Changeset 16399 for branches/2925_AutoDiffForDynamicalModels

Timestamp:

12/19/18 07:43:36 (6 years ago)

Author:

gkronber

Message:

#2925: solution class and solution view

Location:

branches/2925_AutoDiffForDynamicalModels/HeuristicLab.Problems.DynamicalSystemsModelling/3.3

Files:

• HeuristicLab.Problems.DynamicalSystemsModelling-3.3.csproj (modified) (8 diffs)
• Problem.cs (modified) (13 diffs)
• Solution.cs (added)
• SolutionView.Designer.cs (added)
• SolutionView.cs (added)
• SolutionView.resx (added)

branches/2925_AutoDiffForDynamicalModels/HeuristicLab.Problems.DynamicalSystemsModelling/3.3/HeuristicLab.Problems.DynamicalSystemsModelling-3.3.csproj

@@ -104 +104 @@
       <HintPath>..\..\bin\ALGLIB-3.7.0.dll</HintPath>
     </Reference>
+    <Reference Include="Microsoft.CSharp" />
     <Reference Include="System" />
     <Reference Include="System.Core">
@@ -109 +110 @@
     </Reference>
     <Reference Include="System.Drawing" />
+    <Reference Include="System.Windows.Forms" />
   </ItemGroup>
   <ItemGroup>
@@ -116 +118 @@
     <Compile Include="Problem.cs" />
     <Compile Include="Properties\AssemblyInfo.cs" />
+    <Compile Include="Solution.cs" />
+    <Compile Include="SolutionView.cs">
+      <SubType>UserControl</SubType>
+    </Compile>
+    <Compile Include="SolutionView.Designer.cs">
+      <DependentUpon>SolutionView.cs</DependentUpon>
+    </Compile>
     <Compile Include="Vector.cs" />
   </ItemGroup>
@@ -129 +138 @@
       <Name>HeuristicLab.Algorithms.DataAnalysis-3.4</Name>
     </ProjectReference>
+    <ProjectReference Include="..\..\HeuristicLab.Analysis.Views\3.3\HeuristicLab.Analysis.Views-3.3.csproj">
+      <Project>{76945D76-CA61-4147-9DC2-0ACDCDDF87F9}</Project>
+      <Name>HeuristicLab.Analysis.Views-3.3</Name>
+    </ProjectReference>
     <ProjectReference Include="..\..\HeuristicLab.Analysis\3.3\HeuristicLab.Analysis-3.3.csproj">
       <Project>{887425B4-4348-49ED-A457-B7D2C26DDBF9}</Project>
@@ -148 +161 @@
       <Private>False</Private>
     </ProjectReference>
+    <ProjectReference Include="..\..\HeuristicLab.Core.Views\3.3\HeuristicLab.Core.Views-3.3.csproj">
+      <Project>{E226881D-315F-423D-B419-A766FE0D8685}</Project>
+      <Name>HeuristicLab.Core.Views-3.3</Name>
+    </ProjectReference>
     <ProjectReference Include="..\..\HeuristicLab.Core\3.3\HeuristicLab.Core-3.3.csproj">
       <Project>{C36BD924-A541-4A00-AFA8-41701378DDC5}</Project>
@@ -153 +170 @@
       <Private>False</Private>
     </ProjectReference>
+    <ProjectReference Include="..\..\HeuristicLab.Data.Views\3.3\HeuristicLab.Data.Views-3.3.csproj">
+      <Project>{72104A0B-90E7-42F3-9ABE-9BBBADD4B943}</Project>
+      <Name>HeuristicLab.Data.Views-3.3</Name>
+    </ProjectReference>
     <ProjectReference Include="..\..\HeuristicLab.Data\3.3\HeuristicLab.Data-3.3.csproj">
       <Project>{BBAB9DF5-5EF3-4BA8-ADE9-B36E82114937}</Project>
@@ -162 +183 @@
       <Name>HeuristicLab.Encodings.SymbolicExpressionTreeEncoding-3.4</Name>
       <Private>False</Private>
+    </ProjectReference>
+    <ProjectReference Include="..\..\HeuristicLab.MainForm.WindowsForms\3.3\HeuristicLab.MainForm.WindowsForms-3.3.csproj">
+      <Project>{AB687BBE-1BFE-476B-906D-44237135431D}</Project>
+      <Name>HeuristicLab.MainForm.WindowsForms-3.3</Name>
+    </ProjectReference>
+    <ProjectReference Include="..\..\HeuristicLab.MainForm\3.3\HeuristicLab.MainForm-3.3.csproj">
+      <Project>{3BD61258-31DA-4B09-89C0-4F71FEF5F05A}</Project>
+      <Name>HeuristicLab.MainForm-3.3</Name>
     </ProjectReference>
     <ProjectReference Include="..\..\HeuristicLab.Operators\3.3\HeuristicLab.Operators-3.3.csproj">
@@ -235 +264 @@
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
     </Content>
+  </ItemGroup>
+  <ItemGroup>
+    <EmbeddedResource Include="SolutionView.resx">
+      <DependentUpon>SolutionView.cs</DependentUpon>
+    </EmbeddedResource>
   </ItemGroup>
   <Import Project="$(MSBuildToolsPath)\Microsoft.CSharp.targets" />

