def test_diagmatrix(self):
traj = pt.iterload("./data/tz2.nc", "./data/tz2.parm7")
state = pt.load_batch(traj, '''
#matrix covar @CA name mymat
matrix dist @CA name mymat
diagmatrix mymat vecs 6''')
state.run()
mat = state.data['mymat']
cpp_evals, cpp_evecs = state.data[-1].eigenvalues, state.data[
-1].eigenvectors
# test triu_indices
mat2 = mat.__class__()
indices = np.triu_indices(mat.n_cols)
mat2._set_data_half_matrix(mat._to_cpptraj_sparse_matrix(), mat.size,
mat.n_cols)
# OK
data = matrix.diagonalize(mat.values, n_vecs=6, dtype='dataset')[-1]
data2 = matrix.diagonalize(mat, n_vecs=6, dtype='dataset')[0]
aa_eq(data.eigenvalues, cpp_evals, decimal=10)
aa_eq(data2.eigenvalues, cpp_evals, decimal=10)
# try numpy
np_vals, np_vecs = np.linalg.eigh(mat2.values, UPLO='U')
np_vals = np_vals[::-1][:6]
np_vecs = np_vecs[:, ::-1].T[:6]
# at least the absolute values are equal
aa_eq(np.abs(data.eigenvectors), np.abs(cpp_evecs))
aa_eq(np.abs(np_vecs), np.abs(cpp_evecs))
# test raise if not having supported dtype
self.assertRaises(ValueError, lambda: pt.matrix.diagonalize(mat, 3, dtype='ndarray'))
# plot
def plot_():
try:
from pytraj.plot import plot_matrix
from matplotlib import pyplot as plt
fig, ax, bar = plot_matrix(np.abs(data.eigenvectors) - np.abs(
cpp_evecs))
fig.colorbar(bar)
plt.title('data vs cpp_evecs')
fig, ax, bar = plot_matrix(np.abs(np_vecs) - np.abs(cpp_evecs))
fig.colorbar(bar)
plt.title('np vs cpp_evecs')
fig, ax, bar = plot_matrix(np.abs(np_vecs) - np.abs(
data.eigenvectors))
fig.colorbar(bar)
plt.title('np vs data')
pt.show()
except ImportError:
pass
def test_diagmatrix(self):
traj = pt.iterload(fn('tz2.nc'), fn('tz2.parm7'))
state = pt.load_batch(
traj, '''
#matrix covar @CA name mymat
matrix dist @CA name mymat
diagmatrix mymat vecs 6''')
state.run()
mat = state.data['mymat']
cpp_evals, cpp_evecs = state.data[-1].eigenvalues, state.data[
-1].eigenvectors
# test triu_indices
mat2 = mat.__class__()
np.triu_indices(mat.n_cols)
mat2._set_data_half_matrix(mat._to_cpptraj_sparse_matrix(), mat.size,
mat.n_cols)
# OK
data = matrix.diagonalize(mat.values, n_vecs=6, dtype='dataset')[-1]
data2 = matrix.diagonalize(mat, n_vecs=6, dtype='dataset')[0]
aa_eq(data.eigenvalues, cpp_evals, decimal=10)
aa_eq(data2.eigenvalues, cpp_evals, decimal=10)
# try numpy
np_vals, np_vecs = np.linalg.eigh(mat2.values, UPLO='U')
np_vals = np_vals[::-1][:6]
np_vecs = np_vecs[:, ::-1].T[:6]
# at least the absolute values are equal
aa_eq(np.abs(data.eigenvectors), np.abs(cpp_evecs))
aa_eq(np.abs(np_vecs), np.abs(cpp_evecs))
# test raise if not having supported dtype
self.assertRaises(
ValueError, lambda: pt.matrix.diagonalize(mat, 3, dtype='ndarray'))
def nh_order_parameters(traj,
vector_pairs,
order=2,
tstep=1.,
tcorr=10000.,
n_cores=1,
**kwargs):
'''compute NH order parameters
Parameters
----------
traj : Trajectory-like
vector_pairs : 2D array-like, shape (n_pairs, 2)
order : default 2
tstep : default 1.
tcorr : default 10000.
kwargs : additional keyword argument
Returns
-------
S2 : 1D array, order parameters
Examples
--------
>>> import pytraj as pt
>>> traj = pt.datafiles.load_tz2_ortho()
>>> h_indices = pt.select('@H', traj.top)
>>> n_indices = h_indices - 1
>>> nh_pairs = list(zip(n_indices, h_indices))
>>> data = pt.nh_order_parameters(traj, nh_pairs)
'''
from pytraj import matrix
# compute N-H vectors and ired matrix
if n_cores == 1:
vecs_and_mat = ired_vector_and_matrix(
traj, vector_pairs, order=order, dtype='tuple')
else:
# use _pmap to avoid cicular import
from pytraj import _pmap
vecs_and_mat = _pmap(
ired_vector_and_matrix,
traj,
vector_pairs,
order=order,
dtype='tuple',
n_cores=n_cores)
state_vecs = vecs_and_mat[0]
mat_ired = vecs_and_mat[1]
# get eigenvalues and eigenvectors
modes = matrix.diagonalize(
mat_ired, n_vecs=len(state_vecs), dtype='dataset')[0]
evals, evecs = modes.eigenvalues, modes.eigenvectors
data = _ired(
state_vecs,
modes=(evecs, evals),
NHbond=True,
tcorr=tcorr,
tstep=tstep,
**kwargs)
order = [d.values.copy() for d in data if 'S2' in d.key][0]
return order
def NH_order_parameters(traj,
vector_pairs,
order=2,
tstep=1.,
tcorr=10000.,
n_cores=1):
'''compute NH order parameters
Parameters
----------
traj : Trajectory-like
vector_pairs : 2D array-like, shape (n_pairs, 2)
order : default 2
tstep : default 1.
tcorr : default 10000.
Returns
-------
S2 : 1D array, order parameters
Examples
--------
>>> import pytraj as pt
>>> traj = pt.datafiles.load_tz2_ortho()
>>> h_indices = pt.select('@H', traj.top)
>>> n_indices = h_indices - 1
>>> nh_pairs = list(zip(n_indices, h_indices))
>>> data = pt.NH_order_parameters(traj, nh_pairs)
'''
from pytraj import matrix
# compute N-H vectors and ired matrix
if n_cores == 1:
vecs_and_mat = ired_vector_and_matrix(traj,
vector_pairs,
order=order,
dtype='tuple')
else:
# use _pmap to avoid cicular import
from pytraj import _pmap
vecs_and_mat = _pmap(ired_vector_and_matrix,
traj,
vector_pairs,
order=order,
dtype='tuple',
n_cores=n_cores)
state_vecs = vecs_and_mat[0]
mat_ired = vecs_and_mat[1]
# get eigenvalues and eigenvectors
modes = matrix.diagonalize(mat_ired, n_vecs=len(state_vecs), dtype='dataset')[0]
evals, evecs = modes.eigenvalues, modes.eigenvectors
data = _ired(state_vecs,
modes=(evals, evecs),
NHbond=True,
tcorr=tcorr,
tstep=tstep)
order = [d.values.copy() for d in data if 'S2' in d.key][0]
return order