From 52e74e69c364cf8f1fda4c4b25e688a6a8ffc137 Mon Sep 17 00:00:00 2001
From: Farid Mohammadi <farid.mohammadi@iws.uni-stuttgart.de>
Date: Mon, 30 May 2022 09:30:13 +0200
Subject: [PATCH] [surrogate] change large array types to np.float16
---
.../surrogate_models/eval_rec_rule.py | 2 +-
.../surrogate_models/reg_fast_ard.py | 2 +-
.../surrogate_models/surrogate_models.py | 22 +++++--------------
3 files changed, 8 insertions(+), 18 deletions(-)
diff --git a/src/bayesvalidrox/surrogate_models/eval_rec_rule.py b/src/bayesvalidrox/surrogate_models/eval_rec_rule.py
index b583c7eb2..730a3818d 100644
--- a/src/bayesvalidrox/surrogate_models/eval_rec_rule.py
+++ b/src/bayesvalidrox/surrogate_models/eval_rec_rule.py
@@ -183,7 +183,7 @@ def eval_univ_basis(x, max_deg, poly_types, apoly_coeffs=None):
"""
# Initilize the output array
n_samples, n_params = x.shape
- univ_vals = np.zeros((n_samples, n_params, max_deg+1))
+ univ_vals = np.zeros((n_samples, n_params, max_deg+1), dtype=np.float16)
for i in range(n_params):
diff --git a/src/bayesvalidrox/surrogate_models/reg_fast_ard.py b/src/bayesvalidrox/surrogate_models/reg_fast_ard.py
index 2d9eaca8a..1b1582860 100755
--- a/src/bayesvalidrox/surrogate_models/reg_fast_ard.py
+++ b/src/bayesvalidrox/surrogate_models/reg_fast_ard.py
@@ -384,7 +384,7 @@ class RegressionFastARD(LinearModel, RegressorMixin):
if self.normalize:
x = (X - self._x_mean_) / self._x_std
var_hat = 1./self.alpha_
- var_hat += np.sum(np.dot(x[:, self.active_], self.sigma_) *
+ var_hat += np.sum(x[:, self.active_].dot(self.sigma_) *
x[:, self.active_], axis=1)
std_hat = np.sqrt(var_hat)
return y_hat, std_hat
diff --git a/src/bayesvalidrox/surrogate_models/surrogate_models.py b/src/bayesvalidrox/surrogate_models/surrogate_models.py
index 039a5d68e..f519f39a9 100644
--- a/src/bayesvalidrox/surrogate_models/surrogate_models.py
+++ b/src/bayesvalidrox/surrogate_models/surrogate_models.py
@@ -131,7 +131,7 @@ class MetaModel:
if self.ExpDesign.method == 'sequential':
from .sequential_design import SeqDesign
seq_design = SeqDesign(self)
- metamodel = seq_design.train_seq_design(Model)
+ metamodel = seq_design.train_seq_design()
elif self.ExpDesign.method == 'normal':
self.ExpDesignFlag = 'normal'
@@ -488,14 +488,14 @@ class MetaModel:
" match!!")
# Preallocate the Psi matrix for performance
- psi = np.ones((n_samples, n_terms))
+ psi = np.ones((n_samples, n_terms), dtype=np.float16)
# Assemble the Psi matrix
for m in range(basis_indices.shape[1]):
aa = np.where(basis_indices[:, m] > 0)[0]
try:
basisIdx = basis_indices[aa, m]
- bb = np.reshape(univ_p_val[:, m, basisIdx], psi[:, aa].shape)
- psi[:, aa] = np.multiply(psi[:, aa], bb)
+ bb = univ_p_val[:, m, basisIdx].reshape(psi[:, aa].shape)
+ psi[:, aa] *= bb
except ValueError as err:
raise err
@@ -569,7 +569,6 @@ class MetaModel:
n_iter=1000, tol=1e-7)
elif reg_method.lower() == 'lars':
- print("lars")
clf_poly = lm.LassoLarsCV(fit_intercept=False)
elif reg_method.lower() == 'sgdr':
@@ -1106,17 +1105,9 @@ class MetaModel:
model_dict = self.coeffs_dict
if samples is None:
- if nsamples is None:
- self.n_samples = 100000
- else:
- self.n_samples = nsamples
-
- samples = self.ExpDesign.generate_samples(self.n_samples,
+ # Generate
+ samples = self.ExpDesign.generate_samples(nsamples,
sampling_method)
- else:
- self.samples = samples
- self.n_samples = len(samples)
-
# Transform samples
samples = self.ExpDesign.transform(samples)
@@ -1151,7 +1142,6 @@ class MetaModel:
# with error bar
clf_poly = self.clf_poly[ouput][in_key]
y_mean, y_std = clf_poly.predict(psi, return_std=True)
-
except:
# without error bar
coeffs = self.coeffs_dict[ouput][in_key]
--
GitLab