branches/2925_AutoDiffForDynamicalModels/HeuristicLab.Problems.DynamicalSystemsModelling/3.3/Problem.cs

@@ -198 +198 @@
     public override double Evaluate(Individual individual, IRandom random) {
       var trees = individual.Values.Select(v => v.Value).OfType<ISymbolicExpressionTree>().ToArray(); // extract all trees from individual
-      // write back optimized parameters to tree nodes instead of the separate OptTheta variable
-      // retreive optimized parameters from nodes?
-
+                                                                                                      // write back optimized parameters to tree nodes instead of the separate OptTheta variable
+                                                                                                      // retreive optimized parameters from nodes?
+
+      var problemData = ProblemData;
+      var targetVars = TargetVariables.CheckedItems.OrderBy(i => i.Index).Select(i => i.Value.Value).ToArray();
+      var latentVariables = Enumerable.Range(1, NumberOfLatentVariables).Select(i => "λ" + i).ToArray(); // TODO: must coincide with the variables which are actually defined in the grammar and also for which we actually have trees
       if (OptimizeParametersForEpisodes) {
         int eIdx = 0;
@@ -208 +211 @@
           double[] optTheta;
           double nmse;
-          OptimizeForEpisodes(trees, random, new[] { episode }, out optTheta, out nmse);
+          OptimizeForEpisodes(trees, problemData, targetVars, latentVariables, random, new[] { episode }, MaximumParameterOptimizationIterations, NumericIntegrationSteps, OdeSolver, out optTheta, out nmse);
           individual["OptTheta_" + eIdx] = new DoubleArray(optTheta); // write back optimized parameters so that we can use them in the Analysis method
           eIdx++;
@@ -218 +221 @@
         double[] optTheta;
         double nmse;
-        OptimizeForEpisodes(trees, random, TrainingEpisodes, out optTheta, out nmse);
+        OptimizeForEpisodes(trees, problemData, targetVars, latentVariables, random, TrainingEpisodes, MaximumParameterOptimizationIterations, NumericIntegrationSteps, OdeSolver, out optTheta, out nmse);
         individual["OptTheta"] = new DoubleArray(optTheta); // write back optimized parameters so that we can use them in the Analysis method
         return nmse;
@@ -224 +227 @@
     }
 
-    private void OptimizeForEpisodes(
+    public static void OptimizeForEpisodes(
       ISymbolicExpressionTree[] trees,
+      IRegressionProblemData problemData,
+      string[] targetVars,
+      string[] latentVariables,
       IRandom random,
       IEnumerable<IntRange> episodes,
+      int maxParameterOptIterations,
+      int numericIntegrationSteps,
+      string odeSolver,
       out double[] optTheta,
       out double nmse) {
       var rows = episodes.SelectMany(e => Enumerable.Range(e.Start, e.End - e.Start)).ToArray();
-      var problemData = ProblemData;
-      var targetVars = TargetVariables.CheckedItems.OrderBy(i => i.Index).Select(i => i.Value.Value).ToArray();
-      var latentVariables = Enumerable.Range(1, NumberOfLatentVariables).Select(i => "λ" + i).ToArray(); // TODO: must coincide with the variables which are actually defined in the grammar and also for which we actually have trees
       var targetValues = new double[rows.Length, targetVars.Length];
 
