Line data    Source code

       1             : //* This file is part of the MOOSE framework
       2             : //* https://www.mooseframework.org
       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 "StochasticToolsApp.h"
      13             : #include "MooseObject.h"
      14             : #include "libmesh/utility.h"
      15             : 
      16             : /**
      17             :  * All ParallelAcquisition functions should inherit from this class
      18             :  */
      19           0 : class ParallelAcquisitionFunctionBase : public MooseObject
      20             : {
      21             : public:
      22             :   static InputParameters validParams();
      23             :   ParallelAcquisitionFunctionBase(const InputParameters & parameters);
      24             : 
      25             :   /**
      26             :    * Compute the acquisition function values. Performs all argument/size checks,
      27             :    * then dispatches to computeAcquisitionInternal implemented by derived classes.
      28             :    * @param acq The computed acquisition function values
      29             :    * @param gp_mean The provided GP mean values
      30             :    * @param gp_std The provided GP standard deviation values
      31             :    * @param test_inputs All the input values under which the GP has to be tested
      32             :    * @param train_inputs All the input values under which the GP has been trained
      33             :    * @param generic A generic parameter (can be output values under which the GP has been trained or
      34             :    * threshold parameter under the U-Function etc.)
      35             :    */
      36             :   void computeAcquisition(std::vector<Real> & acq,
      37             :                           const std::vector<Real> & gp_mean,
      38             :                           const std::vector<Real> & gp_std,
      39             :                           const std::vector<std::vector<Real>> & test_inputs,
      40             :                           const std::vector<std::vector<Real>> & train_inputs,
      41             :                           const std::vector<Real> & generic) const;
      42             : 
      43             :   /**
      44             :    * Return the modified acquisition function values and sorted indices considering local
      45             :    * penalization (inspired from Zhan et al. 2017)
      46             :    * @param modified_acq The modified acquisition function values
      47             :    * @param sorted_indices The sorted indices modified acquisition function values
      48             :    * @param acq The originally acquisition function values
      49             :    * @param length_scales The length scales to compute the correlation between inputs
      50             :    * @param inputs All the input values under which acquisition needs to be computed
      51             :    * @param penalize Bool to indicate whether to compute the correlations or not
      52             :    */
      53             :   void penalizeAcquisition(std::vector<Real> & modified_acq,
      54             :                            std::vector<unsigned int> & sorted_indices,
      55             :                            const std::vector<Real> & acq,
      56             :                            const std::vector<Real> & length_scales,
      57             :                            const std::vector<std::vector<Real>> & inputs);
      58             : 
      59             :   /**
      60             :    * Compute the correlation between two inputs using the length scales
      61             :    * @param corr The computed correlation value
      62             :    * @param input1 The first input
      63             :    * @param input2 The second input
      64             :    * @param length_scales The length scales to compute the correlation between inputs
      65             :    */
      66             :   void computeCorrelation(Real & corr,
      67             :                           const std::vector<Real> & input1,
      68             :                           const std::vector<Real> & input2,
      69             :                           const std::vector<Real> & length_scales);
      70             : 
      71             : protected:
      72             :   /**
      73             :    * Implementation hook for derived classes (no size checks here).
      74             :    * @param acq The computed acquisition function values
      75             :    * @param gp_mean The provided GP mean values
      76             :    * @param gp_std The provided GP standard deviation values
      77             :    * @param test_inputs All the input values under which the GP has to be tested
      78             :    * @param train_inputs All the input values under which the GP has been trained
      79             :    * @param generic A generic parameter (can be output values under which the GP has been trained or
      80             :    * threshold parameter under the U-Function etc.)
      81             :    */
      82             :   virtual void computeAcquisitionInternal(std::vector<Real> & acq,
      83             :                                           const std::vector<Real> & gp_mean,
      84             :                                           const std::vector<Real> & gp_std,
      85             :                                           const std::vector<std::vector<Real>> & test_inputs,
      86             :                                           const std::vector<std::vector<Real>> & train_inputs,
      87             :                                           const std::vector<Real> & generic) const = 0;
      88             : };