1 # -*- coding: utf-8 -*-
3 # Copyright (C) 2008-2020 EDF R&D
5 # This library is free software; you can redistribute it and/or
6 # modify it under the terms of the GNU Lesser General Public
7 # License as published by the Free Software Foundation; either
8 # version 2.1 of the License.
10 # This library is distributed in the hope that it will be useful,
11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 # Lesser General Public License for more details.
15 # You should have received a copy of the GNU Lesser General Public
16 # License along with this library; if not, write to the Free Software
17 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19 # See http://www.salome-platform.org/ or email : webmaster.salome@opencascade.com
21 # Author: Jean-Philippe Argaud, jean-philippe.argaud@edf.fr, EDF R&D
24 from daCore import BasicObjects
25 import numpy, scipy.optimize, scipy.version
27 # ==============================================================================
28 class ElementaryAlgorithm(BasicObjects.Algorithm):
30 BasicObjects.Algorithm.__init__(self, "NONLINEARLEASTSQUARES")
31 self.defineRequiredParameter(
35 message = "Minimiseur utilisé",
36 listval = ["LBFGSB","TNC", "CG", "NCG", "BFGS", "LM"],
38 self.defineRequiredParameter(
39 name = "MaximumNumberOfSteps",
42 message = "Nombre maximal de pas d'optimisation",
45 self.defineRequiredParameter(
46 name = "CostDecrementTolerance",
49 message = "Diminution relative minimale du coût lors de l'arrêt",
52 self.defineRequiredParameter(
53 name = "ProjectedGradientTolerance",
56 message = "Maximum des composantes du gradient projeté lors de l'arrêt",
59 self.defineRequiredParameter(
60 name = "GradientNormTolerance",
63 message = "Maximum des composantes du gradient lors de l'arrêt",
66 self.defineRequiredParameter(
67 name = "StoreInternalVariables",
70 message = "Stockage des variables internes ou intermédiaires du calcul",
72 self.defineRequiredParameter(
73 name = "StoreSupplementaryCalculations",
76 message = "Liste de calculs supplémentaires à stocker et/ou effectuer",
81 "CostFunctionJAtCurrentOptimum",
83 "CostFunctionJbAtCurrentOptimum",
85 "CostFunctionJoAtCurrentOptimum",
86 "CurrentIterationNumber",
91 "InnovationAtCurrentState",
94 "SimulatedObservationAtBackground",
95 "SimulatedObservationAtCurrentOptimum",
96 "SimulatedObservationAtCurrentState",
97 "SimulatedObservationAtOptimum",
100 self.defineRequiredParameter( # Pas de type
102 message = "Liste des valeurs de bornes",
104 self.requireInputArguments(
105 mandatory= ("Xb", "Y", "HO", "R"),
107 self.setAttributes(tags=(
113 def run(self, Xb=None, Y=None, U=None, HO=None, EM=None, CM=None, R=None, B=None, Q=None, Parameters=None):
114 self._pre_run(Parameters, Xb, Y, U, HO, EM, CM, R, B, Q)
116 # Correction pour pallier a un bug de TNC sur le retour du Minimum
117 if "Minimizer" in self._parameters and self._parameters["Minimizer"] == "TNC":
118 self.setParameterValue("StoreInternalVariables",True)
122 Hm = HO["Direct"].appliedTo
123 Ha = HO["Adjoint"].appliedInXTo
125 # Utilisation éventuelle d'un vecteur H(Xb) précalculé
126 # ----------------------------------------------------
127 if HO["AppliedInX"] is not None and "HXb" in HO["AppliedInX"]:
128 HXb = Hm( Xb, HO["AppliedInX"]["HXb"] )
131 HXb = numpy.asmatrix(numpy.ravel( HXb )).T
132 if Y.size != HXb.size:
133 raise ValueError("The size %i of observations Y and %i of observed calculation H(X) are different, they have to be identical."%(Y.size,HXb.size))
134 if max(Y.shape) != max(HXb.shape):
135 raise ValueError("The shapes %s of observations Y and %s of observed calculation H(X) are different, they have to be identical."%(Y.shape,HXb.shape))
137 # Précalcul des inversions de B et R
138 # ----------------------------------
140 if self._parameters["Minimizer"] == "LM":
141 RdemiI = R.choleskyI()
143 # Définition de la fonction-coût
144 # ------------------------------
146 _X = numpy.asmatrix(numpy.ravel( x )).T
147 if self._parameters["StoreInternalVariables"] or \
