diff --git a/.gitignore b/.gitignore
index acb906e2023293020272c9329370d5d02129c7ce..7aa5fc0a2f338b13b8bd7f9765e9a051d98a96a0 100644
--- a/.gitignore
+++ b/.gitignore
@@ -65,3 +65,54 @@ BayesValidRox/tests/AnalyticalFunction/Outputs_sequential_BayesOptDesign/
BayesValidRox/tests/AnalyticalFunction/Outputs_sequential_alphabetic/
BayesValidRox/tests/AnalyticalFunction/Polynomials/
BayesValidRox/tests/AnalyticalFunction/__pycache__/
+BayesValidRox/BayesInference/__pycache__/BayesInference.cpython-37.pyc
+BayesValidRox/BayesInference/__pycache__/Discrepancy.cpython-37.pyc
+BayesValidRox/BayesInference/__pycache__/__init__.cpython-37.pyc
+BayesValidRox/PostProcessing/__pycache__/PostProcessing.cpython-37.pyc
+BayesValidRox/PostProcessing/__pycache__/__init__.cpython-37.pyc
+BayesValidRox/PyLink/__pycache__/FuncForwardModel.cpython-37.pyc
+BayesValidRox/PyLink/__pycache__/PyLinkForwardModel.cpython-37.pyc
+BayesValidRox/PyLink/__pycache__/__init__.cpython-37.pyc
+BayesValidRox/surrogate_models/__pycache__/ExpDesigns.cpython-37.pyc
+BayesValidRox/surrogate_models/__pycache__/Exploration.cpython-37.pyc
+BayesValidRox/surrogate_models/__pycache__/Input.cpython-37.pyc
+BayesValidRox/surrogate_models/__pycache__/Voronoi.cpython-37.pyc
+BayesValidRox/surrogate_models/__pycache__/__init__.cpython-37.pyc
+BayesValidRox/surrogate_models/__pycache__/aPoly_Construction.cpython-37.pyc
+BayesValidRox/surrogate_models/__pycache__/surrogate_models.cpython-37.pyc
+BayesValidRox/tests/AnalyticalFunction/Outputs_demo/
+BayesValidRox/tests/BeamTest/Beam_Results.pkl
+BayesValidRox/tests/BeamTest/Polynomials/Roots_degree7_Pa1.npy
+BayesValidRox/tests/BeamTest/Polynomials/Roots_degree7_Pa2.npy
+BayesValidRox/tests/BeamTest/Polynomials/Roots_degree7_Pa3.npy
+BayesValidRox/tests/BeamTest/Polynomials/Roots_degree7_Pa4.npy
+BayesValidRox/tests/BeamTest/__pycache__/read_Beam_Deflection.cpython-37.pyc
+BayesValidRox/tests/BeamTest/surrogateBeam/Outputs_PostPorecessing/
+BayesValidRox/tests/BeamTest/surrogateBeam/Polynomials/
+BayesValidRox/tests/BeamTest/surrogateBeam/__pycache__/read_Beam_Deflection.cpython-37.pyc
+BayesValidRox/tests/BraninTest/Branin_Results.pkl
+BayesValidRox/tests/BraninTest/Outputs_Bayes_Calib/
+BayesValidRox/tests/BraninTest/Outputs_PostPorecessing/
+BayesValidRox/tests/BraninTest/Outputs_SeqPosteriorComparison/
+BayesValidRox/tests/BraninTest/Outputs_normal_BayesOptDesign/
+BayesValidRox/tests/BraninTest/Outputs_sequential_BayesOptDesign/
+BayesValidRox/tests/BraninTest/Outputs_sequential_dual annealing/
+BayesValidRox/tests/BraninTest/__pycache__/Branin.cpython-37.pyc
+BayesValidRox/tests/Co2BenchmarkTest/CollocationPoints.npy
+BayesValidRox/tests/Co2BenchmarkTest/ExpDesign_Y.npy
+BayesValidRox/tests/Co2BenchmarkTest/Results/
+"BayesValidRox/tests/Co2BenchmarkTest/uqBenchmarkData/K\303\266ppel2019_Article_ComparisonOfData-drivenUncerta.pdf"
+BayesValidRox/tests/Co2BenchmarkTest/uqBenchmarkData/mc_10.zip
+BayesValidRox/tests/IshigamiTest/Outputs_PostPorecessing/
+BayesValidRox/tests/IshigamiTest/Polynomials/
+BayesValidRox/tests/MECBM/.model_exe_orig.sh.swp
+BayesValidRox/tests/MECBM/SurrogateModel/MECBM.zip
+BayesValidRox/tests/MECBM/SurrogateModel/MECBMValid.zip
+BayesValidRox/tests/MECBM/SurrogateModel/MECBM_SA_Results.pkl
+BayesValidRox/tests/MECBM/SurrogateModel/NormalDesign_100/
+BayesValidRox/tests/MECBM/SurrogateModel/SeqDesign_25_100/
+BayesValidRox/tests/MECBM/SurrogateModel/SeqDesign_25_100_APP/
+BayesValidRox/tests/MECBM/SurrogateModel/SeqDesign_25_100_DPP/
