def get_qm_positions_energies_and_charges(sigma,
overlapping_precision,
database='qm7',
planar=False):
if database == 'qm7':
qm = load_qm7(align=True, only_planar=planar)
elif database == 'qm9':
qm = load_qm9(align=True, only_planar=planar)
else:
raise ValueError('only qm7 and qm9 databases')
positions = qm.R.astype('float32')
atomic_numbers = qm.Z.astype('float32')
atom_valences = get_atom_valences(atomic_numbers).astype('float32')
electron_valences = get_electron_valences(atomic_numbers).astype('float32')
min_dist = np.inf
for i in range(positions.shape[0]):
n_atoms = np.sum(atomic_numbers[i] != 0)
distances = pdist(positions[i, :n_atoms, :])
min_dist = min(min_dist, distances.min())
delta = sigma * np.sqrt(-8 * np.log(overlapping_precision))
print(delta, min_dist)
positions *= delta / min_dist
return (
torch.from_numpy(positions),
torch.from_numpy(atomic_numbers),
torch.from_numpy(atom_valences),
torch.from_numpy(electron_valences),
)
def get_qm_energies(database='qm7', planar=False):
if database == 'qm7':
return load_qm7(align=True, only_planar=planar).T.transpose()
elif database == 'qm9':
return load_qm9(align=True, only_planar=planar).T.transpose()
else:
raise ValueError('only qm7 and qm9 databases')
def get_qm7_positions_energies_and_charges(M, N, O, J, L, sigma):
qm7 = load_qm7(align=True)
positions = qm7.R
charges = qm7.Z
energies = qm7.T.transpose()
valence_charges = get_valence(charges)
positions = renormalize(positions, charges, sigma)
return torch.from_numpy(positions), torch.from_numpy(
energies), torch.from_numpy(charges), torch.from_numpy(valence_charges)
def get_qm_energies_and_folds(random_folds=False, database='qm7'):
if database == 'qm7':
qm = load_qm7(align=True)
P = qm.P
elif database == 'qm9':
qm = load_qm9(align=True)
P = qm.P_stratified_Ua.transpose()
else:
raise ValueError('only qm7 and qm9 databases')
n_folds = P.shape[0]
energies = qm.T.transpose()
if random_folds:
P = np.random.permutation(energies.shape[0]).reshape((n_folds, -1))
cross_val_folds = []
for i_fold in range(n_folds):
fold = (np.concatenate(P[np.arange(n_folds) != i_fold],
axis=0), P[i_fold])
cross_val_folds.append(fold)
return energies, cross_val_folds
def get_qm7_positions_and_charges(sigma, overlapping_precision=1e-1):
"""
Loads the positions and charges of the molecules of the QM7 dataset.
QM7 is a dataset of 7165 organic molecules with up to 7 non-hydrogen
atoms, whose energies were computed with Density Functional Theory.
This dataset has been made available to train machine learning models
to predict these energies.
Parameters
----------
sigma : float
width parameter of the Gaussian that represents a particle
Returns
-------
positions, charges, valence_charges: float array
array containing the positions, charges and valence charges
of the QM7 database molecules
"""
# qm7 = fetch_qm7(align=True)
# positions = qm7['positions']
# charges = qm7['charges'].astype('float32')
from cheml.datasets import load_qm7
qm7 = load_qm7(align=True)
positions = qm7.R
charges = qm7.Z
valence_charges = get_valence(charges)
# normalize positions
min_dist = np.inf
for i in range(positions.shape[0]):
n_atoms = np.sum(charges[i] != 0)
pos = positions[i, :n_atoms, :]
min_dist = min(min_dist, pdist(pos).min())
delta = sigma * np.sqrt(-8 * np.log(overlapping_precision))
positions = positions * delta / min_dist
return (torch.from_numpy(positions), torch.from_numpy(charges),
torch.from_numpy(valence_charges))