Changeset 13247 for branches/HeuristicLab.LinqExpressionTreeInterpreter

Timestamp:

11/18/15 13:39:32 (9 years ago)

Author:

mkommend

Message:

#2442: Merged trunk changes into branch for compiled interpreter.

Location:

branches/HeuristicLab.LinqExpressionTreeInterpreter/HeuristicLab.Problems.DataAnalysis.Symbolic

Files:

• . (modified) (1 prop)
• 3.4/Analyzers/SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer.cs (modified) (6 diffs)
• 3.4/Crossovers/SymbolicDataAnalysisExpressionSizefairCrossover.cs (deleted)
• 3.4/Formatters/SymbolicDataAnalysisExpressionCSharpFormatter.cs (modified) (3 diffs)
• 3.4/HeuristicLab.Problems.DataAnalysis.Symbolic-3.4.csproj (modified) (1 diff)
• 3.4/SymbolicDataAnalysisExpressionTreeSimplificationOperator.cs (copied) (copied from trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/SymbolicDataAnalysisExpressionTreeSimplificationOperator.cs)
• 3.4/SymbolicDataAnalysisModelComplexityCalculator.cs (copied) (copied from trunk/sources/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/SymbolicDataAnalysisModelComplexityCalculator.cs)

branches/HeuristicLab.LinqExpressionTreeInterpreter/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Analyzers/SymbolicDataAnalysisMultiObjectiveTrainingBestSolutionAnalyzer.cs

@@ -20 +20 @@
 #endregion
 
-using System;
 using System.Collections.Generic;
 using System.Linq;
@@ -42 +41 @@
     private const string TrainingBestSolutionQualitiesParameterName = "Best training solution qualities";
     private const string UpdateAlwaysParameterName = "Always update best solutions";
+    private const string TrainingBestSolutionParameterName = "Best training solution";
 
     #region parameter properties
@@ -55 +55 @@
     #endregion
     #region properties
-    public ItemList<T> TrainingBestSolutions {
+    private ItemList<T> TrainingBestSolutions {
       get { return TrainingBestSolutionsParameter.ActualValue; }
       set { TrainingBestSolutionsParameter.ActualValue = value; }
     }
-    public ItemList<DoubleArray> TrainingBestSolutionQualities {
+    private ItemList<DoubleArray> TrainingBestSolutionQualities {
       get { return TrainingBestSolutionQualitiesParameter.ActualValue; }
       set { TrainingBestSolutionQualitiesParameter.ActualValue = value; }
     }
-    public BoolValue UpdateAlways {
-      get { return UpdateAlwaysParameter.Value; }
+    public bool UpdateAlways {
+      get { return UpdateAlwaysParameter.Value.Value; }
+      set { UpdateAlwaysParameter.Value.Value = value; }
     }
     #endregion
@@ -97 +98 @@
       }
 
+      if (!results.ContainsKey(TrainingBestSolutionParameterName)) {
+        results.Add(new Result(TrainingBestSolutionParameterName, "", typeof(ISymbolicDataAnalysisSolution)));
+      }
+
       //if the pareto front of best solutions shall be updated regardless of the quality, the list initialized empty to discard old solutions
-      IList<double[]> trainingBestQualities;
-      if (UpdateAlways.Value) {
+      List<double[]> trainingBestQualities;
+      if (UpdateAlways) {
         trainingBestQualities = new List<double[]>();
       } else {
@@ -105 +110 @@
       }
 