148 self._toStore("CurrentState") or \
149 self._toStore("CurrentOptimum"):
150 self.StoredVariables["CurrentState"].store( _X )
152 _HX = numpy.asmatrix(numpy.ravel( _HX )).T
153 _Innovation = Y - _HX
154 if self._toStore("SimulatedObservationAtCurrentState") or \
155 self._toStore("SimulatedObservationAtCurrentOptimum"):
156 self.StoredVariables["SimulatedObservationAtCurrentState"].store( _HX )
157 if self._toStore("InnovationAtCurrentState"):
158 self.StoredVariables["InnovationAtCurrentState"].store( _Innovation )
161 Jo = float( 0.5 * _Innovation.T * RI * _Innovation )
164 self.StoredVariables["CurrentIterationNumber"].store( len(self.StoredVariables["CostFunctionJ"]) )
165 self.StoredVariables["CostFunctionJb"].store( Jb )
166 self.StoredVariables["CostFunctionJo"].store( Jo )
167 self.StoredVariables["CostFunctionJ" ].store( J )
168 if self._toStore("IndexOfOptimum") or \
169 self._toStore("CurrentOptimum") or \
170 self._toStore("CostFunctionJAtCurrentOptimum") or \
171 self._toStore("CostFunctionJbAtCurrentOptimum") or \
172 self._toStore("CostFunctionJoAtCurrentOptimum") or \
173 self._toStore("SimulatedObservationAtCurrentOptimum"):
174 IndexMin = numpy.argmin( self.StoredVariables["CostFunctionJ"][nbPreviousSteps:] ) + nbPreviousSteps
175 if self._toStore("IndexOfOptimum"):
176 self.StoredVariables["IndexOfOptimum"].store( IndexMin )
177 if self._toStore("CurrentOptimum"):
178 self.StoredVariables["CurrentOptimum"].store( self.StoredVariables["CurrentState"][IndexMin] )
179 if self._toStore("SimulatedObservationAtCurrentOptimum"):
180 self.StoredVariables["SimulatedObservationAtCurrentOptimum"].store( self.StoredVariables["SimulatedObservationAtCurrentState"][IndexMin] )
181 if self._toStore("CostFunctionJbAtCurrentOptimum"):
182 self.StoredVariables["CostFunctionJbAtCurrentOptimum"].store( self.StoredVariables["CostFunctionJb"][IndexMin] )
183 if self._toStore("CostFunctionJoAtCurrentOptimum"):
184 self.StoredVariables["CostFunctionJoAtCurrentOptimum"].store( self.StoredVariables["CostFunctionJo"][IndexMin] )
185 if self._toStore("CostFunctionJAtCurrentOptimum"):
186 self.StoredVariables["CostFunctionJAtCurrentOptimum" ].store( self.StoredVariables["CostFunctionJ" ][IndexMin] )
189 def GradientOfCostFunction(x):
190 _X = numpy.asmatrix(numpy.ravel( x )).T
192 _HX = numpy.asmatrix(numpy.ravel( _HX )).T
194 GradJo = - Ha( (_X, RI * (Y - _HX)) )
195 GradJ = numpy.asmatrix( numpy.ravel( GradJb ) + numpy.ravel( GradJo ) ).T
198 def CostFunctionLM(x):
199 _X = numpy.asmatrix(numpy.ravel( x )).T
201 _HX = numpy.asmatrix(numpy.ravel( _HX )).T
202 _Innovation = Y - _HX
204 Jo = float( 0.5 * _Innovation.T * RI * _Innovation )
206 if self._parameters["StoreInternalVariables"] or \
207 self._toStore("CurrentState"):
208 self.StoredVariables["CurrentState"].store( _X )
209 self.StoredVariables["CostFunctionJb"].store( Jb )
210 self.StoredVariables["CostFunctionJo"].store( Jo )
211 self.StoredVariables["CostFunctionJ" ].store( J )
213 return numpy.ravel( RdemiI*_Innovation )
215 def GradientOfCostFunctionLM(x):
216 _X = numpy.asmatrix(numpy.ravel( x )).T
218 _HX = numpy.asmatrix(numpy.ravel( _HX )).T
220 GradJo = - Ha( (_X, RI * (Y - _HX)) )
221 GradJ = numpy.asmatrix( numpy.ravel( GradJb ) + numpy.ravel( GradJo ) ).T
222 return - RdemiI*HO["Tangent"].asMatrix( _X )
224 # Point de démarrage de l'optimisation : Xini = Xb
225 # ------------------------------------
226 Xini = numpy.ravel(Xb)
228 # Minimisation de la fonctionnelle
229 # --------------------------------
230 nbPreviousSteps = self.StoredVariables["CostFunctionJ"].stepnumber()
232 if self._parameters["Minimizer"] == "LBFGSB":
233 # Minimum, J_optimal, Informations = scipy.optimize.fmin_l_bfgs_b(
234 if "0.19" <= scipy.version.version <= "1.1.0":
235 import lbfgsbhlt as optimiseur
237 import scipy.optimize as optimiseur
238 Minimum, J_optimal, Informations = optimiseur.fmin_l_bfgs_b(
241 fprime = GradientOfCostFunction,
243 bounds = self._parameters["Bounds"],
