source: branches/2886_SymRegGrammarEnumeration/HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration/GrammarEnumeration/GrammarEnumerationAlgorithm.cs @ 15907

using System;
using System.Collections.Generic;
using System.Linq;
using System.Security.Cryptography;
using System.Threading;
using HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration.GrammarEnumeration;
using HeuristicLab.Collections;
using HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Data;
using HeuristicLab.Optimization;
using HeuristicLab.Parameters;
using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
using HeuristicLab.Problems.DataAnalysis;

namespace HeuristicLab.Algorithms.DataAnalysis.SymRegGrammarEnumeration {
  [Item("Grammar Enumeration Symbolic Regression", "Iterates all possible model structures for a fixed grammar.")]
  [StorableClass]
  [Creatable(CreatableAttribute.Categories.DataAnalysisRegression, Priority = 250)]
  public class GrammarEnumerationAlgorithm : FixedDataAnalysisAlgorithm<IRegressionProblem> {
    #region properties and result names
    private readonly string SearchStructureSizeName = "Search Structure Size";
    private readonly string GeneratedPhrasesName = "Generated/Archived Phrases";
    private readonly string GeneratedSentencesName = "Generated Sentences";
    private readonly string DistinctSentencesName = "Distinct Sentences";
    private readonly string PhraseExpansionsName = "Phrase Expansions";
    private readonly string AverageSentenceComplexityName = "Avg. Sentence Complexity among Distinct";
    private readonly string OverwrittenSentencesName = "Sentences overwritten";
    private readonly string AnalyzersParameterName = "Analyzers";
    private readonly string ExpansionsPerSecondName = "Expansions per second";


    private readonly string OptimizeConstantsParameterName = "Optimize Constants";
    private readonly string SearchDataStructureParameterName = "Search Data Structure";
    private readonly string MaxComplexityParameterName = "Max. Complexity";
    private readonly string GuiUpdateIntervalParameterName = "GUI Update Interval";

    public override bool SupportsPause { get { return false; } }

    protected IValueParameter<BoolValue> OptimizeConstantsParameter {
      get { return (IValueParameter<BoolValue>)Parameters[OptimizeConstantsParameterName]; }
    }

    public bool OptimizeConstants {
      get { return OptimizeConstantsParameter.Value.Value; }
      set { OptimizeConstantsParameter.Value.Value = value; }
    }

    protected IValueParameter<IntValue> MaxComplexityParameter {
      get { return (IValueParameter<IntValue>)Parameters[MaxComplexityParameterName]; }
    }
    public int MaxComplexity {
      get { return MaxComplexityParameter.Value.Value; }
      set { MaxComplexityParameter.Value.Value = value; }
    }

    protected IValueParameter<IntValue> GuiUpdateIntervalParameter {
      get { return (IValueParameter<IntValue>)Parameters[GuiUpdateIntervalParameterName]; }
    }
    public int GuiUpdateInterval {
      get { return GuiUpdateIntervalParameter.Value.Value; }
      set { GuiUpdateIntervalParameter.Value.Value = value; }
    }

    protected IValueParameter<EnumValue<StorageType>> SearchDataStructureParameter {
      get { return (IValueParameter<EnumValue<StorageType>>)Parameters[SearchDataStructureParameterName]; }
    }
    public StorageType SearchDataStructure {
      get { return SearchDataStructureParameter.Value.Value; }
      set { SearchDataStructureParameter.Value.Value = value; }
    }

    public IFixedValueParameter<ReadOnlyCheckedItemCollection<IGrammarEnumerationAnalyzer>> AnalyzersParameter {
      get { return (IFixedValueParameter<ReadOnlyCheckedItemCollection<IGrammarEnumerationAnalyzer>>)Parameters[AnalyzersParameterName]; }
    }

    public ICheckedItemCollection<IGrammarEnumerationAnalyzer> Analyzers {
      get { return AnalyzersParameter.Value; }
    }

    public SymbolString BestTrainingSentence { get; set; }     // Currently set in RSquaredEvaluator: quite hacky, but makes testing much easier for now...
    #endregion

    public Dictionary<int, int> DistinctSentencesComplexity { get; private set; }  // Semantically distinct sentences and their length in a run.
    public HashSet<int> ArchivedPhrases { get; private set; }
    internal SearchDataStore OpenPhrases { get; private set; }           // Stack/Queue/etc. for fetching the next node in the search tree.  

