def test_representation_transform(self):
rep = SphericalExpansion(**self.hypers)
features = rep.transform(self.frames)
test = features.get_features(rep)
def test_representation_transform(self):
rep = SphericalExpansion(**self.hypers)
features = rep.transform([self.frame])
test = features.get_dense_feature_matrix(rep)
def benchmark_spherical_representations(frames, optimization_args,
radial_basis):
hypers = {
'interaction_cutoff': INTERACTION_CUTOFF,
'max_radial': 8,
'max_angular': 6,
'gaussian_sigma_constant': 0.5,
'gaussian_sigma_type': "Constant",
'cutoff_smooth_width': 0.5,
'radial_basis': radial_basis,
'optimization_args': optimization_args
}
print("Timing SphericalExpansion")
transform_representation(SphericalExpansion(**hypers),
frames,
nb_iterations=NB_ITERATIONS_PER_REPRESENTATION)
hypers = {
'soap_type': "PowerSpectrum",
'interaction_cutoff': INTERACTION_CUTOFF,
'max_radial': 8,
'max_angular': 6,
'gaussian_sigma_constant': 0.5,
'gaussian_sigma_type': "Constant",
'cutoff_smooth_width': 0.5,
'normalize': False,
'radial_basis': radial_basis,
'optimization_args': optimization_args
}
print("Timing SphericalInvariants")
transform_representation(SphericalInvariants(**hypers),
frames,
nb_iterations=NB_ITERATIONS_PER_REPRESENTATION)
def compute_se_cpp(job):
# setup input for the script
data = job.statepoint()
rep = SphericalExpansion(**data['representation'])
data['calculator'] = rep.hypers
tojson(job.fn(group['fn_in']), data)
# look at memory footprint
p = Popen([
group['executable'][data['nl_type']],
job.fn(group['fn_in']),
job.fn(group['fn_out'])
],
stdout=PIPE,
stderr=PIPE)
max_mem = memory_usage(p, interval=0.1, max_usage=True)
# look at timings
p = Popen([
group['executable'][data['nl_type']],
job.fn(group['fn_in']),
job.fn(group['fn_out'])
],
stdout=PIPE,
stderr=PIPE)
if p.stderr.read(): print(p.stderr.read())
data = fromjson(job.fn(group['fn_out']))
data = data['results']
data['mem_max'] = max_mem
data['mem_unit'] = 'MiB'
tojson(job.fn(group['fn_res']), data)
def test_serialization(self):
rep = SphericalExpansion(**self.hypers)
rep_dict = to_dict(rep)
rep_copy = from_dict(rep_dict)
rep_copy_dict = to_dict(rep_copy)
self.assertTrue(rep_dict == rep_copy_dict)
def test_hypers_construction(self):
"""Checks that manually-constructed and automatic
framework are consistent."""
hypers = deepcopy(self.hypers)
hypers["max_radial"] = self.expanded_max_radial
spex = SphericalExpansion(**hypers)
feats = spex.transform(self.frames).get_features_by_species(spex)
cov = get_radial_basis_covariance(spex, feats)
p_val, p_vec = get_radial_basis_pca(cov)
p_mat = get_radial_basis_projections(p_vec, self.hypers["max_radial"])
# now makes this SOAP
hypers["max_radial"] = self.hypers["max_radial"]
hypers["soap_type"] = "PowerSpectrum"
hypers["optimization"] = {
"RadialDimReduction": {
"projection_matrices": p_mat
},
"Spline": {
"accuracy": 1e-8
},
}
# compute SOAP
soap_opt = SphericalInvariants(**hypers)
soap_feats = soap_opt.transform(self.frames).get_features(soap_opt)
# now we do the same with the compact utils
hypers = deepcopy(self.hypers)
hypers["soap_type"] = "PowerSpectrum"
hypers = get_optimal_radial_basis_hypers(
hypers, self.frames, expanded_max_radial=self.expanded_max_radial)
soap_opt_2 = SphericalInvariants(**hypers)
soap_feats_2 = soap_opt_2.transform(
self.frames).get_features(soap_opt_2)
self.assertTrue(np.allclose(soap_feats, soap_feats_2))
def benchmark_spherical_representations(frames, optimization_args,
radial_basis):
hypers = {
"interaction_cutoff": INTERACTION_CUTOFF,
"max_radial": 8,
"max_angular": 6,
"gaussian_sigma_constant": 0.5,
"gaussian_sigma_type": "Constant",
"cutoff_smooth_width": 0.5,
"radial_basis": radial_basis,
"optimization": optimization_args,
}
print("Timing SphericalExpansion")
transform_representation(
SphericalExpansion(**hypers),
