def train_data(images, setup_only=False):
label = 'nodeplot_test/calc'
train_images = images
calc = Amp(descriptor=Gaussian(),
model=NeuralNetwork(hiddenlayers=(5, 5)),
label=label,
cores=1)
loss = LossFunction(convergence={'energy_rmse': 0.02,
'force_rmse': 0.02})
calc.model.lossfunction = loss
if not setup_only:
calc.train(images=train_images, )
for image in train_images:
print "energy =", calc.get_potential_energy(image)
print "forces =", calc.get_forces(image)
else:
images = hash_images(train_images)
calc.descriptor.calculate_fingerprints(images=images,
log=calc.log,
cores=1,
calculate_derivatives=False)
calc.model.fit(trainingimages=images,
descriptor=calc.descriptor,
log=calc.log,
cores=1,
only_setup=True)
return calc
def test_gaussian():
label = 'noforce_test'
if not os.path.exists(label):
os.mkdir(label)
print('Generating data.')
images = generate_data(10)
print('Training Gaussian-neural network.')
calc = Amp(descriptor=Gaussian(),
label=os.path.join(label, 'Gaussian'),
regression=NeuralNetwork(hiddenlayers=(5, 5)))
calc.train(images, force_goal=None)
def calc_train_images(images, HL, E_conv, f_conv, f_coeff, ncore):
Hidden_Layer=tuple(HL)
print("Hidden Layer: {}".format(Hidden_Layer))
print("Energy convergence: {}".format(E_conv))
cores={'localhost':ncore} # 'localhost' depress SSH, communication between nodes
calc = Amp(descriptor=Gaussian(), model=NeuralNetwork(hiddenlayers=Hidden_Layer), cores=cores)
if f_conv <= 0.0:
convergence={'energy_rmse': E_conv}
else:
convergence={'energy_rmse': E_conv, 'force_rmse':f_conv}
calc.model.lossfunction = LossFunction(convergence=convergence, force_coefficient=f_coeff)
#calc.model.lossfunction = LossFunction(force_coefficient=-0.1)
calc.train(images=images, overwrite=True)
return
def train_test():
label = 'train_test/calc'
train_images = generate_data(2)
calc = Amp(descriptor=Gaussian(),
model=NeuralNetwork(hiddenlayers=(3, 3)),
label=label,
cores=1)
loss = LossFunction(convergence={'energy_rmse': 0.02,
'force_rmse': 0.02})
calc.model.lossfunction = loss
calc.train(images=train_images,)
for image in train_images:
print "energy =", calc.get_potential_energy(image)
print "forces =", calc.get_forces(image)
def test():
"FingerprintPlot test."""
generate_data(2, filename='fpplot-training.traj')
calc = Amp(descriptor=Gaussian(),
model=NeuralNetwork(),
label='fpplot-test'
)
calc.model.lossfunction = LossFunction(convergence={'energy_rmse': 1.00,
'force_rmse': 1.00})
calc.train(images='fpplot-training.traj')
images = ase.io.Trajectory('fpplot-training.traj')
fpplot = FingerprintPlot(calc)
fpplot(images)
fpplot(images, overlay=images[0])
fpplot(images, overlay=[images[1][2], images[0][-1]])
def calc_train_images(images, HL, E_conv, f_conv, ncore):
Hidden_Layer = tuple(HL)
print("Hidden Layer: {}".format(Hidden_Layer))
print("Energy convergence: {}".format(E_conv))
calc = Amp(descriptor=Gaussian(),
model=NeuralNetwork(hiddenlayers=Hidden_Layer),
cores=ncore)
if f_conv <= 0.0:
calc.model.lossfunction = LossFunction(
convergence={'energy_rmse': E_conv})
else:
calc.model.lossfunction = LossFunction(convergence={
'energy_rmse': E_conv,
'force_rmse': f_conv
})
#calc.model.lossfunction = LossFunction(force_coefficient=-0.1)
calc.train(images=images, overwrite=True)
return
def test_none():
label = 'force_test'
if not os.path.exists(label):
os.mkdir(label)
print('Generating data.')
all_images = generate_data(4)
train_images, test_images = randomize_images(all_images)
print('Training none-neural network.')
