def predict_E(self, structure):
energy = torch.zeros(1).double()
for count in range(len(structure.species)):
environment = ace.LocalEnvironment(structure, count, self.cutoff)
energy += self.predict_local_E(environment)
return energy
def predict_EFS(self, structure):
noa = len(structure.species)
efs_tens = torch.zeros(1 + 3 * noa + 6).double()
for count in range(noa):
environment = ace.LocalEnvironment(structure, count, self.cutoff)
efs_tens += self.predict_local_EFS(environment)
return efs_tens
def predict_F(self, structure):
noa = len(structure.species)
f_tens = torch.zeros(3 * noa).double()
for count in range(noa):
environment = ace.LocalEnvironment(structure, count, self.cutoff)
f_tens += self.predict_local_F(environment)
return f_tens
def predict_EF(self, structure):
# Initialize energy/force tensor.
noa = len(structure.species)
ef_tens = torch.zeros(1 + 3 * noa).double()
for count in range(noa):
environment = ace.LocalEnvironment(structure, count, self.cutoff)
ef_tens += self.predict_local_EF(environment)
return ef_tens
[-2.01399916, 1.26567364, -0.34566542],
[-1.24917847, 0.11058142, -0.35152255],
[-0.26895874, -1.24669592, 3.20771389],
[2.65635959, -1.73157442, 0.26207606],
[-0.27368027, 1.27614842, -2.37945504],
[3.57300086, -0.82058207, -1.6675233],
[2.19445749, 0.74663571, -3.1649488],
[-3.6649996, 1.36982664, 0.68209262],
[-3.84709602, 1.91009051, -0.95156392],
[-3.87331649, 0.10910619, -0.56318171]])
noa = len(positions)
test_struc = ace.Structure(cell, coded_species, positions)
cutoff = 7.
test_env = ace.LocalEnvironment(test_struc, 0, cutoff)
radial_basis = "chebyshev"
cutoff_function = "cosine"
radial_hyps = [0, cutoff]
cutoff_hyps = []
nos = 3
N = 2
lmax = 1
descriptor_settings = [nos, N, lmax]
descriptor = \
ace.DescriptorCalculator(radial_basis, cutoff_function, radial_hyps,
cutoff_hyps, descriptor_settings)
descriptor.compute_B2(test_env)