-      #region find best trees
-      IList<int> nonDominatedIndexes = new List<int>();
-      ISymbolicExpressionTree[] tree = SymbolicExpressionTree.ToArray();
+      ISymbolicExpressionTree[] trees = SymbolicExpressionTree.ToArray();
       List<double[]> qualities = Qualities.Select(x => x.ToArray()).ToList();
       bool[] maximization = Maximization.ToArray();
-      List<double[]> newNonDominatedQualities = new List<double[]>();
-      for (int i = 0; i < tree.Length; i++) {
-        if (IsNonDominated(qualities[i], trainingBestQualities, maximization) &&
-          IsNonDominated(qualities[i], qualities, maximization)) {
-          if (!newNonDominatedQualities.Contains(qualities[i], new DoubleArrayComparer())) {
-            newNonDominatedQualities.Add(qualities[i]);
-            nonDominatedIndexes.Add(i);
+
+      var nonDominatedInvididuals = new[] { new { Tree = default(ISymbolicExpressionTree), Qualities = default(double[]) } }.ToList();
+      nonDominatedInvididuals.Clear();
+
+      // build list of new non-dominated solutions
+      for (int i = 0; i < trees.Length; i++) {
+        if (IsNonDominated(qualities[i], nonDominatedInvididuals.Select(ind => ind.Qualities), maximization) &&
+            IsNonDominated(qualities[i], trainingBestQualities, maximization)) {
+          for (int j = nonDominatedInvididuals.Count - 1; j >= 0; j--) {
+            if (IsBetterOrEqual(qualities[i], nonDominatedInvididuals[j].Qualities, maximization)) {
+              nonDominatedInvididuals.RemoveAt(j);
+            }
           }
+          nonDominatedInvididuals.Add(new { Tree = trees[i], Qualities = qualities[i] });
         }
       }
-      #endregion
+
+      var nonDominatedSolutions = nonDominatedInvididuals.Select(x => new { Solution = CreateSolution(x.Tree, x.Qualities), Qualities = x.Qualities }).ToList();
+      nonDominatedSolutions.ForEach(s => s.Solution.Name = string.Join(",", s.Qualities.Select(q => q.ToString())));
+
       #region update Pareto-optimal solution archive
-      if (nonDominatedIndexes.Count > 0) {
-        ItemList<DoubleArray> nonDominatedQualities = new ItemList<DoubleArray>();
-        ItemList<T> nonDominatedSolutions = new ItemList<T>();
-        // add all new non-dominated solutions to the archive
-        foreach (var index in nonDominatedIndexes) {
-          T solution = CreateSolution(tree[index], qualities[index]);
-          nonDominatedSolutions.Add(solution);
-          nonDominatedQualities.Add(new DoubleArray(qualities[index]));
-        }
-        // add old non-dominated solutions only if they are not dominated by one of the new solutions
+      if (nonDominatedSolutions.Count > 0) {
+        //add old non-dominated solutions only if they are not dominated by one of the new solutions
         for (int i = 0; i < trainingBestQualities.Count; i++) {
-          if (IsNonDominated(trainingBestQualities[i], newNonDominatedQualities, maximization)) {
-            if (!newNonDominatedQualities.Contains(trainingBestQualities[i], new DoubleArrayComparer())) {
-              nonDominatedSolutions.Add(TrainingBestSolutions[i]);
-              nonDominatedQualities.Add(TrainingBestSolutionQualities[i]);
-            }
+          if (IsNonDominated(trainingBestQualities[i], nonDominatedSolutions.Select(x => x.Qualities), maximization)) {
+            nonDominatedSolutions.Add(new { Solution = TrainingBestSolutions[i], Qualities = TrainingBestSolutionQualities[i].ToArray() });
           }
         }
 
-        results[TrainingBestSolutionsParameter.Name].Value = nonDominatedSolutions;
-        results[TrainingBestSolutionQualitiesParameter.Name].Value = nonDominatedQualities;
+        //assumes the the first objective is always the accuracy
+        var sortedNonDominatedSolutions = maximization[0]
+          ? nonDominatedSolutions.OrderByDescending(x => x.Qualities[0])
+          : nonDominatedSolutions.OrderBy(x => x.Qualities[0]);
+        var trainingBestSolution = sortedNonDominatedSolutions.Select(s => s.Solution).First();
+        results[TrainingBestSolutionParameterName].Value = trainingBestSolution;
+        TrainingBestSolutions = new ItemList<T>(sortedNonDominatedSolutions.Select(x => x.Solution));
+        results[TrainingBestSolutionsParameter.Name].Value = TrainingBestSolutions;
+        TrainingBestSolutionQualities = new ItemList<DoubleArray>(sortedNonDominatedSolutions.Select(x => new DoubleArray(x.Qualities)));
+        results[TrainingBestSolutionQualitiesParameter.Name].Value = TrainingBestSolutionQualities;
       }
       #endregion
@@ -148 +157 @@
     }
 