calc1 = Amp(descriptor=None,
label=os.path.join(label, 'none'),
regression=NeuralNetwork(hiddenlayers=(5, 5)))
calc1.train(train_images,
energy_goal=0.01,
force_goal=0.05,
global_search=None)
print('Testing none-neural network.')
energies1 = []
for image in all_images:
energies1.append(calc1.get_potential_energy(atoms=image))
print('Verify making new calc works.')
params = calc1.todict()
calc2 = Amp(**params)
energies2 = []
for image in all_images:
energies2.append(calc2.get_potential_energy(atoms=image))
assert energies1 == energies2
print('Verifying can move an atom and get new energy.')
image = all_images[0]
image.set_calculator(calc2)
e1 = image.get_potential_energy(apply_constraint=False)
f1 = image.get_forces(apply_constraint=False)
image[0].x += 0.5 # perturb
e2 = image.get_potential_energy(apply_constraint=False)
f2 = image.get_forces(apply_constraint=False)
assert e1 != e2
assert not (f1 == f2).all()
def calc_train_images(images, HL, E_conv, f_conv, f_coeff, ncore, amp_pot=None):
Hidden_Layer=tuple(HL)
print("Hidden Layer: {}".format(Hidden_Layer))
print("Energy convergence: {}".format(E_conv))
cores={'localhost':ncore} # 'localhost' depress SSH, communication between nodes
### load "amp.amp"
if amp_pot:
calc = Amp.load(amp_pot)
calc = Amp(descriptor=Gaussian(), model=NeuralNetwork(hiddenlayers=Hidden_Layer), cores=cores)
### Global Search in Param Space
Annealer(calc=calc, images=images, Tmax=20, Tmin=1, steps=4000)
if f_conv <= 0.0:
E_maxresid = E_conv*3
#convergence={'energy_rmse': E_conv}
convergence={'energy_rmse': E_conv, 'energy_maxresid': E_maxresid}
else:
convergence={'energy_rmse': E_conv, 'force_rmse':f_conv}
calc.model.lossfunction = LossFunction(convergence=convergence, force_coefficient=f_coeff) # setting
calc.train(images=images, overwrite=True)
return
def train_test():
"""Gaussian/tflow train test."""
perform, reason = check_perform()
if not perform:
print('Skipping this test because {}.'.format(reason))
return
from amp.model.tflow import NeuralNetwork
label = 'train_test/calc'
train_images = generate_data(2)
convergence = {'energy_rmse': 0.02, 'force_rmse': 0.02}
calc = Amp(descriptor=Gaussian(),
model=NeuralNetwork(hiddenlayers=(3, 3),
convergenceCriteria=convergence),
label=label,
cores=1)
calc.train(images=train_images, )
for image in train_images:
print("energy =", calc.get_potential_energy(image))
print("forces =", calc.get_forces(image))
def train_test():
"""Gaussian/Neural train test."""
label = 'train_test/calc'
train_images = generate_data(2)
calc = Amp(descriptor=Gaussian(),
model=NeuralNetwork(hiddenlayers=(3, 3)),
label=label,
cores=1)
loss = LossFunction(convergence={'energy_rmse': 0.02, 'force_rmse': 0.03})
calc.model.lossfunction = loss
calc.train(images=train_images, )
for image in train_images:
print("energy = %s" % str(calc.get_potential_energy(image)))
print("forces = %s" % str(calc.get_forces(image)))
