1 # -*- coding: utf-8 -*-
3 # Copyright (C) 2008-2024 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
26 __author__ = "Jean-Philippe ARGAUD"
28 import numpy, scipy, scipy.optimize, scipy.version
29 from daCore.NumericObjects import HessienneEstimation, QuantilesEstimations
30 from daCore.PlatformInfo import vt, vfloat
32 # ==============================================================================
33 def std3dvar(selfA, Xb, Y, U, HO, CM, R, B, __storeState = False):
40 Hm = HO["Direct"].appliedTo
41 Ha = HO["Adjoint"].appliedInXTo
43 if HO["AppliedInX"] is not None and "HXb" in HO["AppliedInX"]:
44 HXb = numpy.asarray(Hm( Xb, HO["AppliedInX"]["HXb"] ))
46 HXb = numpy.asarray(Hm( Xb ))
47 HXb = HXb.reshape((-1, 1))
48 if Y.size != HXb.size:
49 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)) # noqa: E501
50 if max(Y.shape) != max(HXb.shape):
51 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)) # noqa: E501
53 if selfA._toStore("JacobianMatrixAtBackground"):
54 HtMb = HO["Tangent"].asMatrix(Xb)
55 HtMb = HtMb.reshape(Y.size, Xb.size) # ADAO & check shape
56 selfA.StoredVariables["JacobianMatrixAtBackground"].store( HtMb )
61 Xini = selfA._parameters["InitializationPoint"]
63 # Définition de la fonction-coût
64 # ------------------------------
67 _X = numpy.asarray(x).reshape((-1, 1))
68 if selfA._parameters["StoreInternalVariables"] or \
69 selfA._toStore("CurrentState") or \
70 selfA._toStore("CurrentOptimum"):
71 selfA.StoredVariables["CurrentState"].store( _X )
72 _HX = numpy.asarray(Hm( _X )).reshape((-1, 1))
74 if selfA._toStore("SimulatedObservationAtCurrentState") or \
75 selfA._toStore("SimulatedObservationAtCurrentOptimum"):
76 selfA.StoredVariables["SimulatedObservationAtCurrentState"].store( _HX )
77 if selfA._toStore("InnovationAtCurrentState"):
78 selfA.StoredVariables["InnovationAtCurrentState"].store( _Innovation )
80 Jb = vfloat( 0.5 * (_X - Xb).T * (BI * (_X - Xb)) )
81 Jo = vfloat( 0.5 * _Innovation.T * (RI * _Innovation) )
84 selfA.StoredVariables["CurrentIterationNumber"].store( len(selfA.StoredVariables["CostFunctionJ"]) )
85 selfA.StoredVariables["CostFunctionJb"].store( Jb )
86 selfA.StoredVariables["CostFunctionJo"].store( Jo )
87 selfA.StoredVariables["CostFunctionJ" ].store( J )
88 if selfA._toStore("IndexOfOptimum") or \
89 selfA._toStore("CurrentOptimum") or \
90 selfA._toStore("CostFunctionJAtCurrentOptimum") or \
91 selfA._toStore("CostFunctionJbAtCurrentOptimum") or \
92 selfA._toStore("CostFunctionJoAtCurrentOptimum") or \
93 selfA._toStore("SimulatedObservationAtCurrentOptimum"):
94 IndexMin = numpy.argmin( selfA.StoredVariables["CostFunctionJ"][nbPreviousSteps:] ) + nbPreviousSteps
95 if selfA._toStore("IndexOfOptimum"):
96 selfA.StoredVariables["IndexOfOptimum"].store( IndexMin )
97 if selfA._toStore("CurrentOptimum"):
98 selfA.StoredVariables["CurrentOptimum"].store( selfA.StoredVariables["CurrentState"][IndexMin] )
99 if selfA._toStore("SimulatedObservationAtCurrentOptimum"):
100 selfA.StoredVariables["SimulatedObservationAtCurrentOptimum"].store( selfA.StoredVariables["SimulatedObservationAtCurrentState"][IndexMin] ) # noqa: E501
101 if selfA._toStore("CostFunctionJbAtCurrentOptimum"):
102 selfA.StoredVariables["CostFunctionJbAtCurrentOptimum"].store( selfA.StoredVariables["CostFunctionJb"][IndexMin] ) # noqa: E501
103 if selfA._toStore("CostFunctionJoAtCurrentOptimum"):
104 selfA.StoredVariables["CostFunctionJoAtCurrentOptimum"].store( selfA.StoredVariables["CostFunctionJo"][IndexMin] ) # noqa: E501
105 if selfA._toStore("CostFunctionJAtCurrentOptimum"):
106 selfA.StoredVariables["CostFunctionJAtCurrentOptimum" ].store( selfA.StoredVariables["CostFunctionJ" ][IndexMin] ) # noqa: E501