+BayesValidRox/tests/MECBM/SurrogateModel/SeqDesign_25_150_DKL/
+BayesValidRox/tests/MECBM/SurrogateModel/__pycache__/read_MECBM.cpython-37.pyc
+BayesValidRox/tests/MECBM/model_exe_orig.sh
diff --git a/BayesValidRox/PostProcessing/PostProcessing.py b/BayesValidRox/PostProcessing/PostProcessing.py
index 660f8e1a2cb332d43d51f9d71a0d265fb4f1952c..c6f9dde320f468e7f1c9533a2339091407e3f04e 100644
--- a/BayesValidRox/PostProcessing/PostProcessing.py
+++ b/BayesValidRox/PostProcessing/PostProcessing.py
@@ -54,7 +54,7 @@ class PostProcessing:
# Mean = c_0
PCEMean[idx] = coeffs[0]
# Std = sum(coeffs[1:]^2)
- PCEStd[idx] = np.sum(np.square(coeffs[1:]))
+ PCEStd[idx] = np.sqrt(np.sum(np.square(coeffs[1:])))
self.PCEMeans[Outkey] = PCEMean
self.PCEStd[Outkey] = PCEStd
@@ -542,33 +542,33 @@ class PostProcessing:
"""
PCEModel = self.PCEModel
- plotList = ['Modified LOO error', 'KLD', 'BME']
- seqList = [PCEModel.SeqModifiedLOO, PCEModel.SeqKLD, PCEModel.SeqBME]
-
- # Plot the evolution of the diagnostic criteria of the Sequential Experimental Design.
-
- markers = ['x', 'o', 'd']
- colors = ['k', 'r', 'b']
+ plotList = ['Modified LOO error', 'KLD', 'BME', 'RMSEMean', 'RMSEStd']
+ seqList = [PCEModel.SeqModifiedLOO, PCEModel.SeqKLD, PCEModel.SeqBME,
+ PCEModel.seqRMSEMean, PCEModel.seqRMSEStd]
+ markers = ('x', 'o', 'd', '*', '+')
+ colors = ('k', 'darkgreen', 'b', 'navy', 'darkred')
+ # Plot the evolution of the diagnostic criteria of the Sequential Experimental Design.
for plotidx, plot in enumerate(plotList):
- fig = plt.figure(figsize=(24, 16))
SeqDict = seqList[plotidx]
nameUtil = list(SeqDict.keys())
+ if len(nameUtil) == 0:
+ continue
+
+ fig = plt.figure(figsize=(24, 16))
for idx, name in enumerate(nameUtil):
SeqValues = SeqDict[name]
x_idx = np.arange(NrofInitSamples, SeqValues.shape[0]+NrofInitSamples)
- marker = markers[idx]
- color = colors[idx]
-
- if plot == 'Modified LOO error':
- plt.semilogy(x_idx, SeqValues, marker=marker, color=color, label=name)
- else:
- plt.plot(x_idx, SeqValues, marker=marker, color=color, label=name)
+ if plot == 'KLD' or plot == 'BME':
+ plt.plot(x_idx, SeqValues, marker=markers[idx], color=colors[idx], label=name)
+ else:
+ plt.semilogy(x_idx, SeqValues, marker=markers[idx], color=colors[idx], label=name)
+
plt.xlabel('Number of runs')
plt.ylabel(plot)
plt.legend()
diff --git a/BayesValidRox/surrogate_models/surrogate_models.py b/BayesValidRox/surrogate_models/surrogate_models.py
index ba5044ccf9d6a310e1ac8e99c9273a68cc0cea4c..1f74984535f141d4593dcf57e15bc3e04ab61f54 100644
--- a/BayesValidRox/surrogate_models/surrogate_models.py
+++ b/BayesValidRox/surrogate_models/surrogate_models.py
@@ -69,6 +69,8 @@ class aPCE:
self.SeqModifiedLOO = []
self.SeqBME = {}
self.SeqKLD = {}
+ self.seqRMSEMean = {}
+ self.seqRMSEStd = {}
self.Observations = []
self.Likelihoods = None
@@ -1212,11 +1214,11 @@ class aPCE:
# Mean = c_0
PCEMean[idx] = coeffs[0]
# Std = sum(coeffs[1:]^2)
- PCEStd[idx] = np.sum(np.square(coeffs[1:]))
+ PCEStd[idx] = np.sqrt(np.sum(np.square(coeffs[1:])))
# Compute the error between mean and std of PCEModel and OrigModel
- MSE_Mean = mean_squared_error(df_MCReference['mean'], PCEMean)
- MSE_std = mean_squared_error(df_MCReference['std'], PCEStd)
+ MSE_Mean = np.sqrt(mean_squared_error(df_MCReference['mean'], PCEMean))
+ MSE_std = np.sqrt(mean_squared_error(df_MCReference['std'], PCEStd))
PCEMeans[Outkey] = PCEMean
PCEStds[Outkey] = PCEStd
@@ -1283,7 +1285,10 @@ class aPCE:
print('Posterior snapshot (initial) is being plotted...')