# Test that we can re-load this calculator and call it again.
del calc
calc2 = Amp.load(label + '.amp')
for image in train_images:
print("energy = %s" % str(calc2.get_potential_energy(image)))
print("forces = %s" % str(calc2.get_forces(image)))
# -*- coding: utf-8 -*-
"""
Created on Sun Apr 12 21:54:34 2020
@author: srava
"""
from ase.io import read
from ase.io.trajectory import Trajectory
import numpy as np
from ase.calculators.emt import EMT
from amp import Amp
from amp.descriptor.gaussian import Gaussian
from amp.utilities import hash_images
from amp.model.neuralnetwork import NeuralNetwork
import matplotlib.pyplot as plt
calc = Amp(descriptor=Gaussian(), model=NeuralNetwork(),
label='calc')
calc.model.lossfunction.parameters['convergence'].update(
{'energy_rmse': 0.05,})
calc.train(images='training_data.traj')
def test():
pwd = os.getcwd()
os.mkdir(os.path.join(pwd, 'consistnone'))
images = generate_images()
count = 0
for global_search in [None, 'SA']:
for fortran in [False, True]:
for extend_variables in [False, True]:
for data_format in ['db', 'json']:
for save_memory in [False]:
for cores in range(1, 7):
string = 'consistnone/%s-%s-%s-%s-%s-%i'
label = string % (global_search, fortran,
extend_variables, data_format,
save_memory, cores)
if global_search is 'SA':
global_search = \
SimulatedAnnealing(temperature=10, steps=5)
calc = Amp(descriptor=None,
regression=NeuralNetwork(
hiddenlayers=(2, 1),
activation='tanh',
weights=weights,
scalings=scalings,
),
fortran=fortran,
label=label)
calc.train(images=images,
energy_goal=10.**10.,
force_goal=10.**10.,
cores=cores,
data_format=data_format,
save_memory=save_memory,
global_search=global_search,
extend_variables=extend_variables)
if count == 0:
reference_cost_function = calc.cost_function
reference_energy_rmse = \
calc.energy_per_atom_rmse
reference_force_rmse = calc.force_rmse
ref_cost_fxn_variable_derivatives = \
calc.der_variables_cost_function
else:
assert (abs(calc.cost_function -
reference_cost_function) <
10.**(-10.)), \
'''Cost function value for %r fortran, %r
data format, %r save_memory, and %i cores is
not consistent with the value of python version
on single core.''' % (fortran, data_format,
save_memory, cores)
assert (abs(calc.energy_per_atom_rmse -
reference_energy_rmse) <
10.**(-10.)), \
'''Energy rmse value for %r fortran, %r data
format, %r save_memory, and %i cores is not
consistent with the value of python version on
single core.''' % (fortran, data_format,
save_memory, cores)
assert (abs(calc.force_rmse -
reference_force_rmse) < 10.**(-10.)), \
'''Force rmse value for %r fortran, %r data
format, %r save_memory, and %i cores is not
consistent with the value of python version on
single core.''' % (fortran, data_format,
save_memory, cores)
for _ in range(
len(ref_cost_fxn_variable_derivatives)):
assert (calc.der_variables_cost_function[_] -
ref_cost_fxn_variable_derivatives[_] <
10.**(-10.))
'''Derivative of the cost function for %r
fortran, %r data format, %r save_memory, and %i
cores is not consistent with the value of
python version on single
core. ''' % (fortran, data_format, save_memory,
cores)
count = count + 1
#!/usr/bin/env python
from amp import Amp
from amp.descriptor import Behler
from amp.regression import NeuralNetwork
calc = Amp(label="./",
descriptor=Behler(cutoff=6.0),
regression=NeuralNetwork(hiddenlayers=(2, 16)))
calc.train("../train.db",
cores=12,
force_goal=None,
extend_variables=False)
def test():
"""Gaussian/Neural numeric-analytic consistency."""
images = generate_data()
regressor = Regressor(optimizer='BFGS')
_G = make_symmetry_functions(type='G2',
etas=[0.05, 5.],
elements=['Cu', 'Pt'])
_G += make_symmetry_functions(type='G4',
etas=[0.005],
zetas=[1., 4.],
gammas=[1.],
elements=['Cu', 'Pt'])
Gs = {'Cu': _G, 'Pt': _G}
calc = Amp(descriptor=Gaussian(Gs=Gs),
model=NeuralNetwork(
hiddenlayers=(2, 1),
regressor=regressor,
randomseed=42,
),
cores=1)
step = 0
for d in [None, 0.00001]:
for fortran in [True, False]:
for cores in [1, 2]:
step += 1
label = \
'numeric_analytic_test/analytic-%s-%i' % (fortran, cores) \
if d is None \
else 'numeric_analytic_test/numeric-%s-%i' \
% (fortran, cores)
print(label)
loss = LossFunction(convergence={
'energy_rmse': 10**10,
'force_rmse': 10**10
},
d=d)
calc.set_label(label)
calc.dblabel = 'numeric_analytic_test/analytic-True-1'
calc.model.lossfunction = loss
calc.descriptor.fortran = fortran
calc.model.fortran = fortran
calc.cores = cores
calc.train(images=images, )
if step == 1:
ref_energies = []
ref_forces = []
for image in images:
ref_energies += [calc.get_potential_energy(image)]
ref_forces += [calc.get_forces(image)]
ref_dloss_dparameters = \
calc.model.lossfunction.dloss_dparameters
else:
energies = []
forces = []
for image in images:
energies += [calc.get_potential_energy(image)]
forces += [calc.get_forces(image)]
dloss_dparameters = \
calc.model.lossfunction.dloss_dparameters
for image_no in range(2):
diff = abs(energies[image_no] - ref_energies[image_no])
assert (diff < 10.**(-13.)), \
'The calculated value of energy of image %i is ' \
'wrong!' % (image_no + 1)
for atom_no in range(len(images[0])):
for i in range(3):
diff = abs(forces[image_no][atom_no][i] -
ref_forces[image_no][atom_no][i])
assert (diff < 10.**(-10.)), \
'The calculated %i force of atom %i of ' \
'image %i is wrong!' \
% (i, atom_no, image_no + 1)
# Checks analytical and numerical dloss_dparameters
for _ in range(len(ref_dloss_dparameters)):
diff = abs(dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-10.)), \
'The calculated value of loss function ' \
'derivative is wrong!'
