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, "4DVAR")
31 self.defineRequiredParameter(
32 name = "ConstrainedBy",
33 default = "EstimateProjection",
35 message = "Prise en compte des contraintes",
36 listval = ["EstimateProjection"],
38 self.defineRequiredParameter(
39 name = "EstimationOf",
42 message = "Estimation d'etat ou de parametres",
43 listval = ["State", "Parameters"],
45 self.defineRequiredParameter(
49 message = "Minimiseur utilisé",
50 listval = ["LBFGSB","TNC", "CG", "NCG", "BFGS"],
52 self.defineRequiredParameter(
53 name = "MaximumNumberOfSteps",
56 message = "Nombre maximal de pas d'optimisation",
59 self.defineRequiredParameter(
60 name = "CostDecrementTolerance",
63 message = "Diminution relative minimale du coût lors de l'arrêt",
66 self.defineRequiredParameter(
67 name = "ProjectedGradientTolerance",
70 message = "Maximum des composantes du gradient projeté lors de l'arrêt",
73 self.defineRequiredParameter(
74 name = "GradientNormTolerance",
77 message = "Maximum des composantes du gradient lors de l'arrêt",
80 self.defineRequiredParameter(
81 name = "StoreInternalVariables",
84 message = "Stockage des variables internes ou intermédiaires du calcul",
86 self.defineRequiredParameter(
87 name = "StoreSupplementaryCalculations",
90 message = "Liste de calculs supplémentaires à stocker et/ou effectuer",
95 "CostFunctionJAtCurrentOptimum",
97 "CostFunctionJbAtCurrentOptimum",
99 "CostFunctionJoAtCurrentOptimum",
105 self.defineRequiredParameter( # Pas de type
107 message = "Liste des valeurs de bornes",
109 self.requireInputArguments(
110 mandatory= ("Xb", "Y", "HO", "EM", "R", "B" ),
111 optional = ("U", "CM"),
113 self.setAttributes(tags=(
120 def run(self, Xb=None, Y=None, U=None, HO=None, EM=None, CM=None, R=None, B=None, Q=None, Parameters=None):
121 self._pre_run(Parameters, Xb, Y, U, HO, EM, CM, R, B, Q)
123 # Correction pour pallier a un bug de TNC sur le retour du Minimum
124 if "Minimizer" in self._parameters and self._parameters["Minimizer"] == "TNC":
125 self.setParameterValue("StoreInternalVariables",True)
129 Hm = HO["Direct"].appliedControledFormTo
131 Mm = EM["Direct"].appliedControledFormTo
133 if CM is not None and "Tangent" in CM and U is not None:
134 Cm = CM["Tangent"].asMatrix(Xb)
140 if hasattr(U,"store") and 1<=_step<len(U) :
141 _Un = numpy.asmatrix(numpy.ravel( U[_step] )).T
142 elif hasattr(U,"store") and len(U)==1:
143 _Un = numpy.asmatrix(numpy.ravel( U[0] )).T
145 _Un = numpy.asmatrix(numpy.ravel( U )).T
150 if Cm is not None and _un is not None: # Attention : si Cm est aussi dans M, doublon !
151 _Cm = Cm.reshape(_xn.size,_un.size) # ADAO & check shape
157 # Remarque : les observations sont exploitées à partir du pas de temps
158 # numéro 1, et sont utilisées dans Yo comme rangées selon ces indices.
159 # Donc le pas 0 n'est pas utilisé puisque la première étape commence
160 # avec l'observation du pas 1.