244 maxfun = self._parameters["MaximumNumberOfSteps"]-1,
245 factr = self._parameters["CostDecrementTolerance"]*1.e14,
246 pgtol = self._parameters["ProjectedGradientTolerance"],
247 iprint = self._parameters["optiprint"],
249 nfeval = Informations['funcalls']
250 rc = Informations['warnflag']
251 elif self._parameters["Minimizer"] == "TNC":
252 Minimum, nfeval, rc = scipy.optimize.fmin_tnc(
255 fprime = GradientOfCostFunction,
257 bounds = self._parameters["Bounds"],
258 maxfun = self._parameters["MaximumNumberOfSteps"],
259 pgtol = self._parameters["ProjectedGradientTolerance"],
260 ftol = self._parameters["CostDecrementTolerance"],
261 messages = self._parameters["optmessages"],
263 elif self._parameters["Minimizer"] == "CG":
264 Minimum, fopt, nfeval, grad_calls, rc = scipy.optimize.fmin_cg(
267 fprime = GradientOfCostFunction,
269 maxiter = self._parameters["MaximumNumberOfSteps"],
270 gtol = self._parameters["GradientNormTolerance"],
271 disp = self._parameters["optdisp"],
274 elif self._parameters["Minimizer"] == "NCG":
275 Minimum, fopt, nfeval, grad_calls, hcalls, rc = scipy.optimize.fmin_ncg(
278 fprime = GradientOfCostFunction,
280 maxiter = self._parameters["MaximumNumberOfSteps"],
281 avextol = self._parameters["CostDecrementTolerance"],
282 disp = self._parameters["optdisp"],
285 elif self._parameters["Minimizer"] == "BFGS":
286 Minimum, fopt, gopt, Hopt, nfeval, grad_calls, rc = scipy.optimize.fmin_bfgs(
289 fprime = GradientOfCostFunction,
291 maxiter = self._parameters["MaximumNumberOfSteps"],
292 gtol = self._parameters["GradientNormTolerance"],
293 disp = self._parameters["optdisp"],
296 elif self._parameters["Minimizer"] == "LM":
297 Minimum, cov_x, infodict, mesg, rc = scipy.optimize.leastsq(
298 func = CostFunctionLM,
300 Dfun = GradientOfCostFunctionLM,
302 ftol = self._parameters["CostDecrementTolerance"],
303 maxfev = self._parameters["MaximumNumberOfSteps"],
304 gtol = self._parameters["GradientNormTolerance"],
307 nfeval = infodict['nfev']
309 raise ValueError("Error in Minimizer name: %s"%self._parameters["Minimizer"])
311 IndexMin = numpy.argmin( self.StoredVariables["CostFunctionJ"][nbPreviousSteps:] ) + nbPreviousSteps
312 MinJ = self.StoredVariables["CostFunctionJ"][IndexMin]
314 # Correction pour pallier a un bug de TNC sur le retour du Minimum
315 # ----------------------------------------------------------------
316 if self._parameters["StoreInternalVariables"] or self._toStore("CurrentState"):
317 Minimum = self.StoredVariables["CurrentState"][IndexMin]
319 # Obtention de l'analyse
320 # ----------------------
321 Xa = numpy.asmatrix(numpy.ravel( Minimum )).T
323 self.StoredVariables["Analysis"].store( Xa.A1 )
325 if self._toStore("OMA") or \
326 self._toStore("SimulatedObservationAtOptimum"):
327 if self._toStore("SimulatedObservationAtCurrentState"):
328 HXa = self.StoredVariables["SimulatedObservationAtCurrentState"][IndexMin]
329 elif self._toStore("SimulatedObservationAtCurrentOptimum"):
330 HXa = self.StoredVariables["SimulatedObservationAtCurrentOptimum"][-1]
335 # Calculs et/ou stockages supplémentaires
336 # ---------------------------------------
337 if self._toStore("Innovation") or \
338 self._toStore("OMB"):
340 if self._toStore("Innovation"):
341 self.StoredVariables["Innovation"].store( numpy.ravel(d) )
342 if self._toStore("BMA"):
343 self.StoredVariables["BMA"].store( numpy.ravel(Xb) - numpy.ravel(Xa) )
344 if self._toStore("OMA"):
345 self.StoredVariables["OMA"].store( numpy.ravel(Y) - numpy.ravel(HXa) )
346 if self._toStore("OMB"):
347 self.StoredVariables["OMB"].store( numpy.ravel(d) )
348 if self._toStore("SimulatedObservationAtBackground"):
349 self.StoredVariables["SimulatedObservationAtBackground"].store( numpy.ravel(HXb) )
350 if self._toStore("SimulatedObservationAtOptimum"):
351 self.StoredVariables["SimulatedObservationAtOptimum"].store( numpy.ravel(HXa) )
356 # ==============================================================================
357 if __name__ == "__main__":
358 print('\n AUTODIAGNOSTIC\n')