Changeset 15957 for branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/GrammarEnumerationAlgorithm.cs

Timestamp:

06/13/18 14:41:46 (6 years ago)

Author:

bburlacu

Message:

#2886: Minor refactor; fix multiple analyzer event registration

File:

• branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/GrammarEnumerationAlgorithm.cs (modified) (8 diffs)

branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/GrammarEnumerationAlgorithm.cs

@@ -4 +4 @@
 using System.Threading;
 using HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration.GrammarEnumeration;
+using HeuristicLab.Collections;
 using HeuristicLab.Common;
 using HeuristicLab.Core;
@@ -128 +129 @@
 
     public GrammarEnumerationAlgorithm() {
-      Problem = new RegressionProblem() {
-        ProblemData = new HeuristicLab.Problems.Instances.DataAnalysis.PolyTen(seed: 1234).GenerateRegressionData()
-      };
-
       Parameters.Add(new ValueParameter<DoubleValue>(VariableImportanceWeightName, "Variable Weight.", new DoubleValue(1.0)));
       Parameters.Add(new ValueParameter<BoolValue>(OptimizeConstantsParameterName, "Run constant optimization in sentence evaluation.", new BoolValue(false)));
@@ -148 +145 @@
         new CheckedItemCollection<IGrammarEnumerationAnalyzer>(availableAnalyzers).AsReadOnly()));
 
+      Analyzers.CheckedItemsChanged += Analyzers_CheckedItemsChanged;
+
       foreach (var analyzer in Analyzers) {
         Analyzers.SetItemCheckedState(analyzer, false);
       }
       Analyzers.SetItemCheckedState(Analyzers.First(analyzer => analyzer is RSquaredEvaluator), true);
-      //Analyzers.SetItemCheckedState(Analyzers.First(analyzer => analyzer is SentenceLogger), true);
 
       var grammarSymbols = Enum.GetValues(typeof(GrammarRule))
@@ -165 +163 @@
         GrammarSymbols.SetItemCheckedState(grammarSymbol, true);
       }
+
+      // set a default problem
+      Problem = new RegressionProblem() {
+        ProblemData = new HeuristicLab.Problems.Instances.DataAnalysis.PolyTen(seed: 1234).GenerateRegressionData()
+      };
     }
 
     public GrammarEnumerationAlgorithm(GrammarEnumerationAlgorithm original, Cloner cloner) : base(original, cloner) {
+      foreach (var analyzer in Analyzers.CheckedItems)
+        analyzer.Register(this);
+      Analyzers.CheckedItemsChanged += Analyzers_CheckedItemsChanged;
     }
     #endregion
@@ -173 +179 @@
     private Dictionary<VariableTerminalSymbol, double> variableImportance;
 
-    protected override void Run(CancellationToken cancellationToken) {
-      #region init
-      InitResults();
-
-      foreach (IGrammarEnumerationAnalyzer grammarEnumerationAnalyzer in Analyzers) {
-        if (Analyzers.ItemChecked(grammarEnumerationAnalyzer)) {
-          grammarEnumerationAnalyzer.Register(this);
-        } else {
-          grammarEnumerationAnalyzer.Deregister(this);
-        }
-      }
-
-      Analyzers.OfType<RSquaredEvaluator>().First().OptimizeConstants = OptimizeConstants;
-
+    public override void Prepare() {
+      DistinctSentencesComplexity = new Dictionary<int, int>();
       ArchivedPhrases = new HashSet<int>();
-
-      DistinctSentencesComplexity = new Dictionary<int, int>();
       AllGeneratedSentencesCount = 0;
       OverwrittenSentencesCount = 0;
       PhraseExpansionCount = 0;
 
+      Analyzers.OfType<RSquaredEvaluator>().First().OptimizeConstants = OptimizeConstants;
       Grammar = new Grammar(Problem.ProblemData.AllowedInputVariables.ToArray(), GrammarSymbols.CheckedItems.Select(v => v.Value));
-
       OpenPhrases = new SearchDataStore(SearchDataStructure); // Select search strategy
+
+      base.Prepare(); // this actually clears the results which will get reinitialized on Run()
+    }
+
+    protected override void Run(CancellationToken cancellationToken) {
+      InitResults();
       var phrase0 = new SymbolString(new[] { Grammar.StartSymbol });
       var phrase0Hash = Grammar.Hasher.CalcHashCode(phrase0);
-      #endregion
 
       #region Variable Importance
@@ -249 +246 @@
           int newPhraseComplexity = Grammar.GetComplexity(newPhrase);
 