162 # Nombre de pas identique au nombre de pas d'observations
163 # -------------------------------------------------------
164 if hasattr(Y,"stepnumber"):
165 duration = Y.stepnumber()
169 # Précalcul des inversions de B et R
170 # ----------------------------------
174 # Définition de la fonction-coût
175 # ------------------------------
176 self.DirectCalculation = [None,] # Le pas 0 n'est pas observé
177 self.DirectInnovation = [None,] # Le pas 0 n'est pas observé
179 _X = numpy.asmatrix(numpy.ravel( x )).T
180 if self._parameters["StoreInternalVariables"] or \
181 self._toStore("CurrentState") or \
182 self._toStore("CurrentOptimum"):
183 self.StoredVariables["CurrentState"].store( _X )
184 Jb = 0.5 * (_X - Xb).T * BI * (_X - Xb)
185 self.DirectCalculation = [None,]
186 self.DirectInnovation = [None,]
189 for step in range(0,duration-1):
190 self.DirectCalculation.append( _Xn )
191 if hasattr(Y,"store"):
192 _Ynpu = numpy.asmatrix(numpy.ravel( Y[step+1] )).T
194 _Ynpu = numpy.asmatrix(numpy.ravel( Y )).T
198 if self._parameters["EstimationOf"] == "State":
199 _Xn = Mm( (_Xn, _Un) ) + CmUn(_Xn, _Un)
200 elif self._parameters["EstimationOf"] == "Parameters":
203 if self._parameters["Bounds"] is not None and self._parameters["ConstrainedBy"] == "EstimateProjection":
204 _Xn = numpy.max(numpy.hstack((_Xn,numpy.asmatrix(self._parameters["Bounds"])[:,0])),axis=1)
205 _Xn = numpy.min(numpy.hstack((_Xn,numpy.asmatrix(self._parameters["Bounds"])[:,1])),axis=1)
207 # Etape de différence aux observations
208 if self._parameters["EstimationOf"] == "State":
209 _YmHMX = _Ynpu - numpy.asmatrix(numpy.ravel( Hm( (_Xn, None) ) )).T
210 elif self._parameters["EstimationOf"] == "Parameters":
211 _YmHMX = _Ynpu - numpy.asmatrix(numpy.ravel( Hm( (_Xn, _Un) ) )).T - CmUn(_Xn, _Un)
212 self.DirectInnovation.append( _YmHMX )
213 # Ajout dans la fonctionnelle d'observation
214 Jo = Jo + _YmHMX.T * RI * _YmHMX
216 J = float( Jb ) + float( Jo )
217 self.StoredVariables["CostFunctionJb"].store( Jb )
218 self.StoredVariables["CostFunctionJo"].store( Jo )
219 self.StoredVariables["CostFunctionJ" ].store( J )
220 if self._toStore("IndexOfOptimum") or \
221 self._toStore("CurrentOptimum") or \
222 self._toStore("CostFunctionJAtCurrentOptimum") or \
223 self._toStore("CostFunctionJbAtCurrentOptimum") or \
224 self._toStore("CostFunctionJoAtCurrentOptimum"):
225 IndexMin = numpy.argmin( self.StoredVariables["CostFunctionJ"][nbPreviousSteps:] ) + nbPreviousSteps
226 if self._toStore("IndexOfOptimum"):
227 self.StoredVariables["IndexOfOptimum"].store( IndexMin )
228 if self._toStore("CurrentOptimum"):
229 self.StoredVariables["CurrentOptimum"].store( self.StoredVariables["CurrentState"][IndexMin] )
230 if self._toStore("CostFunctionJAtCurrentOptimum"):
231 self.StoredVariables["CostFunctionJAtCurrentOptimum" ].store( self.StoredVariables["CostFunctionJ" ][IndexMin] )
232 if self._toStore("CostFunctionJbAtCurrentOptimum"):
233 self.StoredVariables["CostFunctionJbAtCurrentOptimum"].store( self.StoredVariables["CostFunctionJb"][IndexMin] )
234 if self._toStore("CostFunctionJoAtCurrentOptimum"):
235 self.StoredVariables["CostFunctionJoAtCurrentOptimum"].store( self.StoredVariables["CostFunctionJo"][IndexMin] )
238 def GradientOfCostFunction(x):
239 _X = numpy.asmatrix(numpy.ravel( x )).T
240 GradJb = BI * (_X - Xb)
242 for step in range(duration-1,0,-1):
243 # Etape de récupération du dernier stockage de l'évolution
244 _Xn = self.DirectCalculation.pop()
245 # Etape de récupération du dernier stockage de l'innovation
246 _YmHMX = self.DirectInnovation.pop()
247 # Calcul des adjoints
248 Ha = HO["Adjoint"].asMatrix(ValueForMethodForm = _Xn)
249 Ha = Ha.reshape(_Xn.size,_YmHMX.size) # ADAO & check shape