109 def GradientOfCostFunction(x):
110 _X = numpy.asarray(x).reshape((-1, 1))
111 _HX = numpy.asarray(Hm( _X )).reshape((-1, 1))
112 GradJb = BI * (_X - Xb)
113 GradJo = - Ha( (_X, RI * (Y - _HX)) )
114 GradJ = numpy.ravel( GradJb ) + numpy.ravel( GradJo )
117 # Minimisation de la fonctionnelle
118 # --------------------------------
119 nbPreviousSteps = selfA.StoredVariables["CostFunctionJ"].stepnumber()
121 if selfA._parameters["Minimizer"] == "LBFGSB":
122 if vt("0.19") <= vt(scipy.version.version) <= vt("1.4.99"):
123 import daAlgorithms.Atoms.lbfgsb14hlt as optimiseur
124 elif vt("1.5.0") <= vt(scipy.version.version) <= vt("1.7.99"):
125 import daAlgorithms.Atoms.lbfgsb17hlt as optimiseur
126 elif vt("1.8.0") <= vt(scipy.version.version) <= vt("1.8.99"):
127 import daAlgorithms.Atoms.lbfgsb18hlt as optimiseur
128 elif vt("1.9.0") <= vt(scipy.version.version) <= vt("1.10.99"):
129 import daAlgorithms.Atoms.lbfgsb19hlt as optimiseur
130 elif vt("1.11.0") <= vt(scipy.version.version) <= vt("1.11.99"):
131 import daAlgorithms.Atoms.lbfgsb111hlt as optimiseur
132 elif vt("1.12.0") <= vt(scipy.version.version) <= vt("1.12.99"):
133 import daAlgorithms.Atoms.lbfgsb112hlt as optimiseur
134 elif vt("1.13.0") <= vt(scipy.version.version) <= vt("1.13.99"):
135 import daAlgorithms.Atoms.lbfgsb113hlt as optimiseur
136 elif vt("1.14.0") <= vt(scipy.version.version) <= vt("1.14.99"):
137 import daAlgorithms.Atoms.lbfgsb114hlt as optimiseur
139 import scipy.optimize as optimiseur
140 Minimum, J_optimal, Informations = optimiseur.fmin_l_bfgs_b(
143 fprime = GradientOfCostFunction,
145 bounds = selfA._parameters["Bounds"],
146 maxfun = selfA._parameters["MaximumNumberOfIterations"] - 1,
147 factr = selfA._parameters["CostDecrementTolerance"] * 1.e14,
148 pgtol = selfA._parameters["ProjectedGradientTolerance"],
149 iprint = selfA._parameters["optiprint"],
151 # nfeval = Informations['funcalls']
152 # rc = Informations['warnflag']
153 elif selfA._parameters["Minimizer"] == "TNC":
154 Minimum, nfeval, rc = scipy.optimize.fmin_tnc(
157 fprime = GradientOfCostFunction,
159 bounds = selfA._parameters["Bounds"],
160 maxfun = selfA._parameters["MaximumNumberOfIterations"],
161 pgtol = selfA._parameters["ProjectedGradientTolerance"],
162 ftol = selfA._parameters["CostDecrementTolerance"],
163 messages = selfA._parameters["optmessages"],
165 elif selfA._parameters["Minimizer"] == "CG":
166 Minimum, fopt, nfeval, grad_calls, rc = scipy.optimize.fmin_cg(
169 fprime = GradientOfCostFunction,
171 maxiter = selfA._parameters["MaximumNumberOfIterations"],
172 gtol = selfA._parameters["GradientNormTolerance"],
173 disp = selfA._parameters["optdisp"],
176 elif selfA._parameters["Minimizer"] == "NCG":
177 Minimum, fopt, nfeval, grad_calls, hcalls, rc = scipy.optimize.fmin_ncg(
180 fprime = GradientOfCostFunction,
182 maxiter = selfA._parameters["MaximumNumberOfIterations"],
183 avextol = selfA._parameters["CostDecrementTolerance"],
184 disp = selfA._parameters["optdisp"],
187 elif selfA._parameters["Minimizer"] == "BFGS":
188 Minimum, fopt, gopt, Hopt, nfeval, grad_calls, rc = scipy.optimize.fmin_bfgs(
191 fprime = GradientOfCostFunction,
193 maxiter = selfA._parameters["MaximumNumberOfIterations"],
194 gtol = selfA._parameters["GradientNormTolerance"],
195 disp = selfA._parameters["optdisp"],
199 raise ValueError("Error in minimizer name: %s is unkown"%selfA._parameters["Minimizer"])
201 IndexMin = numpy.argmin( selfA.StoredVariables["CostFunctionJ"][nbPreviousSteps:] ) + nbPreviousSteps
202 MinJ = selfA.StoredVariables["CostFunctionJ"][IndexMin]
204 # Correction pour pallier a un bug de TNC sur le retour du Minimum