-          if (newPhraseComplexity <= MaxComplexity) {
-            var phraseHash = Grammar.Hasher.CalcHashCode(newPhrase);
-
-            OnPhraseDerived(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
-
-            if (newPhrase.IsSentence()) {
-              AllGeneratedSentencesCount++;
-
-              OnSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
-
-              // Is the best solution found? (only if RSquaredEvaluator is activated)
-              if (Results.ContainsKey(RSquaredEvaluator.BestTrainingQualityResultName)) {
-                double r2 = ((DoubleValue)Results[RSquaredEvaluator.BestTrainingQualityResultName].Value).Value;
-                if (r2.IsAlmost(1.0)) {
-                  UpdateView(force: true);
-                  return;
-                }
+          if (newPhraseComplexity > MaxComplexity)
+            continue;
+
+          var phraseHash = Grammar.Hasher.CalcHashCode(newPhrase);
+
+          OnPhraseDerived(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
+
+          if (newPhrase.IsSentence()) {
+            AllGeneratedSentencesCount++;
+
+            OnSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
+
+            // Is the best solution found? (only if RSquaredEvaluator is activated)
+            if (Results.ContainsKey(RSquaredEvaluator.BestTrainingQualityResultName)) {
+              double r2 = ((DoubleValue)Results[RSquaredEvaluator.BestTrainingQualityResultName].Value).Value;
+              if (r2.IsAlmost(1.0)) {
+                UpdateView(force: true);
+                return;
               }
-
-              if (!DistinctSentencesComplexity.ContainsKey(phraseHash) || DistinctSentencesComplexity[phraseHash] > newPhraseComplexity) {
-                if (DistinctSentencesComplexity.ContainsKey(phraseHash)) OverwrittenSentencesCount++; // for analysis only
-
-                DistinctSentencesComplexity[phraseHash] = newPhraseComplexity;
-                OnDistinctSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
-              }
-              UpdateView();
-
-            } else if (!OpenPhrases.Contains(phraseHash) && !ArchivedPhrases.Contains(phraseHash)) {
-
-              double r2 = GetR2(newPhrase, fetchedSearchNode.R2);
-              double phrasePriority = GetPriority(newPhrase, r2, maxSentenceLength, errorWeight, variableImportanceWeight);
-
-              SearchNode newSearchNode = new SearchNode(phraseHash, phrasePriority, r2, newPhrase);
-              OpenPhrases.Store(newSearchNode);
             }
+
+            if (!DistinctSentencesComplexity.ContainsKey(phraseHash) || DistinctSentencesComplexity[phraseHash] > newPhraseComplexity) {
+              if (DistinctSentencesComplexity.ContainsKey(phraseHash)) OverwrittenSentencesCount++; // for analysis only
+
+              DistinctSentencesComplexity[phraseHash] = newPhraseComplexity;
+              OnDistinctSentenceGenerated(fetchedSearchNode.Hash, fetchedSearchNode.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
+            }
+            UpdateView();
+
+          } else if (!OpenPhrases.Contains(phraseHash) && !ArchivedPhrases.Contains(phraseHash)) {
+
+            double r2 = GetR2(newPhrase, fetchedSearchNode.R2);
+            double phrasePriority = GetPriority(newPhrase, r2, maxSentenceLength, errorWeight, variableImportanceWeight);
+
+            SearchNode newSearchNode = new SearchNode(phraseHash, phrasePriority, r2, newPhrase);
+            OpenPhrases.Store(newSearchNode);
           }
         }
@@ -349 +347 @@
     // Initialize entries in result set.
     private void InitResults() {
+      Results.Clear();
       Results.Add(new Result(GeneratedPhrasesName, new IntValue(0)));
       Results.Add(new Result(SearchStructureSizeName, new IntValue(0)));
@@ -379 +378 @@
 
     #region events
+
+    // private event handlers for analyzers
+    private void Analyzers_CheckedItemsChanged(object sender, CollectionItemsChangedEventArgs<IGrammarEnumerationAnalyzer> args) {
+      // newly added items
+      foreach (var item in args.Items.Except(args.OldItems).Union(args.OldItems.Except(args.Items))) {
+        if (Analyzers.ItemChecked(item)) {
+          item.Register(this);
+        } else {
+          item.Deregister(this);
+        }
+      }
+    }
+
     public event EventHandler<PhraseEventArgs> PhraseFetched;
     private void OnPhraseFetched(int hash, SymbolString symbolString) {