From 8c40ea3e6fe2e343b39681a6e8cf2dc7ed72272c Mon Sep 17 00:00:00 2001
From: Farid Mohammadi <farid.mohammadi@iws.uni-stuttgart.de>
Date: Sat, 28 Mar 2020 19:54:28 +0100
Subject: [PATCH] [surrogate] modified FastARD.
---
.../PostProcessing/PostProcessing.py | 32 +++--
.../surrogate_models/RegressionFastARD.py | 55 ++++++--
.../surrogate_models/surrogate_models.py | 133 ++++++++++++------
.../AnalyticalFunctionComplex_Test.py | 16 +--
.../AnalyticalFunction_Test.py | 20 +--
.../Co2BenchmarkTest/Test_CO2Benchmark.py | 12 +-
6 files changed, 179 insertions(+), 89 deletions(-)
diff --git a/BayesValidRox/PostProcessing/PostProcessing.py b/BayesValidRox/PostProcessing/PostProcessing.py
index 95d4ddff2..5e597f6c6 100644
--- a/BayesValidRox/PostProcessing/PostProcessing.py
+++ b/BayesValidRox/PostProcessing/PostProcessing.py
@@ -143,21 +143,27 @@ class PostProcessing:
idx = 0
for Inkey, InIdxValues in ValuesDict.items():
- PolynomialDegrees = PCEModel.BasisDict[Outkey][Inkey]
- clf_poly = PCEModel.clf_poly[Outkey][Inkey]
+ if PCEModel.RegMethod == 'GPR':
+ gp = PCEModel.clf_poly[Outkey][Inkey]
+ y_mean, y_std = gp.predict(Samples, return_std=True)
- PSI_Val = PCEModel.PCE_create_Psi(PolynomialDegrees, univ_p_val)
-
- # Perdiction
- try:
- # with error bar
- y_mean, y_std = clf_poly.predict(PSI_Val, return_std=True)
- except:
- # without error bar
- coeffs = PCEModel.CoeffsDict[Outkey][Inkey]
- y_mean = np.dot(PSI_Val, coeffs)
- y_std = np.zeros(shape=y_mean.shape)
+ else:
+ PolynomialDegrees = PCEModel.BasisDict[Outkey][Inkey]
+
+ clf_poly = PCEModel.clf_poly[Outkey][Inkey]
+
+ PSI_Val = PCEModel.PCE_create_Psi(PolynomialDegrees, univ_p_val)
+
+ # Perdiction
+ try:
+ # with error bar
+ y_mean, y_std = clf_poly.predict(PSI_Val, return_std=True)
+ except:
+ # without error bar
+ coeffs = PCEModel.CoeffsDict[Outkey][Inkey]
+ y_mean = np.dot(PSI_Val, coeffs)
+ y_std = np.zeros(shape=y_mean.shape)
PCEOutputs_mean[:, idx] = y_mean
PCEOutputs_std[:, idx] = y_std
diff --git a/BayesValidRox/surrogate_models/RegressionFastARD.py b/BayesValidRox/surrogate_models/RegressionFastARD.py
index b53aea525..cad82103f 100755
--- a/BayesValidRox/surrogate_models/RegressionFastARD.py
+++ b/BayesValidRox/surrogate_models/RegressionFastARD.py
@@ -12,6 +12,7 @@ import warnings
from sklearn.utils import check_X_y,check_array,as_float_array
from scipy.linalg import pinvh
+
def update_precisions(Q,S,q,s,A,active,tol,n_samples,clf_bias):
'''
Selects one feature to be added/recomputed/deleted to model based on
@@ -42,7 +43,7 @@ def update_precisions(Q,S,q,s,A,active,tol,n_samples,clf_bias):
deltaL[recompute] = Qrec**2 / (Srec + 1. / delta_alpha) - np.log(1 + Srec*delta_alpha)
deltaL[delete] = Qdel**2 / (Sdel - Adel) - np.log(1 - Sdel / Adel)
deltaL = deltaL / n_samples
-
+
# find feature which caused largest change in likelihood
feature_index = np.argmax(deltaL)
@@ -106,8 +107,12 @@ class RegressionFastARD():
copy_X : boolean, optional (DEFAULT = True)
If True, X will be copied; else, it may be overwritten.