    #region execution stats
    public int AllGeneratedSentencesCount { get; private set; }

    public int OverwrittenSentencesCount { get; private set; } // It is not guaranteed that shorter solutions are found first. 
                                                               // When longer solutions are overwritten with shorter ones, 
                                                               // this counter is increased.
    public int PhraseExpansionCount { get; private set; }      // Number, how many times a nonterminal symbol is replaced with a production rule.
    #endregion

    public Grammar Grammar { get; private set; }


    #region ctors
    public override IDeepCloneable Clone(Cloner cloner) {
      return new GrammarEnumerationAlgorithm(this, cloner);
    }

    public GrammarEnumerationAlgorithm() {
      Problem = new RegressionProblem() {
        ProblemData = new HeuristicLab.Problems.Instances.DataAnalysis.NguyenFunctionNine(seed: 1234).GenerateRegressionData()
      };

      Parameters.Add(new ValueParameter<BoolValue>(OptimizeConstantsParameterName, "Run constant optimization in sentence evaluation.", new BoolValue(true)));
      Parameters.Add(new ValueParameter<IntValue>(MaxComplexityParameterName, "The maximum number of variable symbols in a sentence.", new IntValue(5)));
      Parameters.Add(new ValueParameter<IntValue>(GuiUpdateIntervalParameterName, "Number of generated sentences, until GUI is refreshed.", new IntValue(100000)));
      Parameters.Add(new ValueParameter<EnumValue<StorageType>>(SearchDataStructureParameterName, new EnumValue<StorageType>(StorageType.Stack)));

      var availableAnalyzers = new IGrammarEnumerationAnalyzer[] {
        new SearchGraphVisualizer(),
        new SentenceLogger(),
        new RSquaredEvaluator()
      };
      Parameters.Add(new FixedValueParameter<ReadOnlyCheckedItemCollection<IGrammarEnumerationAnalyzer>>(
        AnalyzersParameterName,
        new CheckedItemCollection<IGrammarEnumerationAnalyzer>(availableAnalyzers).AsReadOnly()));

      foreach (var analyzer in Analyzers) {
        Analyzers.SetItemCheckedState(analyzer, false);
      }
      Analyzers.CheckedItemsChanged += AnalyzersOnCheckedItemsChanged;
      Analyzers.SetItemCheckedState(Analyzers.First(analyzer => analyzer is RSquaredEvaluator), true);
      Analyzers.SetItemCheckedState(Analyzers.First(analyzer => analyzer is SentenceLogger), true);
    }

    public GrammarEnumerationAlgorithm(GrammarEnumerationAlgorithm original, Cloner cloner) : base(original, cloner) { }
    #endregion

    protected override void Run(CancellationToken cancellationToken) {
      #region init
      InitResults();

      Analyzers.OfType<RSquaredEvaluator>().First().OptimizeConstants = OptimizeConstants;

      ArchivedPhrases = new HashSet<int>();

      DistinctSentencesComplexity = new Dictionary<int, int>();
      AllGeneratedSentencesCount = 0;
      OverwrittenSentencesCount = 0;
      PhraseExpansionCount = 0;

      Grammar = new Grammar(Problem.ProblemData.AllowedInputVariables.ToArray());

      OpenPhrases = new SearchDataStore(SearchDataStructure); // Select search strategy
      var phrase0 = new SymbolString(new[] { Grammar.StartSymbol });
      var phrase0Hash = Grammar.Hasher.CalcHashCode(phrase0);
      #endregion