@@ -267 +273 @@
         alglib.minlbfgsreport report;
         alglib.minlbfgscreate(Math.Min(theta.Length, 5), theta, out state);
-        alglib.minlbfgssetcond(state, 0.0, 0.0, 0.0, MaximumParameterOptimizationIterations);
+        alglib.minlbfgssetcond(state, 0.0, 0.0, 0.0, maxParameterOptIterations);
         //alglib.minlbfgssetgradientcheck(state, 1e-6);
         alglib.minlbfgsoptimize(state, EvaluateObjectiveAndGradient, null,
-          new object[] { trees, targetVars, problemData, targetValues, episodes.ToArray(), NumericIntegrationSteps, latentVariables, OdeSolver }); //TODO: create a type
+          new object[] { trees, targetVars, problemData, targetValues, episodes.ToArray(), numericIntegrationSteps, latentVariables, odeSolver }); //TODO: create a type
 
         alglib.minlbfgsresults(state, out optTheta, out report);
@@ -307 +313 @@
       nmse = double.NaN;
       EvaluateObjectiveAndGradient(optTheta, ref nmse, grad,
-        new object[] { trees, targetVars, problemData, targetValues, episodes.ToArray(), NumericIntegrationSteps, latentVariables, OdeSolver });
+        new object[] { trees, targetVars, problemData, targetValues, episodes.ToArray(), numericIntegrationSteps, latentVariables, odeSolver });
       if (double.IsNaN(nmse) || double.IsInfinity(nmse)) { nmse = 10E6; return; } // return a large value (TODO: be consistent by using NMSE)
     }
@@ -385 +391 @@
         results.Add(new Result("Models", typeof(VariableCollection)));
       }
-      if(!results.ContainsKey("SNMSE")) {
+      if (!results.ContainsKey("SNMSE")) {
         results.Add(new Result("SNMSE", typeof(DoubleValue)));
+      }
+      if (!results.ContainsKey("Solution")) {
+        results.Add(new Result("Solution", typeof(Solution)));
       }
 
@@ -517 +526 @@
         results["Prediction (training)"].Value = trainingList.AsReadOnly();
         results["Prediction (test)"].Value = testList.AsReadOnly();
+
+
         #region simplification of models
         // TODO the dependency of HeuristicLab.Problems.DataAnalysis.Symbolic is not ideal
         var models = new VariableCollection();    // to store target var names and original version of tree
 
+        var optimizedTrees = new List<ISymbolicExpressionTree>();
         int nextParIdx = 0;
+        for (int idx = 0; idx < trees.Length; idx++) {
+          var tree = trees[idx];
+          optimizedTrees.Add(new SymbolicExpressionTree(FixParameters(tree.Root, optTheta.ToArray(), ref nextParIdx)));
+        }
+        var ds = problemData.Dataset;
+        var newVarNames = Enumerable.Range(0, nextParIdx).Select(i => "c_" + i).ToArray();
+        var allVarNames = ds.DoubleVariables.Concat(newVarNames);
+        var newVarValues = Enumerable.Range(0, nextParIdx).Select(i => "c_" + i).ToArray();
+        var allVarValues = ds.DoubleVariables.Select(varName => ds.GetDoubleValues(varName).ToList())
+          .Concat(Enumerable.Range(0, nextParIdx).Select(i => Enumerable.Repeat(optTheta[i], ds.Rows).ToList()))
+          .ToList();
+        var newDs = new Dataset(allVarNames, allVarValues);
+        var newProblemData = new RegressionProblemData(newDs, problemData.AllowedInputVariables.Concat(newVarValues).ToArray(), problemData.TargetVariable);
+        results["Solution"].Value = new Solution(optimizedTrees.ToArray(),
+                   // optTheta,
+                   newProblemData,
+                   targetVars,
+                   latentVariables,
+                   TrainingEpisodes,
+                   OdeSolver,
+                   NumericIntegrationSteps);
+
+
+        nextParIdx = 0;
         for (int idx = 0; idx < trees.Length; idx++) {
           var varName = string.Empty;
@@ -558 +594 @@
 