-
- verbose : boolean, optional (DEFAULT = True)
+
+ compute_score : bool, default=False
+ If True, compute the log marginal likelihood at each iteration of the
+ optimization.
+
+ verbose : boolean, optional (DEFAULT = FALSE)
Verbose mode when fitting the model
@@ -128,7 +133,10 @@ class RegressionFastARD():
sigma_ : array, shape = (n_features, n_features)
estimated covariance matrix of the weights, computed only
for non-zero coefficients
-
+
+ scores_ : array-like of shape (n_iter_+1,)
+ If computed_score is True, value of the log marginal likelihood (to be
+ maximized) at each iteration of the optimization.
References
----------
@@ -140,12 +148,13 @@ class RegressionFastARD():
'''
def __init__( self, n_iter = 300, tol = 1e-3, fit_intercept = True,
- copy_X = True, verbose = False):
+ copy_X = True, compute_score=False, verbose = False):
self.n_iter = n_iter
self.tol = tol
self.scores_ = list()
self.fit_intercept = fit_intercept
self.copy_X = copy_X
+ self.compute_score = compute_score
self.verbose = verbose
@@ -206,7 +215,7 @@ class RegressionFastARD():
# in case of almost perfect multicollinearity between some features
# start from feature 0
- if np.sum( XXd - X_mean**2 < np.finfo(np.float32).eps ) > 0:
+ if np.sum( XXd - X_mean**2 < np.finfo(np.float32).eps ) >= 0:
A[0] = np.finfo(np.float16).eps
active[0] = True
else:
@@ -217,6 +226,7 @@ class RegressionFastARD():
A[start] = XXd[start]/( proj[start] - var_y)
warning_flag = 0
+ scores_ = []
for i in range(self.n_iter):
XXa = XX[active,:][:,active]
XYa = XY[active]
@@ -247,6 +257,10 @@ class RegressionFastARD():
# update precision parameters of coefficients
A,converged = update_precisions(Q,S,q,s,A,active,self.tol,
n_samples,False)
+
+ if self.compute_score:
+ scores_.append(self.log_marginal_likelihood(XXa,XYa, Aa, beta))
+
if self.verbose:
print(('Iteration: {0}, number of features '
'in the model: {1}').format(i,np.sum(active)))
@@ -254,7 +268,8 @@ class RegressionFastARD():
if converged and self.verbose:
print('Algorithm converged !')
break
-
+
+
# after last update of alpha & beta update parameters
# of posterior distribution
XXa,XYa,Aa = XX[active,:][:,active],XY[active],A[active]
@@ -265,11 +280,27 @@ class RegressionFastARD():
self.active_ = active
self.lambda_ = A
self.alpha_ = beta
+ if self.compute_score:
+ self.scores_ = np.array(scores_)
#self._set_intercept(X_mean,y_mean,X_std)
return self
+
+ def log_marginal_likelihood(self, XXa,XYa, Aa, beta):
+ """Computes the log of the marginal likelihood."""
+ N, M = XXa.shape
+ A = np.diag(Aa)
+
+ Mn, sigma_, cholesky = self._posterior_dist(Aa,beta,XXa,XYa,full_covar=True)
+
+ C = sigma_ + np.dot(np.dot(XXa.T, np.linalg.pinv(A)), XXa)
+ score = np.dot(np.dot(XYa.T, np.linalg.pinv(C)), XYa) + \
+ np.log(np.linalg.det(C)) + \
+ N * np.log(2 * np.pi)
+
+ return -0.5 * score
+
- #def predict_dist(self,X):
def predict(self,X, return_std=False):
'''
Computes predictive distribution for test set.