self.posteriorPlot(initPosterior, MAP, parNames, 'SeqPosterior_init')
-
+ # Check the convergence of the Mean&Std
+ if self.ModelObj.MCReference:
+ initRMSEMean, initRMSEStd = self.error_Mean_Std()
+ print("Initial Mean and Std error: %s, %s"%(initRMSEMean, initRMSEStd))
# Read the initial experimental design
from copy import deepcopy
@@ -1300,6 +1305,8 @@ class aPCE:
self.SeqModifiedLOO = {}
self.SeqBME = {}
+ self.seqRMSEMean = {}
+ self.seqRMSEStd = {}
for UtilityFunction in UtilityFunctions:
@@ -1320,6 +1327,9 @@ class aPCE:
SeqModifiedLOO = np.array(ModifiedLOO)
SeqBME = np.array([initBME])
SeqKLD = np.array([initKLD])
+ if self.ModelObj.MCReference:
+ seqRMSEMean = np.array([initRMSEMean])
+ seqRMSEStd = np.array([initRMSEStd])
postcnt = 1
@@ -1395,20 +1405,33 @@ class aPCE:
postcnt += 1
+ # Check the convergence of the Mean&Std
+ if self.ModelObj.MCReference:
+ print('\n')
+ RMSE_Mean, RMSE_std = self.error_Mean_Std()
+ print("Updated Mean and Std error: %s, %s"%(RMSE_Mean, RMSE_std))
+ print('\n')
+
print('\n')
print("Updated BME:", BME)
print("Updated KLD:", KLD)
print('\n')
- # Store the updated BME & KLD
+ # Store the updated BME & KLD &
SeqBME = np.vstack((SeqBME, BME))
SeqKLD = np.vstack((SeqKLD, KLD))
+ if self.ModelObj.MCReference:
+ seqRMSEMean = np.vstack((seqRMSEMean, RMSE_Mean))
+ seqRMSEStd = np.vstack((seqRMSEStd, RMSE_std))
# Store updated ModifiedLOO and BME in dictonary
self.SeqModifiedLOO[UtilityFunction] = SeqModifiedLOO
self.SeqBME[UtilityFunction] = SeqBME
self.SeqKLD[UtilityFunction] = SeqKLD
-
+ if self.ModelObj.MCReference:
+ self.seqRMSEMean[UtilityFunction] = seqRMSEMean
+ self.seqRMSEStd[UtilityFunction] = seqRMSEStd
+
return PCEModel
#--------------------------------------------------------------------------------------------------------
@@ -1432,7 +1455,6 @@ class aPCE:
# TODO: Loop over outputs
OutputName = Model.Output.Names
-
# Read the initial experimental design
from copy import deepcopy
Xinit = PCEModel.ExpDesign.X
@@ -1446,7 +1468,15 @@ class aPCE:
Scores = [item for sublist in Scores_all for item in sublist]
initModifiedLOO = [1-score for score in Scores]
+ # Check the initial EMSE error of the Mean&Std
+ if self.ModelObj.MCReference:
+ initRMSEMean, initRMSEStd = self.error_Mean_Std()
+ print("Initial Mean and Std error: %s, %s"%(initRMSEMean, initRMSEStd))
+
self.SeqModifiedLOO = {}
+ self.seqRMSEMean = {}
+ self.seqRMSEStd = {}
+
for UtilityFunction in UtilityFunctions:
print ('>>>> UtilityFunction: %s<<<<'%UtilityFunction)
@@ -1463,11 +1493,10 @@ class aPCE:
print("Initial ModifiedLOO:", initModifiedLOO)
ModifiedLOO = initModifiedLOO
SeqModifiedLOO = np.array(ModifiedLOO)
-
- # Check the convergence of the Mean&Std
if self.ModelObj.MCReference:
- MSE_Mean, MSE_std = self.error_Mean_Std()
- print("Initial Mean and Std error: %s, %s"%(MSE_Mean, MSE_std))
+ seqRMSEMean = np.array([initRMSEMean])
+ seqRMSEStd = np.array([initRMSEStd])
+
