]
-    private readonly HyperParameter<DoubleArray> weightsParameter;
-    public IValueParameter<DoubleArray> WeightsParameter { get { return weightsParameter; } }
+    public IValueParameter<DoubleArray> WeightsParameter {
+      get { return (IValueParameter<DoubleArray>)Parameters["Weights"]; }
+    }
 
     [StorableConstructor]
@@ -41 +41 @@
     private MeanLinear(MeanLinear original, Cloner cloner)
       : base(original, cloner) {
-      if (original.weights != null) {
-        this.weights = new double[original.weights.Length];
-        Array.Copy(original.weights, weights, original.weights.Length);
-      }
-      weightsParameter = cloner.Clone(original.weightsParameter);
-      RegisterEvents();
     }
     public MeanLinear()
       : base() {
-      this.weightsParameter = new HyperParameter<DoubleArray>("Weights", "The weights parameter for the linear mean function.");
-      Parameters.Add(weightsParameter);
-      RegisterEvents();
+      Parameters.Add(new OptionalValueParameter<DoubleArray>("Weights", "The weights parameter for the linear mean function."));
     }
 
@@ -59 +51 @@
     }
 
-    [StorableHook(HookType.AfterDeserialization)]
-    private void AfterDeserialization() {
-      RegisterEvents();
+    public int GetNumberOfParameters(int numberOfVariables) {
+      return WeightsParameter.Value != null ? 0 : numberOfVariables;
     }
 
-    private void RegisterEvents() {
-      Util.AttachArrayChangeHandler<DoubleArray, double>(weightsParameter, () => {
-        weights = weightsParameter.Value.ToArray();
-      });
+    public void SetParameter(double[] p) {
+      double[] weights;
+      GetParameter(p, out weights);
+      WeightsParameter.Value = new DoubleArray(weights);
     }
 
-    public int GetNumberOfParameters(int numberOfVariables) {
-      return weightsParameter.Fixed ? 0 : numberOfVariables;
+    public void GetParameter(double[] p, out double[] weights) {
+      if (WeightsParameter.Value == null) {
+        weights = p;
+      } else {
+        if (p.Length != 0) throw new ArgumentException("The length of the parameter vector does not match the number of free parameters for the linear mean function.", "p");
+        weights = WeightsParameter.Value.ToArray();
+      }
     }
 
-    public void SetParameter(double[] hyp) {
-      if (!weightsParameter.Fixed) {
-        weightsParameter.SetValue(new DoubleArray(hyp));
-      } else if (hyp.Length != 0) throw new ArgumentException("The length of the parameter vector does not match the number of free parameters for the linear mean function.", "hyp");
-    }
-
-    public double[] GetMean(double[,] x) {
-      // sanity check
-      if (weights.Length != x.GetLength(1)) throw new ArgumentException("The number of hyperparameters must match the number of variables for the linear mean function.");
-      int cols = x.GetLength(1);
-      int n = x.GetLength(0);
-      return (from i in Enumerable.Range(0, n)
-              let rowVector = Enumerable.Range(0, cols).Select(j => x[i, j])
-              select Util.ScalarProd(weights, rowVector))
-        .ToArray();
-    }
-
-    public double[] GetGradients(int k, double[,] x) {
-      int cols = x.GetLength(1);
-      int n = x.GetLength(0);
-      if (k > cols) throw new ArgumentException();
-      return (Enumerable.Range(0, n).Select(r => x[r, k])).ToArray();
+    public ParameterizedMeanFunction GetParameterizedMeanFunction(double[] p, IEnumerable<int> columnIndices) {
+      double[] weights;
+      int[] columns = columnIndices.ToArray();
+      GetParameter(p, out weights);
+      var mf = new ParameterizedMeanFunction();
+      mf.Mean = (x, i) => {
+        // sanity check
+        if (weights.Length != columns.Length) throw new ArgumentException("The number of rparameters must match the number of variables for the linear mean function.");
+        return Util.ScalarProd(weights, Util.GetRow(x, i, columns));
+      };
+      mf.Gradient = (x, i, k) => {
+        if (k > columns.Length) throw new ArgumentException();
+        return x[i, columns[k]];
+      };
+      return mf;
     }
   }