@@ -292,7 +323,7 @@ class RegressionFastARD():
var_hat: numpy array of size (n_samples_test,)
Variance of predictive distribution
'''
- y_hat = np.dot(X[:,self.active_],self.coef_[self.active_]) #+ self.intercept_ #self._decision_function(X)
+ y_hat = np.dot(X[:,self.active_],self.coef_[self.active_])
if return_std:
var_hat = 1./self.alpha_
var_hat += np.sum( np.dot(X[:,self.active_],self.sigma_) * X[:,self.active_], axis = 1)
@@ -316,9 +347,9 @@ class RegressionFastARD():
# find posterior mean : R*R.T*mean = beta*X.T*Y
# solve(R*z = beta*X.T*Y) => find z => solve(R.T*mean = z) => find mean
R = np.linalg.cholesky(Sinv)
- Z = solve_triangular(R,beta*XY, check_finite=False, lower = True)
- Mn = solve_triangular(R.T,Z, check_finite=False, lower = False)
-
+ Z = solve_triangular(R,beta*XY, check_finite=True, lower = True)
+ Mn = solve_triangular(R.T,Z, check_finite=True, lower = False)
+
# invert lower triangular matrix from cholesky decomposition
Ri = solve_triangular(R,np.eye(A.shape[0]), check_finite=False, lower=True)
if full_covar:
diff --git a/BayesValidRox/surrogate_models/surrogate_models.py b/BayesValidRox/surrogate_models/surrogate_models.py
index 4ed089fe3..313f82f27 100644
--- a/BayesValidRox/surrogate_models/surrogate_models.py
+++ b/BayesValidRox/surrogate_models/surrogate_models.py
@@ -427,6 +427,7 @@ class aPCE:
"""
clf_poly = []
+ compute_score = True if self.DisplayFlag else False
# inverse of the observed variance of the data
Lambda = 1 / np.var(ModelOutput) if np.var(ModelOutput) !=0 else 1e-6
@@ -435,17 +436,17 @@ class aPCE:
if self.RegMethod != 'OLS':
if self.RegMethod == 'BRR':
clf_poly = linear_model.BayesianRidge(n_iter=1000, tol=1e-7,
- fit_intercept = False,
- compute_score=True,
+ fit_intercept=False,
+ compute_score=compute_score,
lambda_1=Lambda, lambda_2=Lambda)
-
+
elif self.RegMethod == 'ARD':
# clf_poly = ARDRegression(fit_intercept = False,compute_score=True,
# n_iter=1000, tol= 0.001)
# alpha_1=1e-6, alpha_2=1e-6,
# lambda_1=Lambda, lambda_2=Lambda)
- clf_poly = RegressionFastARD(fit_intercept = False)
-
+ clf_poly = RegressionFastARD(fit_intercept=False,compute_score=compute_score,
+ tol= 0.001)
elif self.RegMethod == 'LARS':
clf_poly = linear_model.Lars(fit_intercept=False)
@@ -542,28 +543,29 @@ class aPCE:
qNormScores = -np.inf*np.ones(nqnorms);
scores_BRR = -np.inf*np.ones(self.DegreeArray.shape[0])
+ # Extract the univariate polynomials on ExpDesign
+ univ_p_val = self.univ_p_val
+
for degIdx, deg in enumerate(self.DegreeArray):
for qidx, q in enumerate(qnorm):
# Generate the polynomial basis indices
BasisIndices = self.PolyBasisIndices(degree = deg, q=q)
-
- # Evaluate the univariate polynomials on ExpDesig
- univ_p_val = self.univ_basis_vals(CollocationPoints)
-
# Assemble the Psi matrix
Psi = self.PCE_create_Psi(BasisIndices,univ_p_val)
# ---- Calulate the cofficients of aPCE ----------------------
- PolynomialDegrees_Sparse, PSI_Sparse, Coeffs, clf_poly = self.RS_Builder(Psi, ModelOutput,BasisIndices)