# Start Sequential Experimental Design
while (TotalNSamples < MaxNSamples) and any(LOO > ModifiedLOOThreshold for LOO in ModifiedLOO):
@@ -1530,13 +1559,18 @@ class aPCE:
# Check the convergence of the Mean&Std
if self.ModelObj.MCReference:
print('\n')
- MSE_Mean, MSE_std = self.error_Mean_Std()
- print("Updated Mean and Std error: %s, %s"%(MSE_Mean, MSE_std))
+ RMSE_Mean, RMSE_std = self.error_Mean_Std()
+ print("Updated Mean and Std error: %s, %s"%(RMSE_Mean, RMSE_std))
print('\n')
+ seqRMSEMean = np.vstack((seqRMSEMean, RMSE_Mean))
+ seqRMSEStd = np.vstack((seqRMSEStd, RMSE_std))
# Store updated ModifiedLOO and BME in dictonary
self.SeqModifiedLOO[UtilityFunction] = SeqModifiedLOO
-
+ if self.ModelObj.MCReference:
+ self.seqRMSEMean[UtilityFunction] = seqRMSEMean
+ self.seqRMSEStd[UtilityFunction] = seqRMSEStd
+
return PCEModel
#--------------------------------------------------------------------------------------------------------
diff --git a/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunction_Test.py b/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunction_Test.py
index 400359bdbfe538d1a688df5a0a02dc94d3a0e183..03dbdb21312541f11123b9e977e1fac2f3535ce9 100644
--- a/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunction_Test.py
+++ b/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunction_Test.py
@@ -102,14 +102,14 @@ if __name__ == "__main__":
# hypercube sampling of the input model or user-defined values of X and/or Y:
MetaModelOpts.addExpDesign()
- NrofInitSamples = 5
+ NrofInitSamples = 15
MetaModelOpts.ExpDesign.NrSamples = NrofInitSamples
- MetaModelOpts.ExpDesign.SamplingMethod = 'MC' # 1)MC 2)LHS 3)PCM 4)LSCM 5)user
+ MetaModelOpts.ExpDesign.SamplingMethod = 'LHS' # 1)MC 2)LHS 3)PCM 4)LSCM 5)user
MetaModelOpts.ExpDesign.Method = 'sequential' # 1) normal 2) sequential
# Sequential experimental design (needed only for sequential ExpDesign)
MetaModelOpts.ExpDesign.MaxNSamples = 50
- MetaModelOpts.ExpDesign.ModifiedLOOThreshold = 1e-3
+ MetaModelOpts.ExpDesign.ModifiedLOOThreshold = 1e-5
# Plot the posterior snapshots for SeqDesign
# MetaModelOpts.ExpDesign.PostSnapshot = True
diff --git a/BayesValidRox/tests/Co2BenchmarkTest/Test_CO2Benchmark.py b/BayesValidRox/tests/Co2BenchmarkTest/Test_CO2Benchmark.py
index 49c525e0bdfc899dd05c3551c95d99d270946e6c..27f39d08efc0014b4dff5baa7dd1bccced6e08f1 100644
--- a/BayesValidRox/tests/Co2BenchmarkTest/Test_CO2Benchmark.py
+++ b/BayesValidRox/tests/Co2BenchmarkTest/Test_CO2Benchmark.py
@@ -203,13 +203,15 @@ if __name__ == "__main__":
# ----- Define the discrepancy model -------
obsData = pd.read_csv(Model.MeasurementFile)['CO2 saturation[-]']
+ error = 0.02*obsData.to_numpy().reshape(1,-1)[0]
+ error[0] = 1e-5
# (Option A)
# Do nothing
# (Option B) # 0.2% of the data
DiscrepancyOpts = Discrepancy('')
DiscrepancyOpts.Type = 'Gaussian'
- DiscrepancyOpts.Parameters = 0.02*obsData.to_numpy().reshape(1,-1)[0]
+ DiscrepancyOpts.Parameters = error
BayesOpts.Discrepancy = DiscrepancyOpts
# (Option C)