frames,
nb_iterations=NB_ITERATIONS_PER_REPRESENTATION,
)
hypers = {
"soap_type": "PowerSpectrum",
"interaction_cutoff": INTERACTION_CUTOFF,
"max_radial": 8,
"max_angular": 6,
"gaussian_sigma_constant": 0.5,
"gaussian_sigma_type": "Constant",
"cutoff_smooth_width": 0.5,
"normalize": False,
"radial_basis": radial_basis,
"optimization": optimization_args,
}
print("Timing SphericalInvariants")
transform_representation(
SphericalInvariants(**hypers),
frames,
nb_iterations=NB_ITERATIONS_PER_REPRESENTATION,
)
def compute_ri_cpp(job):
# setup input for the script
data = job.statepoint()
rep = SphericalExpansion(**data['representation'])
data['calculator'] = rep.hypers
tojson(job.fn(group['fn_in']), data)
# look at memory footprint
p = Popen(
[group['executable'],
job.fn(group['fn_in']),
job.fn(group['fn_out'])],
stdout=PIPE,
stderr=PIPE)
max_mem = memory_usage(p, interval=0.001, max_usage=True)
# look at timings
carry_on = True
N_ITERATIONS = [
int(data['N_ITERATIONS'] * (1 + f))
for f in [0., 0.5, 1, 2, 3, 5, 10, 20, 30, 50, 100]
]
for N_ITERATION in N_ITERATIONS:
if run(job, N_ITERATION, max_mem):
break
def test_radial_dimension_reduction_test(self):
rep = SphericalExpansion(**self.hypers)
features_ref = rep.transform(self.frames).get_features(rep)
K_ref = features_ref.dot(features_ref.T)
# identity test
hypers = deepcopy(self.hypers)
projection_matrices = [
np.eye(self.max_radial).tolist()
for _ in range(self.max_angular + 1)
]
hypers["optimization"] = {
"Spline": {
"accuracy": 1e-8
},
"RadialDimReduction": {
"projection_matrices":
{sp: projection_matrices
for sp in self.species}
},
}
rep = SphericalExpansion(**hypers)
features_test = rep.transform(self.frames).get_features(rep)
self.assertTrue(np.allclose(features_ref, features_test))
# random orthogonal matrix test,
# for angular_l=0 we can do the projection on the python side
hypers = deepcopy(self.hypers)
np.random.seed(self.seed)
projection_matrices = {
sp: [
ortho_group.rvs(self.max_radial)[:self.max_radial -
1, :].tolist()
for _ in range(self.max_angular + 1)
]
for sp in self.species
}
hypers["max_radial"] = self.max_radial - 1
hypers["optimization"] = {
"Spline": {
"accuracy": 1e-8
},
"RadialDimReduction": {
"projection_matrices": projection_matrices
},
}
rep = SphericalExpansion(**hypers)
features_test = rep.transform(self.frames).get_features(rep)
features_test = features_test.reshape(
len(features_test),
len(self.species),
self.max_radial - 1,
(self.max_angular + 1)**2,
)
hypers["max_radial"] = self.max_radial
hypers["optimization"] = {
"Spline": {
"accuracy": 1e-8
},
}
rep = SphericalExpansion(**hypers)
features_ref = rep.transform(self.frames).get_features(rep)
features_ref = features_ref.reshape(
len(features_ref),
len(self.species),
self.max_radial,
(self.max_angular + 1)**2,
)
for a, species in enumerate(self.species):
self.assertTrue(
np.allclose(
features_ref[:, a, :, 0] @ np.array(
projection_matrices[species][0]).T,
features_test[:, a, :, 0],
))
# checks if error is raised if wrong number of species is given
with self.assertRaises(RuntimeError):
species = [1]
hypers["optimization"] = {
"Spline": {
"accuracy": 1e-8
},
"RadialDimReduction": {
"projection_matrices":
{sp: projection_matrices
for sp in species}
},
}
rep = SphericalExpansion(**hypers)
features_test = rep.transform(self.frames).get_features(rep)
def test_pickle(self):
rep = SphericalExpansion(**self.hypers)
serialized = pickle.dumps(rep)
rep_ = pickle.loads(serialized)
self.assertTrue(to_dict(rep) == to_dict(rep_))
def dump_reference_json():
import ubjson
from copy import copy
from itertools import product
sys.path.insert(0, os.path.join(root, 'build/'))
sys.path.insert(0, os.path.join(root, 'tests/'))
cutoffs = [2, 3]
gaussian_sigmas = [0.2, 0.5]
max_radials = [4, 10]
max_angulars = [3, 6]
cutoff_smooth_widths = [0., 1.]