205 # ----------------------------------------------------------------
206 if selfA._parameters["StoreInternalVariables"] or selfA._toStore("CurrentState"):
207 Minimum = selfA.StoredVariables["CurrentState"][IndexMin]
211 selfA._setInternalState("Xn", Xa)
212 # --------------------------
214 selfA.StoredVariables["Analysis"].store( Xa )
216 if selfA._toStore("OMA") or \
217 selfA._toStore("InnovationAtCurrentAnalysis") or \
218 selfA._toStore("SigmaObs2") or \
219 selfA._toStore("SimulationQuantiles") or \
220 selfA._toStore("SimulatedObservationAtOptimum"):
221 if selfA._toStore("SimulatedObservationAtCurrentState"):
222 HXa = selfA.StoredVariables["SimulatedObservationAtCurrentState"][IndexMin]
223 elif selfA._toStore("SimulatedObservationAtCurrentOptimum"):
224 HXa = selfA.StoredVariables["SimulatedObservationAtCurrentOptimum"][-1]
227 oma = Y - numpy.asarray(HXa).reshape((-1, 1))
229 if selfA._toStore("APosterioriCovariance") or \
230 selfA._toStore("SimulationQuantiles") or \
231 selfA._toStore("JacobianMatrixAtOptimum") or \
232 selfA._toStore("KalmanGainAtOptimum"):
233 HtM = HO["Tangent"].asMatrix(ValueForMethodForm = Xa)
234 HtM = HtM.reshape(Y.size, Xa.size) # ADAO & check shape
235 if selfA._toStore("APosterioriCovariance") or \
236 selfA._toStore("SimulationQuantiles") or \
237 selfA._toStore("KalmanGainAtOptimum"):
238 HaM = HO["Adjoint"].asMatrix(ValueForMethodForm = Xa)
239 HaM = HaM.reshape(Xa.size, Y.size) # ADAO & check shape
240 if selfA._toStore("APosterioriCovariance") or \
241 selfA._toStore("SimulationQuantiles"):
242 A = HessienneEstimation(selfA, Xa.size, HaM, HtM, BI, RI)
243 if selfA._toStore("APosterioriCovariance"):
244 selfA.StoredVariables["APosterioriCovariance"].store( A )
245 if selfA._toStore("JacobianMatrixAtOptimum"):
246 selfA.StoredVariables["JacobianMatrixAtOptimum"].store( HtM )
247 if selfA._toStore("KalmanGainAtOptimum"):
248 if (Y.size <= Xb.size):
249 KG = B * HaM * (R + numpy.dot(HtM, B * HaM)).I
250 elif (Y.size > Xb.size):
251 KG = (BI + numpy.dot(HaM, RI * HtM)).I * HaM * RI
252 selfA.StoredVariables["KalmanGainAtOptimum"].store( KG )
254 # Calculs et/ou stockages supplémentaires
255 # ---------------------------------------
256 if selfA._toStore("Innovation") or \
257 selfA._toStore("SigmaObs2") or \
258 selfA._toStore("MahalanobisConsistency") or \
259 selfA._toStore("OMB"):
261 if selfA._toStore("Innovation"):
262 selfA.StoredVariables["Innovation"].store( Innovation )
263 if selfA._toStore("BMA"):
264 selfA.StoredVariables["BMA"].store( numpy.ravel(Xb) - numpy.ravel(Xa) )
265 if selfA._toStore("OMA"):
266 selfA.StoredVariables["OMA"].store( oma )
267 if selfA._toStore("InnovationAtCurrentAnalysis"):
268 selfA.StoredVariables["InnovationAtCurrentAnalysis"].store( oma )
269 if selfA._toStore("OMB"):
270 selfA.StoredVariables["OMB"].store( Innovation )
271 if selfA._toStore("SigmaObs2"):
272 TraceR = R.trace(Y.size)
273 selfA.StoredVariables["SigmaObs2"].store( vfloat( (Innovation.T @ oma) ) / TraceR )
274 if selfA._toStore("MahalanobisConsistency"):
275 selfA.StoredVariables["MahalanobisConsistency"].store( float( 2. * MinJ / Innovation.size ) )
276 if selfA._toStore("SimulationQuantiles"):
277 QuantilesEstimations(selfA, A, Xa, HXa, Hm, HtM)
278 if selfA._toStore("SimulatedObservationAtBackground"):
279 selfA.StoredVariables["SimulatedObservationAtBackground"].store( HXb )
280 if selfA._toStore("SimulatedObservationAtOptimum"):
281 selfA.StoredVariables["SimulatedObservationAtOptimum"].store( HXa )
285 # ==============================================================================
286 if __name__ == "__main__":
287 print('\n AUTODIAGNOSTIC\n')