Changeset 15906 for branches/HeuristicLab.EvolutionTracking/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Tracking/Analyzers/SymbolicDataAnalysisSchemaFrequencyAnalyzer.cs

Timestamp:

04/16/18 08:35:59 (6 years ago)

Author:

bburlacu

Message:

#1772: Refactoring and speed optimization of diversification operators

File:

• branches/HeuristicLab.EvolutionTracking/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Tracking/Analyzers/SymbolicDataAnalysisSchemaFrequencyAnalyzer.cs (modified) (5 diffs)

branches/HeuristicLab.EvolutionTracking/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Tracking/Analyzers/SymbolicDataAnalysisSchemaFrequencyAnalyzer.cs

@@ -43 +43 @@
     private const string ExecuteInParallelParameterName = "ExecuteInParallel";
     private const string MaximumDegreeOfParallelismParameterName = "MaximumDegreeOfParallelism";
+    private const string SchemaDefinitionParameterName = "SchemaDefinition";
 
     private static readonly Dictionary<string, string> ShortNames = new Dictionary<string, string> {
-    { "Addition", "+" },
-    { "Subtraction", "-" },
-    { "Multiplication", "*" },
-    { "Division", "/" },
-    { "Exponential", "exp" },
-    { "Logarithm", "log" }
-  };
+      { "Addition", "+" },
+      { "Subtraction", "-" },
+      { "Multiplication", "*" },
+      { "Division", "/" },
+      { "Exponential", "exp" },
+      { "Logarithm", "log" }
+    };
 
     [Storable]
@@ -63 +64 @@
     private QueryMatch qm;
 
+
+    public IConstrainedValueParameter<StringValue> SchemaDefinitionParameter {
+      get { return (IConstrainedValueParameter<StringValue>)Parameters[SchemaDefinitionParameterName]; }
+    }
     public IFixedValueParameter<BoolValue> ExecuteInParallelParameter {
       get { return (IFixedValueParameter<BoolValue>)Parameters[ExecuteInParallelParameterName]; }
@@ -122 +127 @@
       Parameters.Add(new LookupParameter<IRegressionProblemData>(ProblemDataParameterName));
       Parameters.Add(new LookupParameter<ISymbolicDataAnalysisExpressionTreeInterpreter>(InterpreterParameterName));
+
+      var schemaDefinitions = new ItemSet<StringValue>(new[] {
+         new StringValue("="),
+         new StringValue("#"),
+         new StringValue("=,#")
+      });
+      var schemaDefinitionParameter = new ConstrainedValueParameter<StringValue>(SchemaDefinitionParameterName, schemaDefinitions);
+      schemaDefinitionParameter.Value = schemaDefinitions.First();
+      Parameters.Add(schemaDefinitionParameter);
     }
 
@@ -185 +199 @@
       }
 
-      var schemas = SchemaCreator.GenerateCombinedSchemas(vertices, MinimumSchemaLength, 0, StrictSchemaMatching).ToList();
+      List<ISymbolicExpressionTree> schemas;
+      var def = SchemaDefinitionParameter.Value.Value;
+      switch (def) {
+        case "=":
+          schemas = SchemaCreator.GenerateAnyNodeSchemas(vertices, MinimumSchemaLength, 0, StrictSchemaMatching).ToList();
+          break;
+        case "#":
+          schemas = SchemaCreator.GenerateAnySubtreeSchemas(vertices, MinimumSchemaLength, 0, StrictSchemaMatching).ToList();
+          break;
+        case "=,#":
+          schemas = SchemaCreator.GenerateCombinedSchemas(vertices, MinimumSchemaLength, 0, StrictSchemaMatching).ToList();
+          break;
+        default:
+          return base.Apply();
+      }
       var schemaStrings = schemas.Select(x => x.Root.GetSubtree(0).GetSubtree(0).FormatToString(StrictSchemaMatching)).ToList();
-      int[][] matchingIndices;
+      int[][] matchingIndices = new int[schemas.Count][];
+
+      var tNodes = trees.Select(x => QueryMatch.InitializePostOrder(x.Root.GetSubtree(0).GetSubtree(0))).ToList();
       if (ExecuteInParallel) {
-        matchingIndices = new int[schemas.Count][];
         Parallel.For(0, schemas.Count, new ParallelOptions { MaxDegreeOfParallelism = MaximumDegreeOfParallelism }, i => {
           var schema = schemas[i];
-          matchingIndices[i] = Enumerable.Range(0, trees.Count).Where(v => qm.Match(trees[v], schema)).ToArray();
+          var sNodes = QueryMatch.InitializePostOrder(schema.Root.GetSubtree(0).GetSubtree(0));
+          matchingIndices[i] = Enumerable.Range(0, trees.Count).Where(idx => qm.Match(tNodes[idx], sNodes)).ToArray();
         });
       } else {
-        matchingIndices = schemas.Select(x => Enumerable.Range(0, trees.Count).Where(v => qm.Match(trees[v], x)).ToArray()).ToArray();
+        for (int i = 0; i < schemas.Count; ++i) {
+          var schema = schemas[i];
+          var sNodes = QueryMatch.InitializePostOrder(schema.Root.GetSubtree(0).GetSubtree(0));
+          matchingIndices[i] = Enumerable.Range(0, trees.Count).Where(idx => qm.Match(tNodes[idx], sNodes)).ToArray();
+        }
       }
 
@@ -203 +237 @@
       if (ExecuteInParallel) {
         var locker = new object();
-        Parallel.For(0, schemas.Count, new ParallelOptions { MaxDegreeOfParallelism = MaximumDegreeOfParallelism }, i => {
+        Parallel.For(0, schemas.Count, new ParallelOptions {
+          MaxDegreeOfParallelism = MaximumDegreeOfParallelism
+        }, i => {
           var indices = matchingIndices[i];
           if (indices.Length > 1) {