doc/en/ref_algorithm_DifferentialEvolution.rst

   1 ..
   2    Copyright (C) 2008-2018 EDF R&D
   3
   4    This file is part of SALOME ADAO module.
   5
   6    This library is free software; you can redistribute it and/or
   7    modify it under the terms of the GNU Lesser General Public
   8    License as published by the Free Software Foundation; either
   9    version 2.1 of the License, or (at your option) any later version.
  10
  11    This library is distributed in the hope that it will be useful,
  12    but WITHOUT ANY WARRANTY; without even the implied warranty of
  13    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14    Lesser General Public License for more details.
  15
  16    You should have received a copy of the GNU Lesser General Public
  17    License along with this library; if not, write to the Free Software
  18    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
  19
  20    See http://www.salome-platform.org/ or email : webmaster.salome@opencascade.com
  21
  22    Author: Jean-Philippe Argaud, jean-philippe.argaud@edf.fr, EDF R&D
  23
  24 .. index:: single: DifferentialEvolution
  25 .. _section_ref_algorithm_DifferentialEvolution:
  26
  27 Calculation algorithm "*DifferentialEvolution*"
  28 ----------------------------------------------------
  29
  30 .. warning::
  31
  32   in its present version, this algorithm is experimental, and so changes can be
  33   required in forthcoming versions.
  34
  35 Description
  36 +++++++++++
  37
  38 This algorithm realizes an estimation of the state of a system by minimization
  39 of a cost function :math:`J` by using an evolutionary strategy of differential
  40 evolution. It is a method that does not use the derivatives of the cost
  41 function. It falls in the same category than the
  42 :ref:`section_ref_algorithm_DerivativeFreeOptimization` or the
  43 :ref:`section_ref_algorithm_ParticleSwarmOptimization`.
  44
  45 This is an optimization method allowing for global minimum search of a general
  46 error function :math:`J` of type :math:`L^1`, :math:`L^2` or :math:`L^{\infty}`,
  47 with or without weights. The default error function is the augmented weighted
  48 least squares function, classically used in data assimilation.
  49
  50 Optional and required commands
  51 ++++++++++++++++++++++++++++++
  52
  53 The general required commands, available in the editing user interface, are the
  54 following:
  55
  56   .. include:: snippets/Background.rst
  57
  58   .. include:: snippets/BackgroundError.rst
  59
  60   .. include:: snippets/Observation.rst
  61
  62   .. include:: snippets/ObservationError.rst
  63
  64   .. include:: snippets/ObservationOperator.rst
  65
  66 The general optional commands, available in the editing user interface, are
  67 indicated in :ref:`section_ref_assimilation_keywords`. Moreover, the parameters
  68 of the command "*AlgorithmParameters*" allows to choose the specific options,
  69 described hereafter, of the algorithm. See
  70 :ref:`section_ref_options_Algorithm_Parameters` for the good use of this
  71 command.
  72
  73 The options of the algorithm are the following:
  74
  75   .. include:: snippets/Minimizer_DE.rst
  76
  77   .. include:: snippets/BoundsWithExtremes.rst
  78
  79   .. include:: snippets/CrossOverProbability_CR.rst
  80
  81   .. include:: snippets/MaximumNumberOfSteps.rst
  82
  83   .. include:: snippets/MaximumNumberOfFunctionEvaluations.rst
  84
  85   .. include:: snippets/MutationDifferentialWeight_F.rst
  86
  87   .. include:: snippets/PopulationSize.rst
  88
  89   .. include:: snippets/QualityCriterion.rst
  90
  91   .. include:: snippets/SetSeed.rst
  92
  93   StoreSupplementaryCalculations
  94     .. index:: single: StoreSupplementaryCalculations
  95
  96     This list indicates the names of the supplementary variables that can be
  97     available at the end of the algorithm. It involves potentially costly
  98     calculations or memory consumptions. The default is a void list, none of
  99     these variables being calculated and stored by default. The possible names
 100     are in the following list: ["BMA", "CostFunctionJ",
 101     "CostFunctionJAtCurrentOptimum", "CostFunctionJb",
 102     "CostFunctionJbAtCurrentOptimum", "CostFunctionJo",
 103     "CostFunctionJoAtCurrentOptimum", "CurrentOptimum", "CurrentState",
 104     "IndexOfOptimum", "Innovation", "InnovationAtCurrentState", "OMA", "OMB",
 105     "SimulatedObservationAtBackground", "SimulatedObservationAtCurrentOptimum",
 106     "SimulatedObservationAtCurrentState", "SimulatedObservationAtOptimum"].
 107
 108     Example :
 109     ``{"StoreSupplementaryCalculations":["BMA", "Innovation"]}``
 110
 111 Information and variables available at the end of the algorithm
 112 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 113
 114 At the output, after executing the algorithm, there are variables and
 115 information originating from the calculation. The description of
 116 :ref:`section_ref_output_variables` show the way to obtain them by the method
 117 named ``get`` of the variable "*ADD*" of the post-processing. The input
 118 variables, available to the user at the output in order to facilitate the
 119 writing of post-processing procedures, are described in the
 120 :ref:`subsection_r_o_v_Inventaire`.
 121
 122 The unconditional outputs of the algorithm are the following:
 123
 124   .. include:: snippets/Analysis.rst
 125
 126   .. include:: snippets/CostFunctionJ.rst
 127
 128   .. include:: snippets/CostFunctionJb.rst
 129
 130   .. include:: snippets/CostFunctionJo.rst
 131
 132   .. include:: snippets/CurrentState.rst
 133
 134 The conditional outputs of the algorithm are the following:
 135
 136   .. include:: snippets/BMA.rst
 137
 138   .. include:: snippets/CostFunctionJAtCurrentOptimum.rst
 139
 140   .. include:: snippets/CostFunctionJbAtCurrentOptimum.rst
 141
 142   .. include:: snippets/CostFunctionJoAtCurrentOptimum.rst
 143
 144   .. include:: snippets/CurrentOptimum.rst
 145
 146   .. include:: snippets/IndexOfOptimum.rst
 147
 148   .. include:: snippets/Innovation.rst
 149
 150   .. include:: snippets/InnovationAtCurrentState.rst
 151
 152   .. include:: snippets/OMA.rst
 153
 154   .. include:: snippets/OMB.rst
 155
 156   .. include:: snippets/SimulatedObservationAtBackground.rst
 157
 158   .. include:: snippets/SimulatedObservationAtCurrentOptimum.rst
 159
 160   .. include:: snippets/SimulatedObservationAtCurrentState.rst
 161
 162   .. include:: snippets/SimulatedObservationAtOptimum.rst
 163
 164 See also
 165 ++++++++
 166
 167 References to other sections:
 168   - :ref:`section_ref_algorithm_DerivativeFreeOptimization`
 169   - :ref:`section_ref_algorithm_ParticleSwarmOptimization`
 170
 171 Bibliographical references:
 172   - [Chakraborty08]_
 173   - [Price05]_
 174   - [Storn97]_