source: branches/HeuristicLab.EvolutionTracking/HeuristicLab.Problems.DataAnalysis.Symbolic/3.4/Tracking/SchemaDiversification/SchemaCreator.cs @ 12988

#region License Information
/* HeuristicLab
 * Copyright (C) 2002-2015 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
 *
 * This file is part of HeuristicLab.
 *
 * HeuristicLab is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * HeuristicLab is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
 */
#endregion

using System;
using System.Collections.Generic;
using System.Linq;
using HeuristicLab.Common;
using HeuristicLab.Core;
using HeuristicLab.Data;
using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
using HeuristicLab.EvolutionTracking;
using HeuristicLab.Parameters;
using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;

namespace HeuristicLab.Problems.DataAnalysis.Symbolic {
  [Item("SchemaCreator", "An operator that builds schemas based on the heredity relationship in the genealogy graph.")]
  [StorableClass]
  public class SchemaCreator : EvolutionTrackingOperator<ISymbolicExpressionTree> {
    private const string MinimumSchemaLengthParameterName = "MinimumSchemaLength";
    private const string MinimumSchemaFrequencyParameterName = "MinimumSchemaFrequency";
    private const string MinimumPhenotypicSimilarityParameterName = "MinimumPhenotypicSimilarity";
    private const string ReplacementRatioParameterName = "ReplacementRatio";
    private const string RandomReplacementParameterName = "RandomReplacement";
    private const string ExecuteInParallelParameterName = "ExecuteInParallel";
    private const string MaxDegreeOfParalellismParameterName = "MaxDegreeOfParallelism";
    private const string PercentageOfPopulationParameterName = "PercentageOfPopulationToDiversify";
    private const string ScaleEstimatedValuesParameterName = "ScaleEstimatedValues";
    private const string ExclusiveMatchingParameterName = "ExclusiveMatching";
    private const string NumberOfChangedTreesParameterName = "NumberOfChangedTrees";
    private const string NumberOfSchemasParameterName = "NumberOfSchemas";
    private const string AverageSchemaLengthParameterName = "AverageSchemaLength";

    #region parameters
    public IFixedValueParameter<BoolValue> ExclusiveMatchingParameter {
      get { return (IFixedValueParameter<BoolValue>)Parameters[ExclusiveMatchingParameterName]; }
    }
    public IFixedValueParameter<BoolValue> ScaleEstimatedValuesParameter {
      get { return (IFixedValueParameter<BoolValue>)Parameters[ScaleEstimatedValuesParameterName]; }
    }
    public IFixedValueParameter<PercentValue> PercentageOfPopulationParameter {
      get { return (IFixedValueParameter<PercentValue>)Parameters[PercentageOfPopulationParameterName]; }
    }
    public IFixedValueParameter<IntValue> MinimumSchemaLengthParameter {
      get { return (IFixedValueParameter<IntValue>)Parameters[MinimumSchemaLengthParameterName]; }
    }
    public IFixedValueParameter<BoolValue> ExecuteInParallelParameter {
      get { return (IFixedValueParameter<BoolValue>)Parameters[ExecuteInParallelParameterName]; }
    }
    public IFixedValueParameter<IntValue> MaxDegreeOfParallelismParameter {
      get { return (IFixedValueParameter<IntValue>)Parameters[MaxDegreeOfParalellismParameterName]; }
    }
    public IFixedValueParameter<PercentValue> MinimumSchemaFrequencyParameter {
      get { return (IFixedValueParameter<PercentValue>)Parameters[MinimumSchemaFrequencyParameterName]; }
    }
    public IFixedValueParameter<PercentValue> MinimumPhenotypicSimilarityParameter {
      get { return (IFixedValueParameter<PercentValue>)Parameters[MinimumPhenotypicSimilarityParameterName]; }
    }
    public IFixedValueParameter<PercentValue> ReplacementRatioParameter {
      get { return (IFixedValueParameter<PercentValue>)Parameters[ReplacementRatioParameterName]; }
    }
    public IValueParameter<IntValue> NumberOfSchemasParameter {
      get { return (IValueParameter<IntValue>)Parameters[NumberOfSchemasParameterName]; }
    }
    public IValueParameter<DoubleValue> AverageSchemaLengthParameter {
      get { return (IValueParameter<DoubleValue>)Parameters[AverageSchemaLengthParameterName]; }
    }
    public IValueParameter<IntValue> NumberOfChangedTreesParameter {
      get { return (IValueParameter<IntValue>)Parameters[NumberOfChangedTreesParameterName]; }
    }
    #endregion

