def test_catalogues_sort_by_other_than_zero(self):
cl = bmutils.regspace_cluster_to_target(self.data_for_cluster, 3, n_try_max=10, delta=0)
cat_idxs, cat_cont = bmutils.catalogues(cl, sort_by=1)
# This test is extra, since this is a pure pyemma functions
assert np.allclose(cat_idxs[0], [[1,0]])
assert np.allclose(cat_idxs[1], [[2, 0],
[2, 1],
[2, 2],
[3, 1],
[3, 3],
[3, 5]
])
assert np.allclose(cat_idxs[2], [[0,0],
[0,1],
[0,2],
[3,0],
[3,2],
[3,4]])
# Continous catalogue (all indices inverted)
assert np.allclose(cat_cont[2], [[1, 3],
[2, 3],
[3, 3],
[1, 3],
[2, 3],
[3, 3], ])
assert np.allclose(cat_cont[0], [[17, 1]])
assert np.allclose(cat_cont[1], [[11, 2],
[12, 2],
[13, 3],
[11, 2],
[12, 2],
[13, 2]])
def test_catalogues(self):
cl = bmutils.regspace_cluster_to_target(self.data_for_cluster, 3, n_try_max=10, delta=0)
#print(cl.clustercenters)
cat_idxs, cat_cont = bmutils.catalogues(cl)
# This test is extra, since this is a pure pyemma functions
assert np.allclose(cat_idxs[0], [[0,0], [0,1], [0,2],
[3,0], [3,2], [3,4]
])
assert np.allclose(cat_idxs[1], [[1,0]])
assert np.allclose(cat_idxs[2], [[2,0], [2,1], [2,2],
[3,1], [3,3], [3,5],
])
# Continous catalogue
assert np.allclose(cat_cont[0], [[1, 3],
[2, 3],
[3, 3],
[1, 3],
[2, 3],
[3, 3],])
assert np.allclose(cat_cont[1], [[17,1]])
assert np.allclose(cat_cont[2], [[11, 2],
[12, 2],
[13, 3],
[11, 2],
[12, 2],
[13, 2]])
def test_catalogues_sort_by_zero(self):
cl = bmutils.regspace_cluster_to_target(self.data_for_cluster, 3, n_try_max=10, delta=0)
cat_idxs, cat_cont = bmutils.catalogues(cl, sort_by=0)
# This test is extra, since this is a pure pyemma function
assert np.allclose(cat_idxs[0], [[0, 0], [0, 1], [0, 2],
[3, 0], [3, 2], [3, 4]
])
assert np.allclose(cat_idxs[2], [[1, 0]]) # notice indices inverted
assert np.allclose(cat_idxs[1], [[2, 0], [2, 1], [2, 2], # in these two lines
[3, 1], [3, 3], [3, 5],
])
# Notice that indices 2,1 have inverted wrt to previous test case
# Continous catalogue
assert np.allclose(cat_cont[0], [[1, 3],
[2, 3],
[3, 3],
[1, 3],
[2, 3],
[3, 3],])
assert np.allclose(cat_cont[2], [[17,1]]) # notice indices inverted
assert np.allclose(cat_cont[1], [[11, 2], # in these two lines
[12, 2],
[13, 3],
[11, 2],
[12, 2],
[13, 2]])
def setUp(self):
# The setup creates the typical, "geometries-sampled along cluster-scenario"
n_geom_samples = 20
traj = md.load(os.path.join(pyemma.__path__[0],'coordinates/tests/data/bpti_ca.pdb'))
traj = traj.atom_slice([0,1,3,4]) # create a trajectory with four atoms
# Create a fake bi-modal trajectory with a compact and an open structure
ixyz = np.array([[0., 0., 0.],
[1., 0., 0.],
[0., 1., 0.],
[0., 0., 1.]]) # The position of this atom will be varied. It's a proxy for rgyr
coords = np.zeros((2000, 4, 3))
for ii in range(2000):
coords[ii] = ixyz
# Here's were we create a bimodal distro were the most populated is the open structure
z = np.hstack((np.random.rand(200)*2 + 1, np.random.randn(1800) * 3 + 15))
z = np.random.permutation(z)
coords[:, -1, -1] = z
self.traj = md.Trajectory(coords, traj.top)
self.cl = bmutils.regspace_cluster_to_target(self.traj.xyz[:,-1, -1], 50, n_try_max=10)
self.cat_smpl = self.cl.sample_indexes_by_cluster(np.arange(self.cl.n_clusters), n_geom_samples)
self.geom_smpl = self.traj[np.vstack(self.cat_smpl)[:,1]]
self.geom_smpl = bmutils.re_warp(self.geom_smpl, [n_geom_samples]*self.cl.n_clusters)
def test_cluster_to_target(self):
n_target = 15
data = [np.random.randn(100, 1), np.random.randn(100,1)+10]
cl = bmutils.regspace_cluster_to_target(data, n_target, n_try_max=10, delta=0)
assert n_target - 1 <= cl.n_clusters <= n_target + 1