+ sparseBasisIndices, sparsePSI, Coeffs, clf_poly = self.RS_Builder(Psi,ModelOutput,BasisIndices)
+
+
AllCoeffs[str(degIdx+1)] = Coeffs
- AllIndices_Sparse[str(degIdx+1)] = PolynomialDegrees_Sparse
+ AllIndices_Sparse[str(degIdx+1)] = sparseBasisIndices
Allclf_poly[str(degIdx+1)] = clf_poly
AllnTerms[str(degIdx+1)] = BasisIndices.shape[0]
# Calculate and save the score of LOOCV
- score = self.CorrectedLeaveoutError(PSI_Sparse, Coeffs, ModelOutput, clf_poly)
+ score = self.CorrectedLeaveoutError(sparsePSI, Coeffs, ModelOutput, clf_poly)
qNormScores[qidx] = score
# EarlyStop check
@@ -575,7 +577,10 @@ class aPCE:
if sum(deltas[-n_checks_qNorm+1:]) == 2:
# stop the q-norm loop here
break
-
+ if np.var(ModelOutput) == 0:
+ break
+
+
# check the direction of the error (on average):
# if it increases consistently stop the iterations
if len(scores[scores!=-np.inf]) > n_checks_degree:
@@ -588,6 +593,9 @@ class aPCE:
# Store the score in the scores list
scores[degIdx] = np.max(qNormScores)
+ # Check only one degree
+ if np.var(ModelOutput) == 0:
+ break
# ------------------ Summary of results ------------------
# Select the one with the best score and save the necessary outputs
@@ -606,7 +614,7 @@ class aPCE:
print('\n'+'-'*50)
# ------------------ Summary of results ------------------
- if self.DisplayFlag == True:
+ if self.DisplayFlag:
print('='*50)
print(' '* 10 +' Summary of results ')
print('='*50)
@@ -716,7 +724,7 @@ class aPCE:
varY = np.var(ModelOutputs)
- if varY ==0:
+ if varY == 0:
normEmpErr = 0
ErrLoo = 0
else:
@@ -732,10 +740,12 @@ class aPCE:
trM = np.trace(M)
if trM < 0 or abs(trM) > 1e6:
trM = np.trace(np.linalg.pinv(np.dot(Psi.T,Psi)))
-
+
+ # Over-determined system of Equation
if N > P:
T_factor = N/(N-P) * (1 + trM)
+ # Under-determined system of Equation
else:
T_factor = np.inf
@@ -757,7 +767,36 @@ class aPCE:
value = self[item] = type(self)()
return value
#--------------------------------------------------------------------------------------------------------
-
+ def GPR(self, X, y, varIdx):
+
+ from sklearn.gaussian_process import GaussianProcessRegressor
+ from sklearn.gaussian_process.kernels import (RBF, Matern)
+
+ kernels = [1.0 * RBF(length_scale=1.0, length_scale_bounds=(1e-1, 10.0)),
+ 1.0 * Matern(length_scale=1.0, length_scale_bounds=(1e-1, 10.0),
+ nu=1.5)]
+
+ for kernel in kernels:
+ # Instantiate a Gaussian Process model
+ #kernel = C(1.0, (1e-3, 1e3)) * RBF(10, (1e-2, 1e2))
+ #kernel = RBF(20, (1e-5, 1e5))
+ gp = GaussianProcessRegressor(kernel=kernel, n_restarts_optimizer=9)
+
+ # Fit to data using Maximum Likelihood Estimation of the parameters
+ gp.fit(X, y)
+
+ # Compute score
+ residual = gp.predict(X) - y
+ normEmpErr = np.mean(residual**2)/np.var(y)
+
+ print('\n'+'-'*50)
+ print('The estimation of GPR coefficients converged at for output variable %s.'%(varIdx+1))
+ print('Final normalized emperical error estimate: %s'%(normEmpErr))
+ print('\n'+'-'*50)
