Line data    Source code

       1             : //* This file is part of the MOOSE framework
       2             : //* https://mooseframework.inl.gov
       3             : //*
       4             : //* All rights reserved, see COPYRIGHT for full restrictions
       5             : //* https://github.com/idaholab/moose/blob/master/COPYRIGHT
       6             : //*
       7             : //* Licensed under LGPL 2.1, please see LICENSE for details
       8             : //* https://www.gnu.org/licenses/lgpl-2.1.html
       9             : 
      10             : #pragma once
      11             : 
      12             : #include "ActiveLearningGaussianProcess.h"
      13             : #include "Standardizer.h"
      14             : #include <Eigen/Dense>
      15             : 
      16             : #include "StochasticToolsApp.h"
      17             : #include "LoadSurrogateDataAction.h"
      18             : 
      19             : #include "SurrogateModelInterface.h"
      20             : #include "SurrogateTrainer.h"
      21             : #include "MooseRandom.h"
      22             : 
      23             : #include "Distribution.h"
      24             : 
      25             : #include "CovarianceFunctionBase.h"
      26             : #include "CovarianceInterface.h"
      27             : 
      28             : #include "GaussianProcess.h"
      29             : 
      30             : class ActiveLearningGaussianProcess : public SurrogateTrainerBase,
      31             :                                       public CovarianceInterface,
      32             :                                       public SurrogateModelInterface
      33             : {
      34             : public:
      35             :   static InputParameters validParams();
      36             :   ActiveLearningGaussianProcess(const InputParameters & parameters);
      37             : 
      38        1225 :   virtual void initialize() final {}
      39        1225 :   virtual void execute() final {}
      40             :   virtual void reTrain(const std::vector<std::vector<Real>> & inputs,
      41             :                        const std::vector<Real> & outputs) const final;
      42             : 
      43             :   StochasticTools::GaussianProcess & gp() { return _gp; }
      44             :   const StochasticTools::GaussianProcess & getGP() const { return _gp; }
      45             : 
      46             :   /**
      47             :    * Return the current length scales from GP training
      48             :    */
      49             :   const std::vector<Real> & getLengthScales() const;
      50             : 
      51             :   /**
      52             :    * Return the training data outputs standardizer
      53             :    */
      54             :   const StochasticTools::Standardizer & getTrainingStandardizer() const;
      55             : 
      56             :   /**
      57             :    * Return the normalized training outputs
      58             :    * @param norm_training_outs The normalized traing outputs to return
      59             :    */
      60             :   void getNormTrainingOuts(std::vector<Real> & norm_training_outs) const;
      61             : 
      62             : private:
      63             :   /// Name for the meta data associated with training
      64             :   const std::string _model_meta_data_name;
      65             : 
      66             :   /// The GP handler
      67             :   StochasticTools::GaussianProcess & _gp;
      68             : 
      69             :   /// Paramaters (x) used for training, along with statistics
      70             :   RealEigenMatrix & _training_params;
      71             : 
      72             :   /// Outputs (y) used for training, along with statistics
      73             :   RealEigenMatrix & _training_data;
      74             : 
      75             :   /// Switch for training param (x) standardization
      76             :   bool _standardize_params;
      77             : 
      78             :   /// Switch for training data(y) standardization
      79             :   bool _standardize_data;
      80             : 
      81             :   /// Struct holding parameters necessary for parameter tuning
      82             :   const StochasticTools::GaussianProcess::GPOptimizerOptions _optimization_opts;
      83             : };