mirror of
https://github.com/SheffieldML/GPy.git
synced 2026-05-12 13:32:39 +02:00
47 lines
1.6 KiB
Python
47 lines
1.6 KiB
Python
# Copyright (c) 2012, Nicolo Fusi
|
|
# Licensed under the BSD 3-clause license (see LICENSE.txt)
|
|
|
|
import unittest
|
|
import numpy as np
|
|
import GPy
|
|
|
|
class GPLVMTests(unittest.TestCase):
|
|
def test_bias_kern(self):
|
|
N, M, Q, D = 10, 3, 2, 4
|
|
X = np.random.rand(N, Q)
|
|
k = GPy.kern.rbf(Q) + GPy.kern.white(Q, 0.00001)
|
|
K = k.K(X)
|
|
Y = np.random.multivariate_normal(np.zeros(N),K,D).T
|
|
k = GPy.kern.bias(Q) + GPy.kern.white(Q, 0.00001)
|
|
m = GPy.models.GPLVM(Y, Q, kernel = k)
|
|
m.ensure_default_constraints()
|
|
m.randomize()
|
|
self.assertTrue(m.checkgrad())
|
|
|
|
def test_linear_kern(self):
|
|
N, M, Q, D = 10, 3, 2, 4
|
|
X = np.random.rand(N, Q)
|
|
k = GPy.kern.rbf(Q) + GPy.kern.white(Q, 0.00001)
|
|
K = k.K(X)
|
|
Y = np.random.multivariate_normal(np.zeros(N),K,D).T
|
|
k = GPy.kern.linear(Q) + GPy.kern.white(Q, 0.00001)
|
|
m = GPy.models.GPLVM(Y, Q, kernel = k)
|
|
m.ensure_default_constraints()
|
|
m.randomize()
|
|
self.assertTrue(m.checkgrad())
|
|
|
|
def test_rbf_kern(self):
|
|
N, M, Q, D = 10, 3, 2, 4
|
|
X = np.random.rand(N, Q)
|
|
k = GPy.kern.rbf(Q) + GPy.kern.white(Q, 0.00001)
|
|
K = k.K(X)
|
|
Y = np.random.multivariate_normal(np.zeros(N),K,D).T
|
|
k = GPy.kern.rbf(Q) + GPy.kern.white(Q, 0.00001)
|
|
m = GPy.models.GPLVM(Y, Q, kernel = k)
|
|
m.ensure_default_constraints()
|
|
m.randomize()
|
|
self.assertTrue(m.checkgrad())
|
|
|
|
if __name__ == "__main__":
|
|
print "Running unit tests, please be (very) patient..."
|
|
unittest.main()
|