def gp_matrix(x0, x1, a, K):
"""Constructs GP covariance matrix between two point tuples x0 and x1"""
n0 = len(x0)
n1 = len(x1)
for k0 in range(n0):
for k1 in range(n1):
K[k0, k1] = a[1]*kern_sqexp(x0[k0, :], x1[k1, :], a[0])
def predict(self, x):
if not self.trained:
raise RuntimeError('Need to train() before predict()')
Kstar = self.hyparms[1]*np.fromiter(
(kern_sqexp(xi, xj, self.hyparms[0])
for xi in self.xtrain for xj in x),
float).reshape(self.xtrain.shape[0], x.shape[0])
return Kstar.T.dot(self.Kyinv_y)
def test_kern_sqexp():
points = np.array([[0.0, 0.0], [0.0, 1.5], [1.3, 1.3]])
h0 = 1.2
for x in points:
for x0 in points:
k1 = kernels.kern_sqexp(x, x0, h0)
k1a = kernels.kern_sqexp_multiscale(x, x0, np.array([h0, h0]))
k2 = np.exp(-0.5 * np.sum((x - x0)**2) / h0**2)
print(k1)
print(",,,,")
print(k1a)
assert np.array_equal(k1, k2)
assert np.array_equal(k1a, k2)
def build_K(xa, xb, hyp, K):
for i in np.arange(len(xa)):
for j in np.arange(len(xb)):
K[i, j] = kern_sqexp(xa[i], xb[j], hyp[0])