# Checks analytical and numerical forces
forces = []
for image in images:
image.set_calculator(calc)
forces += [calc.calculate_numerical_forces(image, d=d)]
for atom_no in range(len(images[0])):
for i in range(3):
diff = abs(forces[image_no][atom_no][i] -
ref_forces[image_no][atom_no][i])
print('{:3d} {:1d} {:7.1e}'.format(atom_no, i, diff))
assert (diff < 10.**(-6.)), \
'The calculated %i force of atom %i of ' \
'image %i is wrong! (Diff = %f)' \
% (i, atom_no, image_no + 1, diff)
from amp import Amp
from amp.descriptor.gaussian import Gaussian
from amp.model.neuralnetwork import NeuralNetwork
from amp.model import LossFunction
from amp import analysis
import ase
from ase import io
import sys
calc = Amp(descriptor=Gaussian(), model=NeuralNetwork(), label="calc")
calc.model.lossfunction = LossFunction(convergence={
'energy_rmse': 0.02,
'force_rmse': 0.02
})
images = ase.io.read("water.extxyz", ":")
# images = ase.io.read("../datasets/COCu/COCu_pbc_500.traj", ":")
IMAGES = []
for i in range(100):
IMAGES.append(images[i])
calc.train(IMAGES)
# analysis.plot_parity("calc.amp", IMAGES, overwrite=True)
def train_amp(baseframe=200,
traj='ethane.traj',
convergence={
'energy_rmse': 0.25,
'force_rmse': 0.5
},
elements=['C', 'H', 'O'],
cores=4):
"""Gaussian/tflow train test."""
p = ple()
label = 'amp'
all_images = Trajectory(traj)
nimg, mean_e = get_mean_energy(all_images)
G = make_symmetry_functions(elements=elements,
type='G2',
etas=np.logspace(np.log10(0.05),
np.log10(5.),
num=4))
G += make_symmetry_functions(elements=elements,
type='G5',
etas=[0.005],
zetas=[1., 4.],
gammas=[+1., -1.])
G = {element: G for element in elements} # Gs=G
if not isfile('amp.amp'):
# print('\nset up calculator ...\n')
calc = Amp(descriptor=Gaussian(mode='atom-centered', Gs=G),
model=NeuralNetwork(hiddenlayers=(1024, 1024, 1024, 512,
512, 256, 256, 256, 256,
128, 128),
convergenceCriteria=convergence,
activation='tanh',
energy_coefficient=1.0,
force_coefficient=None,
optimizationMethod='ADAM',
parameters={'energyMeanScale': mean_e},
maxTrainingEpochs=100000),
label=label,
cores=cores) # 'l-BFGS-b' or 'ADAM'
trained_images = [all_images[j] for j in range(0, baseframe)]
calc.train(overwrite=True, images=trained_images)
del calc
else:
calc = Amp.load('amp.amp')
calc.model.parameters['convergence'] = convergence
calc.model.lossfunction = LossFunction(convergence=convergence)
trained_images = [all_images[j] for j in range(0, baseframe)]
calc.train(overwrite=True, images=trained_images)
del calc
edfts, eamps, eamps_ = [], [], []
dolabel = True
basestep = int(baseframe / tframe)
system('epstopdf energies.eps')
p.scatter(x, edft, eamp, eamp_, dolabel=dolabel)
p.plot()
plot_energies(edfts, eamps, eamp_=eamps_)
system('epstopdf energies_scatter.eps')
images = [atoms]
g2_etas = [0.005]
g2_rs_s = [0] * 4
g4_etas = [0.005]
g4_zetas = [1., 4.]