radial_basis = ["GTO", "DVR"]
cutoff_function_types = ['ShiftedCosine', 'RadialScaling']
fns = [
os.path.join(inputs_path, "CaCrP2O7_mvc-11955_symmetrized.json"),
os.path.join(inputs_path, "small_molecule.json")
]
fns_to_write = [
os.path.join(dump_path, "CaCrP2O7_mvc-11955_symmetrized.json"),
os.path.join(dump_path, "small_molecule.json"),
]
data = dict(filenames=fns_to_write,
cutoffs=cutoffs,
gaussian_sigmas=gaussian_sigmas,
max_radials=max_radials,
rep_info=[])
for fn in fns:
for cutoff in cutoffs:
data['rep_info'].append([])
for (gaussian_sigma, max_radial, max_angular, cutoff_smooth_width,
rad_basis, cutoff_function_type) in product(
gaussian_sigmas, max_radials, max_angulars,
cutoff_smooth_widths, radial_basis,
cutoff_function_types):
frames = read(fn)
if cutoff_function_type == 'RadialScaling':
cutoff_function_parameters = dict(rate=1,
scale=cutoff * 0.5,
exponent=3)
else:
cutoff_function_parameters = dict()
hypers = {
"interaction_cutoff": cutoff,
"cutoff_smooth_width": cutoff_smooth_width,
"max_radial": max_radial,
"max_angular": max_angular,
"gaussian_sigma_type": "Constant",
"cutoff_function_type": cutoff_function_type,
'cutoff_function_parameters': cutoff_function_parameters,
"gaussian_sigma_constant": gaussian_sigma,
"radial_basis": rad_basis
}
sph_expn = SphericalExpansion(**hypers)
expansions = sph_expn.transform(frames)
x = expansions.get_features(sph_expn)
x[np.abs(x) < 1e-300] = 0.
data['rep_info'][-1].append(
dict(feature_matrix=x.tolist(),
hypers=copy(sph_expn.hypers)))
with open(
os.path.join(root, dump_path,
"spherical_expansion_reference.ubjson"), 'wb') as f:
ubjson.dump(data, f)
def get_soap_vectors(hypers, frames):
with ostream_redirect():
sph_expn = SphericalExpansion(**hypers)
expansions = sph_expn.transform(frames)
soap_vectors = expansions.get_features(sph_expn)
return soap_vectors
def dump_reference_json():
import ubjson
import os
from copy import copy
path = '../'
sys.path.insert(0, os.path.join(path, 'build/'))
sys.path.insert(0, os.path.join(path, 'tests/'))
cutoffs = [2, 3]
gaussian_sigmas = [0.2, 0.5]
max_radials = [4, 10]
max_angulars = [3, 6]
cutoff_smooth_widths = [0., 1.]
