def setUp(self):
N = 10
x = np.linspace(-5,5,N)
self.X = np.reshape(x, (-1, 1))
self.basis = 'poly3'
H, self.featurizer, self.featurizer_recovery = featurizer_from_string(self.X, self.basis, extract_dim=0)
self.beta = [9.91898792e-01, -1.62113090e+00, 3.15437605e+00, 1.25732838e+00]
self.B = np.eye(4) * 9
self.b = np.zeros((4,))
p = np.dot(H, self.beta)
def covar_matrix(x, k):
n = len(x)
K = np.zeros((n,n))
for i in range(n):
for j in range(n):
K[i,j] = k(x[i], x[j])
return K
k1 = lambda x1, x2 : .001*np.exp( - ((x1-x2)/1.5)**2 ) + (.001 if x1==x2 else 0)
K1 = covar_matrix(x, k1)
np.random.seed(0)
f1 = np.random.multivariate_normal(mean=np.zeros((len(x),)), cov=K1)
self.y1 = f1 + p
self.cov = GPCov(wfn_params=[.001,], dfn_params=[ 1.5,], wfn_str="se", dfn_str="euclidean")
self.noise_var = .001
self.gp = GP(X=self.X,
y=self.y1,
noise_var = self.noise_var,
cov_main = self.cov,
basis = self.basis,
featurizer_recovery = self.featurizer_recovery,
param_mean=self.b,
param_cov=self.B,
compute_ll=True,
compute_grad=True,
sparse_threshold=0,
build_tree=False,
sparse_invert=True)
def setUp(self):
N = 25
x = np.linspace(-5,5,N)
self.X = np.reshape(x, (-1, 1))
self.u = np.array(((-2.0,), (2.0,)))
self.basis = 'poly3'
H, self.featurizer, self.featurizer_recovery = featurizer_from_string(self.X, self.basis, extract_dim=0)
self.beta = np.array([9.91898792e-01, -1.62113090e+00, 3.15437605e+00, 1.25732838e+00])*10
self.B = np.eye(4) * 9
self.b = np.zeros((4,))
self.p = np.dot(H, self.beta)
f1 = np.array([-1.02804007, -1.54448568, -0.31653812, -0.46768499, 0.67463927, 1.06519473, -1.39472442, -0.72392324, -2.99133689, -0.59922449, -3.70430871, -1.75810012, -0.80376896, -0.50514541, -0.5459166, 1.6353825, -1.13032502, 0.80372166, -0.01374143, -1.16083918, -1.6099601, -4.37523678, -1.53780366, -2.98047752, -3.41214803])
self.y1 = f1 + self.p
self.cov_main = GPCov(wfn_params=[0.3,], dfn_params=[ 0.5,], wfn_str="compact2", dfn_str="euclidean")
self.cov_fic = GPCov(wfn_params=[0.2,], dfn_params=[ 2.5,], wfn_str="se", dfn_str="euclidean", Xu = self.u)
self.noise_var = .001
self.gp = GP(X=self.X,
y=self.y1,
noise_var = self.noise_var,
cov_main = self.cov_main,
cov_fic = self.cov_fic,
basis = self.basis,
featurizer_recovery = self.featurizer_recovery,
param_mean=self.b,
param_cov=self.B,
compute_ll=True,
compute_grad=True,
compute_xu_grad=True,
sparse_threshold=0,
build_tree=False,
sparse_invert=True)
