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 "Sampler.h"
      13             : #include "TransientInterface.h"
      14             : 
      15             : /**
      16             :  * A class used to perform Adaptive Importance Sampling using a Markov Chain Monte Carlo algorithm
      17             :  */
      18             : class AdaptiveImportanceSampler : public Sampler, public TransientInterface
      19             : {
      20             : public:
      21             :   static InputParameters validParams();
      22             : 
      23             :   AdaptiveImportanceSampler(const InputParameters & parameters);
      24             : 
      25             :   // Access the initial values vector
      26          33 :   const std::vector<Real> & getInitialValues() const { return _initial_values; }
      27             : 
      28             :   // Access the number of training samples
      29         985 :   const int & getNumSamplesTrain() const { return _num_samples_train; }
      30             : 
      31             :   // Access use absolute value bool
      32         580 :   const bool & getUseAbsoluteValue() const { return _use_absolute_value; }
      33             : 
      34             :   // Access the output limit
      35         189 :   const Real & getOutputLimit() const { return _output_limit; }
      36             : 
      37             :   // Access the mean vector of the importance distribution
      38         175 :   const std::vector<Real> & getImportanceVectorMean() const { return _mean_sto; }
      39             : 
      40             :   // Access the std vector of the importance distribution
      41         175 :   const std::vector<Real> & getImportanceVectorStd() const { return _std_sto; }
      42             : 
      43             :   // Access the std vector of the importance distribution
      44          14 :   const std::vector<const Distribution *> & getDistributionNames() const { return _distributions; }
      45             : 
      46             :   // Access the output limit
      47          14 :   const Real & getStdFactor() const { return _std_factor; }
      48             : 
      49             :   /**
      50             :    * Returns true if the adaptive sampling is completed
      51             :    */
      52         567 :   virtual bool isAdaptiveSamplingCompleted() const override { return _is_sampling_completed; }
      53             : 
      54             : protected:
      55             :   virtual void executeSetUp() override;
      56             :   /// Return the sample for the given row (the sample index) and column (the parameter index)
      57             :   virtual Real computeSample(dof_id_type row_index, dof_id_type col_index) override;
      58             : 
      59             :   /// Storage for distribution objects to be utilized
      60             :   std::vector<const Distribution *> _distributions;
      61             : 
      62             :   /// The proposal distribution standard deviations
      63             :   const std::vector<Real> & _proposal_std;
      64             : 
      65             :   /// Initial values values vector to start the importance sampler
      66             :   const std::vector<Real> & _initial_values;
      67             : 
      68             :   /// The output limit, exceedance of which indicates failure
      69             :   const Real & _output_limit;
      70             : 
      71             :   /// Number of samples to train the importance sampler
      72             :   const int & _num_samples_train;
      73             : 
      74             :   /// Number of importance sampling steps (after the importance distribution has been trained)
      75             :   const int & _num_importance_sampling_steps;
      76             : 
      77             :   /// Factor to be multiplied to the standard deviation of the proposal distribution
      78             :   const Real & _std_factor;
      79             : 
      80             :   /// Absolute value of the model result. Use this when failure is defined as a non-exceedance rather than an exceedance.
      81             :   const bool & _use_absolute_value;
      82             : 
      83             :   /// Initialize a certain number of random seeds. Change from the default only if you have to.
      84             :   const unsigned int & _num_random_seeds;
      85             : 
      86             :   /// True if the sampling is completed
      87             :   bool _is_sampling_completed;
      88             : 
      89             : private:
      90             :   /// Storage for the inputs vector obtained from the reporter
      91             :   const std::vector<std::vector<Real>> & _inputs;
      92             : 
      93             :   /// For proposing the next sample in the MCMC algorithm
      94             :   std::vector<Real> _prev_value;
      95             : 
      96             :   /// Storage for means of input values for proposing the next sample
      97             :   std::vector<Real> _mean_sto;
      98             : 
      99             :   /// Storage for standard deviations of input values for proposing the next sample
     100             :   std::vector<Real> _std_sto;
     101             : 
     102             :   /// Storage for previously accepted samples by the decision reporter system
     103             :   std::vector<std::vector<Real>> _inputs_sto;
     104             : 
     105             :   /// Number of retraining performed
     106             :   int _retraining_steps;
     107             : 
     108             :   /// Indicate whether GP prediction is good or bad to influence next proposed sample
     109             :   const std::vector<bool> * const _gp_flag;
     110             : };