g4_gammas = [1., -1.]
cutoff = 4
make_amp_descriptors_simple_nn(images, g2_etas, g2_rs_s, g4_etas, g4_zetas,
g4_gammas, cutoff)
G = make_symmetry_functions(elements=elements, type='G2', etas=g2_etas)
#Add Rs parameter (0.0 for default) to comply with my version of AMP
#for g in G:
# g['Rs'] = 0.0
G += make_symmetry_functions(elements=elements,
type='G4',
etas=g4_etas,
zetas=g4_zetas,
gammas=g4_gammas)
calc = Amp(descriptor=Gaussian(Gs=G, cutoff=4.),
cores=ncores,
model=NeuralNetwork(hiddenlayers=hiddenlayers))
calc.model.lossfunction = LossFunction(convergence=convergence,
force_coefficient=0.001)
calc.train(images=images)
#!/usr/bin/env python
from amp import Amp
from amp.descriptor import Behler
from amp.regression import NeuralNetwork
calc = Amp(label="./",
descriptor=Behler(cutoff=6.5),
regression=NeuralNetwork(hiddenlayers=(2, 7)))
calc.train("../train.db",
cores=16,
energy_goal=0.0005,
force_goal=None,
extend_variables=False)
{"type":"G2", "element":"Pd", "eta":400., "Rs":4.0},
{"type":"G2", "element":"Pd", "eta":400., "Rs":5.0},
{"type":"G2", "element":"Pd", "eta":400., "Rs":6.0},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":-1.0, "zeta":100.0, "theta_s":-1.0},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":-1.0, "zeta":100.0, "theta_s":-1.5},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":-1.0, "zeta":100.0, "theta_s":-1.75},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":-1.0, "zeta":100.0, "theta_s":-2.0},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":-1.0, "zeta":300.0, "theta_s":-2.15},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":-1.0, "zeta":300.0, "theta_s":-2.25},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":1.0, "zeta":160.0, "theta_s":0.0},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":1.0, "zeta":320.0, "theta_s":0.25},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":1.0, "zeta":640.0, "theta_s":0.35},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":1.0, "zeta":160.0, "theta_s":0.5},
{"type":"G4", "elements":["Pd", "Pd"],"eta":0.005, "gamma":1.0, "zeta":160.0, "theta_s":0.75},
]}
calc = Amp(descriptor=Gaussian(Gs=Gs,
cutoff=Cosine(6.0)
),
cores=24,
model=NeuralNetwork(hiddenlayers=(50,), activation='sigmoid',
maxTrainingEpochs=20000,
energy_coefficient=1.0,
force_coefficient=0.1,
optimizationMethod='l-BFGS-b',
convergenceCriteria={'energy_rmse': 0.05,
'force_rmse': 0.05}))
calc.train(images='train.traj',overwrite=True)
def test():
pwd = os.getcwd()
os.mkdir(os.path.join(pwd, 'rotation_test'))
for descriptor in [Gaussian(), Bispectrum(jmax=2.), Zernike(nmax=5)]:
pwd = os.getcwd()
os.mkdir(os.path.join(pwd, 'rotation_test/before_rot'))
os.mkdir(os.path.join(pwd, 'rotation_test/after_rot'))
# Non-rotated atomic configuration
atoms = Atoms([
Atom('Pt', (0., 0., 0.)),
Atom('Pt', (0., 0., 1.)),
Atom('Pt', (0., 2., 1.))
