#region License Information
/* HeuristicLab
 * Copyright (C) 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.Threading;
using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
using HeuristicLab.MainForm;
using HeuristicLab.Problems.DataAnalysis.Symbolic.Views;

namespace HeuristicLab.Problems.DataAnalysis.Symbolic.Regression.Views {
  public partial class InteractiveSymbolicRegressionSolutionSimplifierView : InteractiveSymbolicDataAnalysisSolutionSimplifierView {
    public new SymbolicRegressionSolution Content {
      get { return (SymbolicRegressionSolution)base.Content; }
      set { base.Content = value; }
    }

    public InteractiveSymbolicRegressionSolutionSimplifierView()
      : base(new SymbolicRegressionSolutionImpactValuesCalculator()) {
      InitializeComponent();
      this.Caption = "Interactive Regression Solution Simplifier";
    }

    protected override void SetEnabledStateOfControls() {
      base.SetEnabledStateOfControls();

      var tree = Content?.Model?.SymbolicExpressionTree;
      btnOptimizeConstants.Enabled = tree != null && NonlinearLeastSquaresConstantOptimizationEvaluator.CanOptimizeConstants(tree);
      btnVectorOptimizeConstants.Enabled = tree != null && TensorFlowConstantOptimizationEvaluator.CanOptimizeConstants(tree);
      nudLearningRate.Enabled = tree != null && TensorFlowConstantOptimizationEvaluator.CanOptimizeConstants(tree);
      btnUnrollingVectorOptimizeConstants.Enabled = tree != null && VectorUnrollingNonlinearLeastSquaresConstantOptimizationEvaluator.CanOptimizeConstants(tree);
#if INCLUDE_DIFFSHARP
      btnDiffSharpOptimizeConstants.Enabled = tree != null && NonlinearLeastSquaresVectorConstantOptimizationEvaluator.CanOptimizeConstants(tree);
#endif
    }

    protected override void UpdateModel(ISymbolicExpressionTree tree) {
      var model = new SymbolicRegressionModel(Content.ProblemData.TargetVariable, tree, Content.Model.Interpreter, Content.Model.LowerEstimationLimit, Content.Model.UpperEstimationLimit);
      model.Scale(Content.ProblemData);
      Content.Model = model;
    }

    protected override ISymbolicExpressionTree OptimizeConstants(ISymbolicExpressionTree tree, CancellationToken cancellationToken, IProgress progress) {
      const int constOptIterations = 50;
      const int maxRepetitions = 100;
      const double minimumImprovement = 1e-10;
      var regressionProblemData = Content.ProblemData;
      var model = Content.Model;
      progress.CanBeStopped = true;
      double prevResult = 0.0, improvement = 0.0;
      var result = 0.0;
      int reps = 0;

      do {
        prevResult = result;
        tree = NonlinearLeastSquaresConstantOptimizationEvaluator.OptimizeTree(tree, regressionProblemData, regressionProblemData.TrainingIndices,
          applyLinearScaling: true, maxIterations: constOptIterations, updateVariableWeights: true,
          cancellationToken: cancellationToken, iterationCallback: (args, func, obj) => {
            double newProgressValue = progress.ProgressValue + (1.0 / (constOptIterations + 2) / maxRepetitions); // (constOptIterations + 2) iterations are reported
            progress.ProgressValue = Math.Min(newProgressValue, 1.0);
            progress.Message = $"MSE: { func / regressionProblemData.TrainingPartition.Size }";
          });
        result = SymbolicRegressionSingleObjectivePearsonRSquaredEvaluator.Calculate(model.Interpreter, tree,
          model.LowerEstimationLimit, model.UpperEstimationLimit, regressionProblemData, regressionProblemData.TrainingIndices, applyLinearScaling: true);
        reps++;
        improvement = result - prevResult;
      } while (improvement > minimumImprovement && reps < maxRepetitions &&
               progress.ProgressState != ProgressState.StopRequested &&
               progress.ProgressState != ProgressState.CancelRequested);
      return tree;
    }

    protected override ISymbolicExpressionTree VectorOptimizeConstants(ISymbolicExpressionTree tree, CancellationToken cancellationToken, IProgress progress) {
      const int maxIterations = 1000;
      var regressionProblemData = Content.ProblemData;
      progress.CanBeStopped = true;

      var learningRate = Math.Pow(10, (double)nudLearningRate.Value);