      OpenPhrases.Store(phrase0Hash, 0.0, phrase0);
      while (OpenPhrases.Count > 0) {
        if (cancellationToken.IsCancellationRequested) break;

        StoredSymbolString fetchedPhrase = OpenPhrases.GetNext();
        SymbolString currPhrase = fetchedPhrase.SymbolString;

        OnPhraseFetched(fetchedPhrase.Hash, currPhrase);

        ArchivedPhrases.Add(fetchedPhrase.Hash);

        // expand next nonterminal symbols
        int nonterminalSymbolIndex = currPhrase.NextNonterminalIndex();
        NonterminalSymbol expandedSymbol = (NonterminalSymbol)currPhrase[nonterminalSymbolIndex];
        var appliedProductions = Grammar.Productions[expandedSymbol];

        for (int i = 0; i < appliedProductions.Count; i++) {
          PhraseExpansionCount++;

          SymbolString newPhrase = currPhrase.DerivePhrase(nonterminalSymbolIndex, appliedProductions[i]);
          int newPhraseComplexity = Grammar.GetComplexity(newPhrase);

          if (newPhraseComplexity <= MaxComplexity) {
            var phraseHash = Grammar.Hasher.CalcHashCode(newPhrase);

            OnPhraseDerived(fetchedPhrase.Hash, fetchedPhrase.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);

            if (newPhrase.IsSentence()) {
              AllGeneratedSentencesCount++;

              OnSentenceGenerated(fetchedPhrase.Hash, fetchedPhrase.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(fetchedPhrase.Hash, fetchedPhrase.SymbolString, phraseHash, newPhrase, expandedSymbol, appliedProductions[i]);
              }
              UpdateView();

            } else if (!OpenPhrases.Contains(phraseHash) && !ArchivedPhrases.Contains(phraseHash)) {
              double phrasePriority = GetPriority(newPhrase);
              OpenPhrases.Store(phraseHash, phrasePriority, newPhrase);
            }
          }
        }
      }
      UpdateView(force: true);
    }

    protected double GetPriority(SymbolString phrase) {
      double complexity = (double)Grammar.GetComplexity(phrase);

      double length = phrase.Count();
      double relLength = (length - 2) / (MaxComplexity * 7);
      double r2 = Grammar.EvaluatePhrase(phrase, Problem.ProblemData, OptimizeConstants);
      double error = 1.0 - r2;

      double variables = 0;
      for (int i = 0; i < phrase.Count(); i++) {
        if (phrase[i] is VariableTerminalSymbol) variables++;
      }

      double variableRatio = 1.0 - variables / complexity;

      return 1.5*relLength + error;
    }

    #region Visualization in HL
    // Initialize entries in result set.
    private void InitResults() {
      Results.Add(new Result(GeneratedPhrasesName, new IntValue(0)));
      Results.Add(new Result(SearchStructureSizeName, new IntValue(0)));
      Results.Add(new Result(GeneratedSentencesName, new IntValue(0)));
      Results.Add(new Result(DistinctSentencesName, new IntValue(0)));
      Results.Add(new Result(PhraseExpansionsName, new IntValue(0)));
      Results.Add(new Result(OverwrittenSentencesName, new IntValue(0)));
      Results.Add(new Result(AverageSentenceComplexityName, new DoubleValue(1.0)));
      Results.Add(new Result(ExpansionsPerSecondName, "In Thousand expansions per second", new IntValue(0)));
    }

    // Update the view for intermediate results in an algorithm run.
    private int updates;
    private void UpdateView(bool force = false) {
      updates++;

      if (force || updates % GuiUpdateInterval == 1) {
        ((IntValue)Results[GeneratedPhrasesName].Value).Value = ArchivedPhrases.Count;
        ((IntValue)Results[SearchStructureSizeName].Value).Value = OpenPhrases.Count;
        ((IntValue)Results[GeneratedSentencesName].Value).Value = AllGeneratedSentencesCount;
        ((IntValue)Results[DistinctSentencesName].Value).Value = DistinctSentencesComplexity.Count;
        ((IntValue)Results[PhraseExpansionsName].Value).Value = PhraseExpansionCount;
        ((DoubleValue)Results[AverageSentenceComplexityName].Value).Value = DistinctSentencesComplexity.Select(pair => pair.Value).Average();
        ((IntValue)Results[OverwrittenSentencesName].Value).Value = OverwrittenSentencesCount;
        ((IntValue)Results[ExpansionsPerSecondName].Value).Value = (int)((PhraseExpansionCount /
                                                                          ExecutionTime.TotalSeconds) / 1000.0);
      }
    }
    #endregion

