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             : #include "ActiveLearningMonteCarloSampler.h"
      11             : #include "Distribution.h"
      12             : 
      13             : registerMooseObject("StochasticToolsApp", ActiveLearningMonteCarloSampler);
      14             : 
      15             : InputParameters
      16         234 : ActiveLearningMonteCarloSampler::validParams()
      17             : {
      18         234 :   InputParameters params = Sampler::validParams();
      19         234 :   params.addClassDescription("Monte Carlo Sampler for active learning with surrogate model.");
      20         468 :   params.addRequiredParam<dof_id_type>("num_batch",
      21             :                                        "The number of full model evaluations in the batch.");
      22         468 :   params.addRequiredParam<std::vector<DistributionName>>(
      23             :       "distributions",
      24             :       "The distribution names to be sampled, the number of distributions provided defines the "
      25             :       "number of columns per matrix.");
      26         468 :   params.addRequiredParam<ReporterName>("flag_sample",
      27             :                                         "Flag samples if the surrogate prediction was inadequate.");
      28         468 :   params.addParam<unsigned int>(
      29             :       "num_random_seeds",
      30         468 :       100000,
      31             :       "Initialize a certain number of random seeds. Change from the default only if you have to.");
      32         468 :   params.addRequiredRangeCheckedParam<int>(
      33             :       "num_samples",
      34             :       "num_samples>0",
      35             :       "Number of samples to use (the total number of steps taken will be equal to this number + "
      36             :       "the number of re-training steps).");
      37         234 :   return params;
      38           0 : }
      39             : 
      40         136 : ActiveLearningMonteCarloSampler::ActiveLearningMonteCarloSampler(const InputParameters & parameters)
      41             :   : Sampler(parameters),
      42         136 :     _flag_sample(getReporterValue<std::vector<bool>>("flag_sample")),
      43         272 :     _step(getCheckedPointerParam<FEProblemBase *>("_fe_problem_base")->timeStep()),
      44         272 :     _num_batch(getParam<dof_id_type>("num_batch")),
      45         136 :     _check_step(std::numeric_limits<int>::min()),
      46         544 :     _num_samples(getParam<int>("num_samples"))
      47             : {
      48         680 :   for (const DistributionName & name : getParam<std::vector<DistributionName>>("distributions"))
      49         408 :     _distributions.push_back(&getDistributionByName(name));
      50         136 :   setNumberOfRows(_num_batch);
      51         136 :   setNumberOfCols(_distributions.size());
      52         136 :   _inputs_sto.resize(_num_batch, std::vector<Real>(_distributions.size()));
      53         272 :   setNumberOfRandomSeeds(getParam<unsigned int>("num_random_seeds"));
      54         136 : }
      55             : 
      56             : void
      57        5480 : ActiveLearningMonteCarloSampler::sampleSetUp(const Sampler::SampleMode /*mode*/)
      58             : {
      59             :   // If we've already done this step, skip
      60        5480 :   if (_check_step == _step)
      61             :     return;
      62             : 
      63        1640 :   if (_is_sampling_completed)
      64           0 :     mooseError("Internal bug: the adaptive sampling is supposed to be completed but another sample "
      65             :                "has been requested.");
      66             : 
      67             :   // Keep data where the GP failed
      68        1640 :   if (_step > 0)
      69        4160 :     for (dof_id_type i = 0; i < _num_batch; ++i)
      70        2520 :       if (_flag_sample[i])
      71             :       {
      72         340 :         _inputs_gp_fails.push_back(_inputs_sto[i]);
      73             : 
      74             :         // When the GP fails, the current time step is 'wasted' and the retraining step doesn't
      75             :         // happen until the next time step. Therefore, keep track of the number of retraining steps
      76             :         // to increase the total number of steps taken.
      77         340 :         ++_retraining_steps;
      78             :       }
      79             : 
      80             :   // If we don't have enough failed inputs, generate new ones
      81        1640 :   if (_inputs_gp_fails.size() < _num_batch)
      82             :   {
      83        3680 :     for (dof_id_type i = 0; i < _num_batch; ++i)
      84        8880 :       for (dof_id_type j = 0; j < _distributions.size(); ++j)
      85        6660 :         _inputs_sto[i][j] = _distributions[j]->quantile(getRand(_step));
      86             :   }
      87             :   // If we do have enough failed inputs, assign them and clear the tracked ones
      88             :   else
      89             :   {
      90         180 :     _inputs_sto.assign(_inputs_gp_fails.begin(), _inputs_gp_fails.begin() + _num_batch);
      91         180 :     _inputs_gp_fails.erase(_inputs_gp_fails.begin(), _inputs_gp_fails.begin() + _num_batch);
      92             :   }
      93             : 
      94        1640 :   _check_step = _step;
      95             : 
      96             :   // check if we have finished the sampling
      97        1640 :   if (_step >= _num_samples + _retraining_steps)
      98          80 :     _is_sampling_completed = true;
      99             : }
     100             : 
     101             : Real
     102       16440 : ActiveLearningMonteCarloSampler::computeSample(dof_id_type row_index, dof_id_type col_index)
     103             : {
     104       16440 :   return _inputs_sto[row_index][col_index];
     105             : }