      return TensorFlowConstantOptimizationEvaluator.OptimizeTree(tree, regressionProblemData,
        regressionProblemData.TrainingIndices,
        applyLinearScaling: true, updateVariableWeights: true, maxIterations: maxIterations, learningRate: learningRate,
        cancellationToken: cancellationToken,
        progress: new SynchronousProgress<double>(cost => {
          var newProgress = progress.ProgressValue + (1.0 / (maxIterations + 1));
          progress.ProgressValue = Math.Min(newProgress, 1.0);
          progress.Message = $"MSE: {cost}";
        })
      );
    }

    protected override ISymbolicExpressionTree UnrollingVectorOptimizeConstants(ISymbolicExpressionTree tree, CancellationToken cancellationToken, IProgress progress) {
      const int constOptIterations = 50;
      const int maxRepetitions = 100;
      const double minimumImprovement = 1e-10;
      var regressionProblemData = Content.ProblemData;
      var model = Content.Model;
      progress.CanBeStopped = true;
      double prevResult = 0.0, improvement = 0.0;
      var result = 0.0;
      int reps = 0;
      var interpreter = new SymbolicDataAnalysisExpressionTreeVectorInterpreter();

      do {
        prevResult = result;
        tree = VectorUnrollingNonlinearLeastSquaresConstantOptimizationEvaluator.OptimizeTree(
          tree, interpreter,
          regressionProblemData,
          regressionProblemData.TrainingIndices,
          applyLinearScaling: true, maxIterations: constOptIterations, updateVariableWeights: true,
          cancellationToken: cancellationToken, iterationCallback: (args, func, obj) => {
            double newProgressValue = progress.ProgressValue + (1.0 / (constOptIterations + 2) / maxRepetitions); // (constOptIterations + 2) iterations are reported
            progress.ProgressValue = Math.Min(newProgressValue, 1.0);
            progress.Message = $"MSE: { func / regressionProblemData.TrainingPartition.Size }";
          });
        result = SymbolicRegressionSingleObjectivePearsonRSquaredEvaluator.Calculate(model.Interpreter, tree,
          model.LowerEstimationLimit, model.UpperEstimationLimit, regressionProblemData, regressionProblemData.TrainingIndices, applyLinearScaling: true);
        reps++;
        improvement = result - prevResult;
      } while (improvement > minimumImprovement && reps < maxRepetitions &&
               progress.ProgressState != ProgressState.StopRequested &&
               progress.ProgressState != ProgressState.CancelRequested);
      return tree;
    }


    protected override ISymbolicExpressionTree DiffSharpVectorOptimizeConstants(ISymbolicExpressionTree tree, CancellationToken cancellationToken, IProgress progress) {
      const int constOptIterations = 50;
      const int maxRepetitions = 100;
      const double minimumImprovement = 1e-10;
      var regressionProblemData = Content.ProblemData;
      var model = Content.Model;
      progress.CanBeStopped = true;
      double prevResult = 0.0, improvement = 0.0;
      var result = 0.0;
      int reps = 0;

#if INCLUDE_DIFFSHARP
      do {
        prevResult = result;
        tree = NonlinearLeastSquaresVectorConstantOptimizationEvaluator.OptimizeTree(tree, regressionProblemData, regressionProblemData.TrainingIndices,
          applyLinearScaling: true, maxIterations: constOptIterations, updateVariableWeights: true,
          cancellationToken: cancellationToken, iterationCallback: (args, func, obj) => {
            double newProgressValue = progress.ProgressValue + (1.0 / (constOptIterations + 2) / maxRepetitions); // (constOptIterations + 2) iterations are reported
            progress.ProgressValue = Math.Min(newProgressValue, 1.0);
            progress.Message = $"MSE: { func / regressionProblemData.TrainingPartition.Size }";
          });
        result = SymbolicRegressionSingleObjectivePearsonRSquaredEvaluator.Calculate(model.Interpreter, tree,
          model.LowerEstimationLimit, model.UpperEstimationLimit, regressionProblemData, regressionProblemData.TrainingIndices, applyLinearScaling: true);
        reps++;
        improvement = result - prevResult;
      } while (improvement > minimumImprovement && reps < maxRepetitions &&
               progress.ProgressState != ProgressState.StopRequested &&
               progress.ProgressState != ProgressState.CancelRequested);
#endif
      return tree;
    }


    internal class SynchronousProgress<T> : IProgress<T> {
      private readonly Action<T> callback;
      public SynchronousProgress(Action<T> callback) {
        this.callback = callback;
      }
      public void Report(T value) {
        callback(value);
      }
    }
  }
}