250 Ma = EM["Adjoint"].asMatrix(ValueForMethodForm = _Xn)
251 Ma = Ma.reshape(_Xn.size,_Xn.size) # ADAO & check shape
252 # Calcul du gradient par etat adjoint
253 GradJo = GradJo + Ha * RI * _YmHMX # Equivaut pour Ha lineaire à : Ha( (_Xn, RI * _YmHMX) )
254 GradJo = Ma * GradJo # Equivaut pour Ma lineaire à : Ma( (_Xn, GradJo) )
255 GradJ = numpy.asmatrix( numpy.ravel( GradJb ) - numpy.ravel( GradJo ) ).T
258 # Point de démarrage de l'optimisation : Xini = Xb
259 # ------------------------------------
260 if isinstance(Xb, type(numpy.matrix([]))):
261 Xini = Xb.A1.tolist()
265 # Minimisation de la fonctionnelle
266 # --------------------------------
267 nbPreviousSteps = self.StoredVariables["CostFunctionJ"].stepnumber()
269 if self._parameters["Minimizer"] == "LBFGSB":
270 # Minimum, J_optimal, Informations = scipy.optimize.fmin_l_bfgs_b(
271 if "0.19" <= scipy.version.version <= "1.1.0":
272 import lbfgsbhlt as optimiseur
274 import scipy.optimize as optimiseur
275 Minimum, J_optimal, Informations = optimiseur.fmin_l_bfgs_b(
278 fprime = GradientOfCostFunction,
280 bounds = self._parameters["Bounds"],
281 maxfun = self._parameters["MaximumNumberOfSteps"]-1,
282 factr = self._parameters["CostDecrementTolerance"]*1.e14,
283 pgtol = self._parameters["ProjectedGradientTolerance"],
284 iprint = self._parameters["optiprint"],
286 nfeval = Informations['funcalls']
287 rc = Informations['warnflag']
288 elif self._parameters["Minimizer"] == "TNC":
289 Minimum, nfeval, rc = scipy.optimize.fmin_tnc(
292 fprime = GradientOfCostFunction,
294 bounds = self._parameters["Bounds"],
295 maxfun = self._parameters["MaximumNumberOfSteps"],
296 pgtol = self._parameters["ProjectedGradientTolerance"],
297 ftol = self._parameters["CostDecrementTolerance"],
298 messages = self._parameters["optmessages"],
300 elif self._parameters["Minimizer"] == "CG":
301 Minimum, fopt, nfeval, grad_calls, rc = scipy.optimize.fmin_cg(
304 fprime = GradientOfCostFunction,
306 maxiter = self._parameters["MaximumNumberOfSteps"],
307 gtol = self._parameters["GradientNormTolerance"],
308 disp = self._parameters["optdisp"],
311 elif self._parameters["Minimizer"] == "NCG":
312 Minimum, fopt, nfeval, grad_calls, hcalls, rc = scipy.optimize.fmin_ncg(
315 fprime = GradientOfCostFunction,
317 maxiter = self._parameters["MaximumNumberOfSteps"],
318 avextol = self._parameters["CostDecrementTolerance"],
319 disp = self._parameters["optdisp"],
322 elif self._parameters["Minimizer"] == "BFGS":
323 Minimum, fopt, gopt, Hopt, nfeval, grad_calls, rc = scipy.optimize.fmin_bfgs(
326 fprime = GradientOfCostFunction,
328 maxiter = self._parameters["MaximumNumberOfSteps"],
329 gtol = self._parameters["GradientNormTolerance"],
330 disp = self._parameters["optdisp"],
334 raise ValueError("Error in Minimizer name: %s"%self._parameters["Minimizer"])
336 IndexMin = numpy.argmin( self.StoredVariables["CostFunctionJ"][nbPreviousSteps:] ) + nbPreviousSteps
337 MinJ = self.StoredVariables["CostFunctionJ"][IndexMin]
339 # Correction pour pallier a un bug de TNC sur le retour du Minimum
340 # ----------------------------------------------------------------
341 if self._parameters["StoreInternalVariables"] or self._toStore("CurrentState"):
342 Minimum = self.StoredVariables["CurrentState"][IndexMin]
344 # Obtention de l'analyse
345 # ----------------------
346 Xa = numpy.asmatrix(numpy.ravel( Minimum )).T
348 self.StoredVariables["Analysis"].store( Xa.A1 )
350 # Calculs et/ou stockages supplémentaires
351 # ---------------------------------------
352 if self._toStore("BMA"):
353 self.StoredVariables["BMA"].store( numpy.ravel(Xb) - numpy.ravel(Xa) )
358 # ==============================================================================
359 if __name__ == "__main__":
360 print('\n AUTODIAGNOSTIC\n')