+
+ return gp, gp.alpha_, 1-normEmpErr
+
+ #--------------------------------------------------------------------------------------------------------
def create_PCE_normDesign(self, Model, OptDesignFlag=False):
"""
This function loops over the outputs and each time step/point to compute the PCE
@@ -782,6 +821,10 @@ class aPCE:
CollocationPoints = self.OptimalCollocationPointsBase
OutputDict = self.ModelOutputDict.copy()
+ # Evaluate the univariate polynomials on ExpDesign
+ self.univ_p_val = self.univ_basis_vals(CollocationPoints)
+
+
if 'x_values' in OutputDict: del OutputDict['x_values']
for key , OutputMatrix in OutputDict.items():
@@ -789,12 +832,16 @@ class aPCE:
if not OptDesignFlag: print(">>>> Calculating the PCE coefficients of %s with %s<<<<<<"%(key, self.RegMethod))
for idx in range(OutputMatrix.shape[1]):
- # Only update the coeffs in the searching alg. of SeqExpDesign
- if OptDesignFlag:
- BasisIndices = self.BasisDict[key]["y_"+str(idx+1)]
- self.CoeffsDict[key]["y_"+str(idx+1)] , self.BasisDict[key]["y_"+str(idx+1)], self.ScoresDict[key]["y_"+str(idx+1)], self.clf_poly[key]["y_"+str(idx+1)] = self.CoeffsUpdateaPCE(CollocationPoints, OutputMatrix[:,idx], BasisIndices)
+ if self.RegMethod== 'GPR': # TODO: Change RegMethod to metamodel type
+ self.clf_poly[key]["y_"+str(idx+1)], self.CoeffsDict[key]["y_"+str(idx+1)], self.ScoresDict[key]["y_"+str(idx+1)] = self.GPR(CollocationPoints, OutputMatrix[:,idx], idx)
else:
- self.CoeffsDict[key]["y_"+str(idx+1)] , self.BasisDict[key]["y_"+str(idx+1)], self.ScoresDict[key]["y_"+str(idx+1)], self.clf_poly[key]["y_"+str(idx+1)] = self.adaptiveSparseaPCE(CollocationPoints, OutputMatrix[:,idx],idx)
+
+ # Only update the coeffs in the searching alg. of SeqExpDesign
+ if OptDesignFlag:
+ BasisIndices = self.BasisDict[key]["y_"+str(idx+1)]
+ self.CoeffsDict[key]["y_"+str(idx+1)] , self.BasisDict[key]["y_"+str(idx+1)], self.ScoresDict[key]["y_"+str(idx+1)], self.clf_poly[key]["y_"+str(idx+1)] = self.CoeffsUpdateaPCE(CollocationPoints, OutputMatrix[:,idx], BasisIndices)
+ else:
+ self.CoeffsDict[key]["y_"+str(idx+1)] , self.BasisDict[key]["y_"+str(idx+1)], self.ScoresDict[key]["y_"+str(idx+1)], self.clf_poly[key]["y_"+str(idx+1)] = self.adaptiveSparseaPCE(CollocationPoints, OutputMatrix[:,idx],idx)
return self
#--------------------------------------------------------------------------------------------------------
@@ -914,12 +961,12 @@ class aPCE:
cov = np.zeros((means.shape[0], means.shape[0]), float)
np.fill_diagonal(cov, stds**2)
- multiNormalDensity = stats.multivariate_normal(mean=means, cov=cov)
+ multiNormalDensity = stats.multivariate_normal(mean=means, cov=cov, allow_singular=True)
Y_PC_MC[key] = multiNormalDensity.rvs(MCsize)
logPriorLikelihoods = np.multiply(logPriorLikelihoods, multiNormalDensity.logpdf(Y_PC_MC[key]))
std_PC_MC[key] = np.zeros((MCsize, means.shape[0]))
- # Likelihood computation (Comparison of data
+ # Likelihood computation (Comparison of data