    #region parameter properties
    public int MinimumSchemaLength { get { return MinimumSchemaLengthParameter.Value.Value; } }
    public int MaxDegreeOfParallelism { get { return MaxDegreeOfParallelismParameter.Value.Value; } }
    public bool ExecuteInParallel { get { return ExecuteInParallelParameter.Value.Value; } }
    public double PercentageOfPopulation { get { return PercentageOfPopulationParameter.Value.Value; } }
    #endregion

    private UpdateEstimatedValuesOperator updateEstimatedValuesOperator;
    private DiversificationStatisticsOperator diversificationStatisticsOperator;

    public SchemaCreator() {
      Parameters.Add(new FixedValueParameter<IntValue>(MinimumSchemaLengthParameterName, new IntValue(10)));
      Parameters.Add(new FixedValueParameter<PercentValue>(MinimumSchemaFrequencyParameterName, new PercentValue(0.01)));
      Parameters.Add(new FixedValueParameter<PercentValue>(MinimumPhenotypicSimilarityParameterName, new PercentValue(0.9)));
      Parameters.Add(new FixedValueParameter<PercentValue>(ReplacementRatioParameterName, new PercentValue(0.9)));
      Parameters.Add(new FixedValueParameter<PercentValue>(PercentageOfPopulationParameterName, new PercentValue(1)));
      Parameters.Add(new FixedValueParameter<BoolValue>(RandomReplacementParameterName, new BoolValue(false)));
      Parameters.Add(new FixedValueParameter<BoolValue>(ExecuteInParallelParameterName, new BoolValue(false)));
      Parameters.Add(new FixedValueParameter<IntValue>(MaxDegreeOfParalellismParameterName, new IntValue(-1)));
      Parameters.Add(new FixedValueParameter<BoolValue>(ScaleEstimatedValuesParameterName, new BoolValue(true)));
      Parameters.Add(new FixedValueParameter<BoolValue>(ExclusiveMatchingParameterName, new BoolValue(false)));
      Parameters.Add(new ValueParameter<IntValue>(NumberOfChangedTreesParameterName, new IntValue(0)));
      Parameters.Add(new ValueParameter<IntValue>(NumberOfSchemasParameterName, new IntValue(0)));
      Parameters.Add(new ValueParameter<DoubleValue>(AverageSchemaLengthParameterName, new DoubleValue(0)));

      NumberOfChangedTreesParameter.Hidden = true;
      NumberOfSchemasParameter.Hidden = true;
      AverageSchemaLengthParameter.Hidden = true;

      ExecuteInParallelParameter.Hidden = true;
      MaxDegreeOfParallelismParameter.Hidden = true;
    }

    protected SchemaCreator(SchemaCreator original, Cloner cloner) : base(original, cloner) { }

    public override IDeepCloneable Clone(Cloner cloner) {
      return new SchemaCreator(this, cloner);
    }

    [StorableConstructor]
    protected SchemaCreator(bool deserializing) : base(deserializing) { }

    public override IOperation Apply() {
      // apply only when at least one generation has passed
      var gen = Generations.Value;
      if (gen < 1 || GenealogyGraph == null)
        return base.Apply();

      var n = (int)Math.Round(ExecutionContext.Scope.SubScopes.Count * PercentageOfPopulation);
      var scopes = new ScopeList(ExecutionContext.Scope.SubScopes.Take(n));