])
atoms.set_constraint(FixAtoms(indices=[0]))
atoms.set_calculator(EMT())
atoms.get_potential_energy()
atoms.get_forces(apply_constraint=False)
calc = Amp(descriptor=descriptor,
fortran=False,
label='rotation_test/before_rot/')
calc.train(images=[atoms],
force_goal=None,
energy_goal=10.**10.,
extend_variables=False,
global_search=None,
data_format='json')
###########################################################################
# Randomly Rotated (and translated) atomic configuration
rot = [random.random(), random.random(), random.random()]
for i in range(1, len(atoms)):
(atoms[i].x, atoms[i].y,
atoms[i].z) = rotate_atom(atoms[i].x, atoms[i].y, atoms[i].z,
rot[0] * np.pi, rot[1] * np.pi,
rot[2] * np.pi)
disp = [random.random(), random.random(), random.random()]
for atom in atoms:
atom.x += disp[0]
atom.y += disp[1]
atom.z += disp[2]
calc = Amp(descriptor=descriptor,
fortran=False,
label='rotation_test/after_rot/')
calc.train(images=[atoms],
force_goal=None,
energy_goal=10.**10.,
extend_variables=False,
global_search=None,
data_format='json')
###########################################################################
fp1 = paropen('rotation_test/before_rot/fingerprints.json', 'rb')
nonrotated_data = json.load(fp1)
fp2 = paropen('rotation_test/after_rot/fingerprints.json', 'rb')
rotated_data = json.load(fp2)
for hash1, hash2 in zip(nonrotated_data.keys(), rotated_data.keys()):
for index in nonrotated_data[hash1]:
for _ in range(len(nonrotated_data[hash1][index])):
assert abs(nonrotated_data[hash1][index][_] -
rotated_data[hash2][index][_]) < 10.**(-7.)
shutil.rmtree('rotation_test/before_rot')
shutil.rmtree('rotation_test/after_rot')
else:
label = label
dblabel = dblabel
desc = Gaussian(cutoff=cutoff)
model = NeuralNetwork(hiddenlayers=framework)
calc = Amp(label=label,
dblabel=dblabel,
descriptor=desc,
regression=model)
{% if optimizer %}
# Change optimization algorithm
from amp.regression import Regressor
from scipy.optimize import {{ optimizer }}
regressor = Regressor(optimizer={{ optimizer }})
calc.model.regressor = regressor
{% endif %}
# Train the network
calc.train(images=images,
data_format='db',
cores=cores,
energy_goal=energy_rmse,
force_goal=force_rmse,
global_search=SimulatedAnnealing(temperature=70,
steps=50),
extend_variables=False)
def train_images(images):
calc = Amp(descriptor=Gaussian(), model=NeuralNetwork(hiddenlayers=(10, 10, 10)))
calc.model.lossfunction = LossFunction(convergence={'energy_rmse': 0.001})
calc.model.lossfunction = LossFunction(force_coefficient=-0.1)
calc.train(images=images, overwrite=True)
os.chdir(work_dir)
# Get atoms from the database
images = []
db = connect(db_path)
for d in db.select(['iteration<={}'.format(iteration), 'train_set=True']):
atoms = db.get_atoms(d.id)
del atoms.constraints
images.append(atoms)
# Build Amp object
framework_str = "-".join([str(f) for f in framework])
label = label
dblabel = dblabel
desc = Gaussian(cutoff=cutoff)
model = NeuralNetwork(hiddenlayers=framework)
calc = Amp(label=label,
dblabel=dblabel,
descriptor=desc,
model=model,
cores=cores)
loss = LossFunction(convergence={'energy_rmse': energy_rmse,
'force_rmse': force_rmse})
calc.model.lossfunction = loss
# Perform simulated annealing for global search
Annealer(calc=calc, images=images)
# Train the network
calc.train(images=images)
#import os
#from ase import Atoms, Atom, units
#import ase.io
from amp import Amp
from amp.descriptor.gaussian import Gaussian
from amp.model.neuralnetwork import NeuralNetwork
from amp.model import LossFunction
calc = Amp(descriptor=Gaussian(),
model=NeuralNetwork(hiddenlayers=(10, 10, 10)))
calc.model.lossfunction = LossFunction(convergence={
'energy_rmse': 0.02,
'force_rmse': 0.03
})
calc.train(images='geoopt_LCAO.traj')
def non_periodic_0th_bfgs_step_test():
pwd = os.getcwd()
os.mkdir(os.path.join(pwd, 'CuOPdbp'))
os.mkdir(os.path.join(pwd, '_CuOPdbp'))
os.mkdir(os.path.join(pwd, 'CuOPdbp/0'))
os.mkdir(os.path.join(pwd, '_CuOPdbp/0'))
os.mkdir(os.path.join(pwd, 'CuOPdbp/1'))
os.mkdir(os.path.join(pwd, '_CuOPdbp/1'))
os.mkdir(os.path.join(pwd, 'CuOPdbp/2'))
os.mkdir(os.path.join(pwd, '_CuOPdbp/2'))
###########################################################################
# Making the list of periodic image
images = [
Atoms(symbols='PdOPd2',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[0., 0., 0.], [0., 2., 0.], [0., 0., 3.],
[1., 0., 0.]])),
Atoms(symbols='PdOPd2',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[0., 1., 0.], [1., 2., 1.], [-1., 1., 2.],
[1., 3., 2.]])),
Atoms(symbols='PdO',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[2., 1., -1.], [1., 2., 1.]])),
Atoms(symbols='Pd2O',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[-2., -1., -1.], [1., 2., 1.], [3., 4.,
4.]])),
Atoms(symbols='Cu',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[0., 0., 0.]]))