radial_basis = ["GTO"]
fns = [
os.path.join(
path, "tests/reference_data/CaCrP2O7_mvc-11955_symmetrized.json"),
os.path.join(path, "tests/reference_data/small_molecule.json")
]
fns_to_write = [
"reference_data/CaCrP2O7_mvc-11955_symmetrized.json",
"reference_data/small_molecule.json",
]
data = dict(filenames=fns_to_write,
cutoffs=cutoffs,
gaussian_sigmas=gaussian_sigmas,
max_radials=max_radials,
rep_info=[])
# An example of the gruesomeness of using 4 spaces for one tab
for fn in fns:
for cutoff in cutoffs:
data['rep_info'].append([])
for gaussian_sigma in gaussian_sigmas:
for max_radial in max_radials:
for max_angular in max_angulars:
for cutoff_smooth_width in cutoff_smooth_widths:
for rad_basis in radial_basis:
frames = [json2ase(load_json(fn))]
hypers = {
"interaction_cutoff": cutoff,
"cutoff_smooth_width": cutoff_smooth_width,
"max_radial": max_radial,
"max_angular": max_angular,
"gaussian_sigma_type": "Constant",
"cutoff_function_type": "Cosine",
"gaussian_sigma_constant": gaussian_sigma,
"radial_basis": rad_basis
}
# x = get_soap_vectors(hypers, frames)
sph_expn = SphericalExpansion(**hypers)
expansions = sph_expn.transform(frames)
x = expansions.get_feature_matrix()
data['rep_info'][-1].append(
dict(feature_matrix=x.tolist(),
hypers=copy(sph_expn.hypers)))
with open(
path + "tests/reference_data/spherical_expansion_reference.ubjson",
'wb') as f:
ubjson.dump(data, f)
def dump_reference_json():
import ubjson
from copy import copy
from itertools import product
sys.path.insert(0, os.path.join(root, "build/"))
sys.path.insert(0, os.path.join(root, "tests/"))
cutoffs = [2, 3]
gaussian_sigmas = [0.2, 0.5]
max_radials = [4]
max_angulars = [3]
cutoff_smooth_widths = [1.0]
radial_basis = ["GTO", "DVR"]
cutoff_function_types = ["ShiftedCosine", "RadialScaling"]
fns = [
os.path.join(inputs_path, "CaCrP2O7_mvc-11955_symmetrized.json"),
os.path.join(inputs_path, "small_molecule.json"),
]
fns_to_write = [
os.path.join(read_inputs_path, "CaCrP2O7_mvc-11955_symmetrized.json"),
os.path.join(read_inputs_path, "small_molecule.json"),
]
optimization_args = [
{
"type": "None",
"accuracy": 1e-8
},
{
"type": "Spline",
"accuracy": 1e-8
},
]
data = dict(
filenames=fns_to_write,
cutoffs=cutoffs,
gaussian_sigmas=gaussian_sigmas,
max_radials=max_radials,
rep_info=[],
)
for fn in fns:
for cutoff in cutoffs:
data["rep_info"].append([])
for (
gaussian_sigma,
max_radial,
max_angular,
cutoff_smooth_width,
rad_basis,
cutoff_function_type,
opt_args,
) in product(
gaussian_sigmas,
max_radials,
max_angulars,
cutoff_smooth_widths,
radial_basis,
cutoff_function_types,
optimization_args,
):
frames = read(fn)
if cutoff_function_type == "RadialScaling":
cutoff_function_parameters = dict(rate=1,
scale=cutoff * 0.5,
exponent=3)
else:
cutoff_function_parameters = dict()
hypers = {
"interaction_cutoff": cutoff,
"cutoff_smooth_width": cutoff_smooth_width,
"max_radial": max_radial,
"max_angular": max_angular,
"gaussian_sigma_type": "Constant",
"cutoff_function_type": cutoff_function_type,
"cutoff_function_parameters": cutoff_function_parameters,
"gaussian_sigma_constant": gaussian_sigma,
"radial_basis": rad_basis,
"optimization_args": opt_args,
}
sph_expn = SphericalExpansion(**hypers)
expansions = sph_expn.transform(frames)
x = expansions.get_features(sph_expn)
x[np.abs(x) < 1e-300] = 0.0
data["rep_info"][-1].append(
dict(feature_matrix=x.tolist(),
hypers=copy(sph_expn.hypers)))
with open(
os.path.join(root, dump_path,
"spherical_expansion_reference.ubjson"),
"wb",
) as f:
ubjson.dump(data, f)