# and simulation results via PCE with candidate design)
Likelihoods = self.normpdf(Y_PC_MC,std_PC_MC, ObservationData, sigma2Dict)
@@ -927,7 +974,7 @@ class aPCE:
# Check the Effective Sample Size (1<ESS<MCsize)
ESS = 1 / np.sum(np.square(Likelihoods/np.sum(Likelihoods)))
-
+
# Enlarge sample size if it doesn't fulfill the criteria
if ((ESS > MCsize) or (ESS < 1)):
MCsize = MCsize * 10
@@ -1517,20 +1564,27 @@ class aPCE:
for Inkey, InIdxValues in ValuesDict.items():
idx = int(Inkey.split('_')[1]) - 1
- PolynomialDegrees = self.BasisDict[Outkey][Inkey]
- PSI_Val = self.PCE_create_Psi(PolynomialDegrees, univ_p_val)
- # Perdiction
- try: # with error bar
- clf_poly = self.clf_poly[Outkey][Inkey]
- y_mean, y_std = clf_poly.predict(PSI_Val, return_std=True)
-
- PCEOutputs_mean[:, idx] = y_mean
- PCEOutputs_std[:, idx] = y_std
+ if self.RegMethod == 'GPR':
+ gp = self.clf_poly[Outkey][Inkey]
+ PCEOutputs_mean[:, idx], PCEOutputs_std[:, idx] = gp.predict(Samples, return_std=True)
+
+ else:
+
+ PolynomialDegrees = self.BasisDict[Outkey][Inkey]
+ PSI_Val = self.PCE_create_Psi(PolynomialDegrees, univ_p_val)
- except: # without error bar
- coeffs = self.CoeffsDict[Outkey][Inkey]
- PCEOutputs_mean[:, idx] = np.dot(PSI_Val, coeffs)
+ # Perdiction
+ try: # with error bar
+ clf_poly = self.clf_poly[Outkey][Inkey]
+ PCEOutputs_mean[:, idx], PCEOutputs_std[:, idx] = clf_poly.predict(PSI_Val, return_std=True)
+
+ # PCEOutputs_mean[:, idx] = y_mean
+ # PCEOutputs_std[:, idx] = y_std
+
+ except: # without error bar
+ coeffs = self.CoeffsDict[Outkey][Inkey]
+ PCEOutputs_mean[:, idx] = np.dot(PSI_Val, coeffs)
meanPCEOutputs[Outkey] = PCEOutputs_mean
stdPCEOutputs[Outkey] = PCEOutputs_std
@@ -1570,7 +1624,7 @@ class aPCE:
covMatrix = covMatrix + covMatrix_PCE
# Compute likelihood
- self.Likelihoods = stats.multivariate_normal.pdf(TotalOutputs, mean=TotalData, cov=covMatrix)
+ self.Likelihoods = stats.multivariate_normal.pdf(TotalOutputs, mean=TotalData, cov=covMatrix, allow_singular=True)
return self.Likelihoods
@@ -2023,7 +2077,6 @@ class aPCE:
print("\n")
# Evaluate the full model response at the new sample points:
- #Ynew = Model.Run_Model(Xnew, TotalNSamples)
Ynew, _ = Model.Run_Model_Parallel(Xnew, prevRun_No=TotalNSamples)
TotalNSamples += Xnew.shape[0]
@@ -2054,7 +2107,7 @@ class aPCE:
prevPCEModel = PCEModel
PCEModel = PCEModel.create_PCE_normDesign(Model)
- # TODO: Compute the mean squre error
+ # Compute the validation error
if len(self.validModelRuns) != 0:
validError = PCEModel.validError_()
Scores_all=[]
diff --git a/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunctionComplex_Test.py b/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunctionComplex_Test.py
index 7b13e7d81..2c92917ea 100644
--- a/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunctionComplex_Test.py
+++ b/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunctionComplex_Test.py
@@ -98,7 +98,7 @@ if __name__ == "__main__":