      // for now, only consider crossover offspring
      var vertices = from s in scopes
                     let t = (ISymbolicExpressionTree)s.Variables["SymbolicExpressionTree"].Value
                     let v = GenealogyGraph.GetByContent(t)
                     where v.InDegree == 2
                     select v;

      var groups = vertices.GroupBy(x => x.Parents.First()).OrderByDescending(g => g.Count()).ToList();
      var anySubtreeSymbol = new AnySubtreeSymbol();

      var updateEstimatedValues = new OperationCollection { Parallel = true };
      if (updateEstimatedValuesOperator == null)
        updateEstimatedValuesOperator = new UpdateEstimatedValuesOperator();

      foreach (var s in scopes) {
        if (!s.Variables.ContainsKey("EstimatedValues"))
          s.Variables.Add(new Core.Variable("EstimatedValues"));
        updateEstimatedValues.Add(ExecutionContext.CreateChildOperation(updateEstimatedValuesOperator, s));
      }

      var evaluateSchemas = new OperationCollection();
      var schemas = new List<ISymbolicExpressionTree>();
      var wildCardCounts = new List<double>();
      #region create schemas and add subscopes representing the individuals
      foreach (var g in groups) {
        var parent = g.Key;
        if (parent.Data.Length < MinimumSchemaLength)
          continue;
        bool replaced = false;
        var schema = (ISymbolicExpressionTree)parent.Data.Clone();
        var nodes = schema.IterateNodesPrefix().ToList();
        var arcs = g.Select(x => x.InArcs.Last()).Where(x => x.Data != null);
        var indices = arcs.Select(x => ((IFragment<ISymbolicExpressionTreeNode>)x.Data).Index1).Distinct().ToArray();
        Array.Sort(indices);
        var nodesToReplace = indices.Select(x => nodes[x]).ToList();
        int w = 0;
        for (int i = nodesToReplace.Count - 1; i >= 0; --i) {
          var node = nodesToReplace[i];

          // do not replace the node with a wildcard if it would result in a length < MinimumSchemaLength
          if ((schema.Length - node.GetLength() + 1) < MinimumSchemaLength)
            continue;

          var replacement = anySubtreeSymbol.CreateTreeNode();
          ReplaceSubtree(node, replacement, false);
          replaced = true;
          ++w;
        }
        if (replaced) {
          // store the schemas and the number of wildcards they contain in two lists
          schemas.Add(schema);
          wildCardCounts.Add(w);
        }
      }

      for (int i = 0; i < schemas.Count; ++i) {
        var schema = schemas[i];
        evaluateSchemas.Add(ExecutionContext.CreateChildOperation(new SchemaEvaluator { Schema = schema }, ExecutionContext.Scope));
      }
      #endregion

      if (diversificationStatisticsOperator == null)
        diversificationStatisticsOperator = new DiversificationStatisticsOperator();

      var calculateStatistics = ExecutionContext.CreateChildOperation(diversificationStatisticsOperator);

      // set parameters for statistics
      AverageSchemaLengthParameter.Value = new DoubleValue(schemas.Average(x => x.Length));
      NumberOfSchemasParameter.Value = new IntValue(schemas.Count);
      NumberOfChangedTreesParameter.Value = new IntValue(0);

      // return an operation collection containing all the scope operations + base.Apply()
      return new OperationCollection { updateEstimatedValues, evaluateSchemas, calculateStatistics, base.Apply() };
    }

    private static void ReplaceSubtree(ISymbolicExpressionTreeNode original, ISymbolicExpressionTreeNode replacement,
      bool preserveChildren = true) {
      var parent = original.Parent;
      if (parent == null)
        throw new ArgumentException("Parent cannot be null for node " + original);
      var index = parent.IndexOfSubtree(original);
      parent.RemoveSubtree(index);
      parent.InsertSubtree(index, replacement);

      if (preserveChildren) {
        var subtrees = original.Subtrees.ToList();

        for (int i = subtrees.Count - 1; i >= 0; --i)
          original.RemoveSubtree(i);

        for (int i = 0; i < subtrees.Count; ++i) {
          replacement.AddSubtree(subtrees[i]);
        }
      }
    }
  }
}