def test_chunksize(self):
reader_xtc = api.source(self.traj_files, top=self.pdb_file)
chunksize = 1001
chain = [reader_xtc, api.tica(), api.cluster_mini_batch_kmeans()]
p = api.pipeline(chain, chunksize=chunksize)
assert p.chunksize == chunksize
for e in p._chain:
assert e.chunksize == chunksize
def test_3gaussian_1d_singletraj(self):
# generate 1D data from three gaussians
X = [
np.random.randn(200) - 2.0,
np.random.randn(300),
np.random.randn(400) + 2.0
]
X = np.hstack(X)
kmeans = cluster_mini_batch_kmeans(X,
batch_size=0.5,
k=100,
max_iter=10000)
cc = kmeans.clustercenters
assert (np.any(cc < 1.0))
assert (np.any((cc > -1.0) * (cc < 1.0)))
assert (np.any(cc > -1.0))
def test_3gaussian_2d_multitraj(self):
# generate 1D data from three gaussians
X1 = np.zeros((200, 2))
X1[:, 0] = np.random.randn(200) - 2.0
X2 = np.zeros((300, 2))
X2[:, 0] = np.random.randn(300)
X3 = np.zeros((400, 2))
X3[:, 0] = np.random.randn(400) + 2.0
X = [X1, X2, X3]
kmeans = cluster_mini_batch_kmeans(X,
batch_size=0.5,
k=100,
max_iter=10000)
cc = kmeans.clustercenters
assert (np.any(cc < 1.0))
assert (np.any((cc > -1.0) * (cc < 1.0)))
assert (np.any(cc > -1.0))
def test_resume(self):
""" check that we can continue with the iteration by passing centers"""
# centers are far off
initial_centers = np.array([[1, 2, 3]]).T
cl = cluster_mini_batch_kmeans(self.X,
clustercenters=initial_centers,
max_iter=1,
k=3)
resume_centers = cl.clustercenters
cl.estimate(self.X, clustercenters=resume_centers, max_iter=50)
new_centers = cl.clustercenters
true = np.array([[-2, 0, 2]]).T
d0 = true - resume_centers
d1 = true - new_centers
diff = np.linalg.norm(d0)
diff_next = np.linalg.norm(d1)
self.assertLess(
diff_next, diff, 'resume_centers=%s, new_centers=%s' %
(resume_centers, new_centers))