         }
+
         results["Models"].Value = models;
         #endregion
@@ -576 +613 @@
     // for a solver with the necessary features see: https://computation.llnl.gov/projects/sundials/cvodes
 
-    private static IEnumerable<Tuple<double, Vector>[]> Integrate(
+    public static IEnumerable<Tuple<double, Vector>[]> Integrate(
       ISymbolicExpressionTree[] trees, IDataset dataset,
       string[] inputVariables, string[] targetVariables, string[] latentVariables, IEnumerable<IntRange> episodes,
@@ -999 +1036 @@
 
 
-        foreach(var node in nodeValues.Keys.ToArray()) {
-          if(node.SubtreeCount == 0 && !IsConstantNode(node)) {
+        foreach (var node in nodeValues.Keys.ToArray()) {
+          if (node.SubtreeCount == 0 && !IsConstantNode(node)) {
             // update values for (latent) variables
             var varName = node.Symbol.Name;
@@ -1168 +1205 @@
 
     private static bool IsConstantNode(ISymbolicExpressionTreeNode n) {
-      return n.Symbol.Name.StartsWith("θ");
+      return n.Symbol.Name[0] == 'θ';
     }
     private static bool IsLatentVariableNode(ISymbolicExpressionTreeNode n) {
-      return n.Symbol.Name.StartsWith("λ");
+      return n.Symbol.Name[0] == 'λ';
     }
     private static bool IsVariableNode(ISymbolicExpressionTreeNode n) {
@@ -1232 +1269 @@
 
       return g;
+    }
+
+
+
+
+
+    private ISymbolicExpressionTreeNode FixParameters(ISymbolicExpressionTreeNode n, double[] parameterValues, ref int nextParIdx) {
+      ISymbolicExpressionTreeNode translatedNode = null;
+      if (n.Symbol is StartSymbol) {
+        translatedNode = new StartSymbol().CreateTreeNode();
+      } else if (n.Symbol is ProgramRootSymbol) {
+        translatedNode = new ProgramRootSymbol().CreateTreeNode();
+      } else if (n.Symbol.Name == "+") {
+        translatedNode = new SimpleSymbol("+", 2).CreateTreeNode();
+      } else if (n.Symbol.Name == "-") {
+        translatedNode = new SimpleSymbol("-", 2).CreateTreeNode();
+      } else if (n.Symbol.Name == "*") {
+        translatedNode = new SimpleSymbol("*", 2).CreateTreeNode();
+      } else if (n.Symbol.Name == "%") {
+        translatedNode = new SimpleSymbol("%", 2).CreateTreeNode();
+      } else if (n.Symbol.Name == "sin") {
+        translatedNode = new SimpleSymbol("sin", 1).CreateTreeNode();
+      } else if (n.Symbol.Name == "cos") {
+        translatedNode = new SimpleSymbol("cos", 1).CreateTreeNode();
+      } else if (n.Symbol.Name == "sqr") {
+        translatedNode = new SimpleSymbol("sqr", 1).CreateTreeNode();
+      } else if (IsConstantNode(n)) {
+        translatedNode = new SimpleSymbol("c_" + nextParIdx, 0).CreateTreeNode();
+        nextParIdx++;
+      } else {
+        translatedNode = new SimpleSymbol(n.Symbol.Name, n.SubtreeCount).CreateTreeNode();
+      }
+      foreach (var child in n.Subtrees) {
+        translatedNode.AddSubtree(FixParameters(child, parameterValues, ref nextParIdx));
+      }
+      return translatedNode;
     }