# error (or the highest score=1-LOO)estimator is chosen as the final
# metamodel.
MetaModelOpts.MinPceDegree = 1 #2
- MetaModelOpts.MaxPceDegree = 8 #8
+ MetaModelOpts.MaxPceDegree = 10 #8
# q-quasi-norm 0<q<1 (default=1)
MetaModelOpts.q = 1.0
@@ -107,9 +107,9 @@ if __name__ == "__main__":
# the PCE coefficients calculation:
# 1)OLS: Ordinary Least Square 2)BRR: Bayesian Ridge Regression
# 3)LARS: Least angle regression 4)ARD: Bayesian ARD Regression
- # 5)SGDRegressor: Stochastic gradient descent learning
+ # 5)SGDR: Stochastic gradient descent learning
- MetaModelOpts.RegMethod = 'BRR'
+ MetaModelOpts.RegMethod = 'ARD'
# Print summary of the regression results
#MetaModelOpts.DisplayFlag = True
@@ -121,7 +121,7 @@ if __name__ == "__main__":
# One-shot (normal) or Sequential Adaptive (sequential) Design
MetaModelOpts.ExpDesign.Method = 'sequential'
- NrofInitSamples = 100
+ NrofInitSamples = 25
MetaModelOpts.ExpDesign.NrSamples = NrofInitSamples
# Sampling methods
@@ -130,9 +130,9 @@ if __name__ == "__main__":
MetaModelOpts.ExpDesign.SamplingMethod = 'latin_hypercube'
# Sequential experimental design (needed only for sequential ExpDesign)
- MetaModelOpts.ExpDesign.NrofNewSample = 5
- MetaModelOpts.ExpDesign.MaxNSamples = 200
- MetaModelOpts.ExpDesign.ModifiedLOOThreshold = 1e-6
+ MetaModelOpts.ExpDesign.NrofNewSample = 1
+ MetaModelOpts.ExpDesign.MaxNSamples = 30
+ MetaModelOpts.ExpDesign.ModifiedLOOThreshold = 1e-10
# Defining the measurement error, if it's known a priori
obsData = pd.DataFrame(Model.Observations, columns=Model.Output.Names)
@@ -152,7 +152,7 @@ if __name__ == "__main__":
# ------- Sequential Design configuration --------
# ------------------------------------------------
# 1) 'None' 2) 'epsilon-decreasing'
- MetaModelOpts.ExpDesign.TradeOffScheme = 'epsilon-decreasing'
+ MetaModelOpts.ExpDesign.TradeOffScheme = 'None'
#MetaModelOpts.ExpDesign.nReprications = 2
# -------- Exploration ------
#1)'Voronoi' 2)'random' 3)'latin_hypercube' 4)'dual annealing'
diff --git a/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunction_Test.py b/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunction_Test.py
index 1679cf304..4441dc6fe 100644
--- a/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunction_Test.py
+++ b/BayesValidRox/tests/AnalyticalFunction/AnalyticalFunction_Test.py
@@ -98,7 +98,7 @@ if __name__ == "__main__":
# 3)LARS: Least angle regression 4)ARD: Bayesian ARD Regression
# 5)SGDRegressor: Stochastic gradient descent learning
- MetaModelOpts.RegMethod = 'ARD'
+ MetaModelOpts.RegMethod = 'BRR'
# Print summary of the regression results
# MetaModelOpts.DisplayFlag = True
@@ -110,7 +110,7 @@ if __name__ == "__main__":
# One-shot (normal) or Sequential Adaptive (sequential) Design
MetaModelOpts.ExpDesign.Method = 'sequential'
- NrofInitSamples = 15
+ NrofInitSamples = 10
MetaModelOpts.ExpDesign.NrSamples = NrofInitSamples
# Sampling methods
@@ -120,8 +120,8 @@ if __name__ == "__main__":
# Sequential experimental design (needed only for sequential ExpDesign)
MetaModelOpts.ExpDesign.NrofNewSample = 1
- MetaModelOpts.ExpDesign.MaxNSamples = 50
- MetaModelOpts.ExpDesign.ModifiedLOOThreshold = 1e-6
+ MetaModelOpts.ExpDesign.MaxNSamples = 100
+ MetaModelOpts.ExpDesign.ModifiedLOOThreshold = 1e-16
# Defining the measurement error, if it's known a priori
obsData = pd.DataFrame(Model.Observations, columns=Model.Output.Names)
@@ -144,34 +144,34 @@ if __name__ == "__main__":
#MetaModelOpts.ExpDesign.nReprications = 5