-    private class DoubleArrayComparer : IEqualityComparer<double[]> {
-      public bool Equals(double[] x, double[] y) {
-        if (y.Length != x.Length) throw new ArgumentException();
-        for (int i = 0; i < x.Length; i++) {
-          if (!x[i].IsAlmost(y[i])) return false;
-        }
-        return true;
-      }
-
-      public int GetHashCode(double[] obj) {
-        int c = obj.Length;
-        for (int i = 0; i < obj.Length; i++)
-          c ^= obj[i].GetHashCode();
-        return c;
-      }
-    }
-
     protected abstract T CreateSolution(ISymbolicExpressionTree bestTree, double[] bestQuality);
 
-    private bool IsNonDominated(double[] point, IList<double[]> points, bool[] maximization) {
+    private bool IsNonDominated(double[] point, IEnumerable<double[]> points, bool[] maximization) {
       foreach (var refPoint in points) {
-        bool refPointDominatesPoint = true;
-        for (int i = 0; i < point.Length; i++) {
-          refPointDominatesPoint &= IsBetterOrEqual(refPoint[i], point[i], maximization[i]);
-        }
+        bool refPointDominatesPoint = IsBetterOrEqual(refPoint, point, maximization);
         if (refPointDominatesPoint) return false;
       }
       return true;
     }
+
+    private bool IsBetterOrEqual(double[] lhs, double[] rhs, bool[] maximization) {
+      for (int i = 0; i < lhs.Length; i++) {
+        var result = IsBetterOrEqual(lhs[i], rhs[i], maximization[i]);
+        if (!result) return false;
+      }
+      return true;
+    }
+
     private bool IsBetterOrEqual(double lhs, double rhs, bool maximization) {
-      if (maximization) return lhs > rhs;
-      else return lhs < rhs;
+      if (maximization) return lhs >= rhs;
+      else return lhs <= rhs;
     }
   }

branches/HeuristicLab.LinqExpressionTreeInterpreter/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Formatters/SymbolicDataAnalysisExpressionCSharpFormatter.cs

@@ -89 +89 @@
           FormatFunction(node, "Math.Sin", strBuilder);
         } else if (node.Symbol is Subtraction) {
-          FormatOperator(node, "-", strBuilder);
+          FormatSubtraction(node, strBuilder);
         } else if (node.Symbol is Tangent) {
           FormatFunction(node, "Math.Tan", strBuilder);
@@ -145 +145 @@
     }
 
+    private void FormatSubtraction(ISymbolicExpressionTreeNode node, StringBuilder strBuilder) {
+      if (node.SubtreeCount == 1) {
+        strBuilder.Append("-");
+        FormatRecursively(node.GetSubtree(0), strBuilder);
+        return;
+      }
+      //Default case: more than 1 child
+      FormatOperator(node, "-", strBuilder);
+    }
+
     private void FormatOperator(ISymbolicExpressionTreeNode node, string symbol, StringBuilder strBuilder) {
       strBuilder.Append("(");
@@ -179 +189 @@
       }
 
-      var orderedNames = varNames.OrderBy(n => n, new NaturalStringComparer()).Select(n=> "double " + n);
+      var orderedNames = varNames.OrderBy(n => n, new NaturalStringComparer()).Select(n => "double " + n);
       strBuilder.Append(string.Join(", ", orderedNames));
 

branches/HeuristicLab.LinqExpressionTreeInterpreter/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/HeuristicLab.Problems.DataAnalysis.Symbolic-3.4.csproj

@@ -204 +204 @@
     <Compile Include="Interfaces\IModelBacktransformator.cs" />
     <Compile Include="Interpreter\SymbolicDataAnalysisExpressionCompiledTreeInterpreter.cs" />
+    <Compile Include="SymbolicDataAnalysisExpressionTreeSimplificationOperator.cs" />
+    <Compile Include="SymbolicDataAnalysisModelComplexityCalculator.cs" />
     <Compile Include="SymbolicExpressionTreeBacktransformator.cs" />
     <Compile Include="SymbolicDataAnalysisExpressionPruningOperator.cs" />