    #region events
    private void AnalyzersOnCheckedItemsChanged(object sender, CollectionItemsChangedEventArgs<IGrammarEnumerationAnalyzer> collectionItemsChangedEventArgs) {
      foreach (IGrammarEnumerationAnalyzer grammarEnumerationAnalyzer in collectionItemsChangedEventArgs.Items) {
        if (Analyzers.ItemChecked(grammarEnumerationAnalyzer)) {
          grammarEnumerationAnalyzer.Register(this);
        } else {
          grammarEnumerationAnalyzer.Deregister(this);
        }
      }
    }

    public event EventHandler<PhraseEventArgs> PhraseFetched;
    private void OnPhraseFetched(int hash, SymbolString symbolString) {
      if (PhraseFetched != null) {
        PhraseFetched(this, new PhraseEventArgs(hash, symbolString));
      }
    }

    public event EventHandler<PhraseAddedEventArgs> PhraseDerived;
    private void OnPhraseDerived(int parentHash, SymbolString parentSymbolString, int addedHash, SymbolString addedSymbolString, Symbol expandedSymbol, Production expandedProduction) {
      if (PhraseDerived != null) {
        PhraseDerived(this, new PhraseAddedEventArgs(parentHash, parentSymbolString, addedHash, addedSymbolString, expandedSymbol, expandedProduction));
      }
    }

    public event EventHandler<PhraseAddedEventArgs> SentenceGenerated;
    private void OnSentenceGenerated(int parentHash, SymbolString parentSymbolString, int addedHash, SymbolString addedSymbolString, Symbol expandedSymbol, Production expandedProduction) {
      if (SentenceGenerated != null) {
        SentenceGenerated(this, new PhraseAddedEventArgs(parentHash, parentSymbolString, addedHash, addedSymbolString, expandedSymbol, expandedProduction));
      }
    }

    public event EventHandler<PhraseAddedEventArgs> DistinctSentenceGenerated;
    private void OnDistinctSentenceGenerated(int parentHash, SymbolString parentSymbolString, int addedHash, SymbolString addedSymbolString, Symbol expandedSymbol, Production expandedProduction) {
      if (DistinctSentenceGenerated != null) {
        DistinctSentenceGenerated(this, new PhraseAddedEventArgs(parentHash, parentSymbolString, addedHash, addedSymbolString, expandedSymbol, expandedProduction));
      }
    }

    #endregion

  }

  #region events for analysis

  public class PhraseEventArgs : EventArgs {
    public int Hash { get; }

    public SymbolString Phrase { get; }

    public PhraseEventArgs(int hash, SymbolString phrase) {
      Hash = hash;
      Phrase = phrase;
    }
  }

  public class PhraseAddedEventArgs : EventArgs {
    public int ParentHash { get; }
    public int NewHash { get; }

    public SymbolString ParentPhrase { get; }
    public SymbolString NewPhrase { get; }

    public Symbol ExpandedSymbol { get; }

    public Production ExpandedProduction { get; }

    public PhraseAddedEventArgs(int parentHash, SymbolString parentPhrase, int newHash, SymbolString newPhrase, Symbol expandedSymbol, Production expandedProduction) {
      ParentHash = parentHash;
      ParentPhrase = parentPhrase;
      NewHash = newHash;
      NewPhrase = newPhrase;
      ExpandedSymbol = expandedSymbol;
      ExpandedProduction = expandedProduction;
    }
  }

  #endregion
}