# -------- Exploration ------
#1)'Voronoi' 2)'random' 3)'latin_hypercube' 4)'dual annealing'
- MetaModelOpts.ExpDesign.ExploreMethod = 'Voronoi'
+ MetaModelOpts.ExpDesign.ExploreMethod = 'random'
# Use when 'dual annealing' chosen
MetaModelOpts.ExpDesign.MaxFunItr = 200
# Use when 'Voronoi' or 'MC' or 'LHS' chosen
- MetaModelOpts.ExpDesign.NCandidate = 5 #200
+ MetaModelOpts.ExpDesign.NCandidate = 100
MetaModelOpts.ExpDesign.NrofCandGroups = 4
# -------- Exploitation ------
# 1)'BayesOptDesign' 2)'VarOptDesign' 3)'alphabetic' 4)'Space-filling'
- MetaModelOpts.ExpDesign.ExploitMethod = 'BayesOptDesign'
+ MetaModelOpts.ExpDesign.ExploitMethod = 'VarOptDesign'
# BayesOptDesign -> when data is available
# 1)DKL (Kullback-Leibler Divergence) 2)DPP (D-Posterior-percision)
# 3)APP (A-Posterior-percision)
# 1)DKL (Kullback-Leibler Divergence) 2)BME (Bayesian model evidence)
# 3)infEntropy (Entropy-based information gain)
- MetaModelOpts.ExpDesign.UtilityFunction = 'DKL' #['DKL', 'DPP']
+ # MetaModelOpts.ExpDesign.UtilityFunction = 'BME' #['DKL', 'DPP']
# VarBasedOptDesign -> when data is not available
# Only with Vornoi >>> 1)Entropy 2)EIGF, 3)ALM, 4)LOOCV
- #MetaModelOpts.ExpDesign.UtilityFunction = 'Entropy' #['EIGF', 'Entropy', 'LOOCV']
+ MetaModelOpts.ExpDesign.UtilityFunction = 'Entropy' #['EIGF', 'Entropy', 'LOOCV']
# alphabetic
# 1)D-Opt (D-Optimality) 2)A-Opt (A-Optimality)
# 3)K-Opt (K-Optimality)
- #MetaModelOpts.ExpDesign.UtilityFunction = 'D-Opt' #['D-Opt', 'A-Opt', 'K-Opt']
+ # MetaModelOpts.ExpDesign.UtilityFunction = 'D-Opt' #['D-Opt', 'A-Opt', 'K-Opt']
# For colculation of validation error for SeqDesign
MetaModelOpts.validModelRuns = np.load("origModelOutput.npy")
diff --git a/BayesValidRox/tests/Co2BenchmarkTest/Test_CO2Benchmark.py b/BayesValidRox/tests/Co2BenchmarkTest/Test_CO2Benchmark.py
index fa6eede4d..1b303b168 100644
--- a/BayesValidRox/tests/Co2BenchmarkTest/Test_CO2Benchmark.py
+++ b/BayesValidRox/tests/Co2BenchmarkTest/Test_CO2Benchmark.py
@@ -102,10 +102,10 @@ if __name__ == "__main__":
# error (or the highest score=1-LOO)estimator is chosen as the final
# metamodel.
MetaModelOpts.MinPceDegree = 1 # default = 1
- MetaModelOpts.MaxPceDegree = 10
+ MetaModelOpts.MaxPceDegree = 3
# q-quasi-norm 0<q<1 (default=1)
- MetaModelOpts.q = 1 #np.linspace(0.25,1, 4)
+ MetaModelOpts.q = np.linspace(0.25,1, 4)
# Select the sparse least-square minimization method for
# the PCE coefficients calculation:
@@ -113,10 +113,10 @@ if __name__ == "__main__":
# 3)LARS: Least angle regression 4)ARD: Bayesian ARD Regression
# 5)SGDRegressor: Stochastic gradient descent learning
- MetaModelOpts.RegMethod = 'BRR'
+ MetaModelOpts.RegMethod = 'ARD'
# Print summary of the regression results
- #MetaModelOpts.DisplayFlag = True
+ # MetaModelOpts.DisplayFlag = True
# ------ Experimental Design --------
# Generate an experimental design of size NrExpDesign based on a latin
@@ -126,13 +126,13 @@ if __name__ == "__main__":
# One-shot (normal) or Sequential Adaptive (sequential) Design
MetaModelOpts.ExpDesign.Method = 'sequential'
- initNSamples = 5
+ initNSamples = 10
MetaModelOpts.ExpDesign.NrSamples = initNSamples
# Sampling methods
# 1)random 2)PCM 3)LSCM 4)user
MetaModelOpts.ExpDesign.SamplingMethod = 'PCM'
- #MetaModelOpts.ExpDesign.X = np.load('EDX_CO2Benchmark.npy')
+ # MetaModelOpts.ExpDesign.X = np.load('EDX_CO2Benchmark.npy')
#MetaModelOpts.ExpDesign.Y = ExpDesign_Y
# Sequential experimental design (needed only for sequential ExpDesign)
--
GitLab