]
for image in images:
image.set_calculator(EMT())
image.get_potential_energy(apply_constraint=False)
image.get_forces(apply_constraint=False)
###########################################################################
# Parameters
Gs = {
'O': [{
'type': 'G2',
'element': 'Pd',
'eta': 0.8
}, {
'type': 'G4',
'elements': ['Pd', 'Pd'],
'eta': 0.2,
'gamma': 0.3,
'zeta': 1
}, {
'type': 'G4',
'elements': ['O', 'Pd'],
'eta': 0.3,
'gamma': 0.6,
'zeta': 0.5
}],
'Pd': [{
'type': 'G2',
'element': 'Pd',
'eta': 0.2
}, {
'type': 'G4',
'elements': ['Pd', 'Pd'],
'eta': 0.9,
'gamma': 0.75,
'zeta': 1.5
}, {
'type': 'G4',
'elements': ['O', 'Pd'],
'eta': 0.4,
'gamma': 0.3,
'zeta': 4
}],
'Cu': [{
'type': 'G2',
'element': 'Cu',
'eta': 0.8
}, {
'type': 'G4',
'elements': ['Cu', 'O'],
'eta': 0.2,
'gamma': 0.3,
'zeta': 1
}, {
'type': 'G4',
'elements': ['Cu', 'Cu'],
'eta': 0.3,
'gamma': 0.6,
'zeta': 0.5
}]
}
hiddenlayers = {'O': (2), 'Pd': (2), 'Cu': (2)}
weights = OrderedDict([
('O',
OrderedDict([(1,
np.matrix([[-2.0, 6.0], [3.0, -3.0], [1.5, -0.9],
[-2.5, -1.5]])),
(2, np.matrix([[5.5], [3.6], [1.4]]))])),
('Pd',
OrderedDict([(1,
np.matrix([[-1.0, 3.0], [2.0, 4.2], [1.0, -0.7],
[-3.0, 2.0]])),
(2, np.matrix([[4.0], [0.5], [3.0]]))])),
('Cu',
OrderedDict([(1,
np.matrix([[0.0, 1.0], [-1.0, -2.0], [2.5, -1.9],
[-3.5, 0.5]])),
(2, np.matrix([[0.5], [1.6], [-1.4]]))]))
])
scalings = OrderedDict([
('O', OrderedDict([('intercept', -2.3), ('slope', 4.5)])),
('Pd', OrderedDict([('intercept', 1.6), ('slope', 2.5)])),
('Cu', OrderedDict([('intercept', -0.3), ('slope', -0.5)]))
])
###########################################################################
# Correct values
correct_cost = 7144.810783950215
correct_energy_rmse = 24.318837496017185
correct_force_rmse = 144.70282475062052
correct_der_cost_fxn = [
0, 0, 0, 0, 0, 0, 0.01374139170953901, 0.36318423812749656,
0.028312691567496464, 0.6012336354445753, 0.9659002689921986,
-1.2897770059416218, -0.5718960935176884, -2.6425667221503035,
-1.1960399246712894, 0, 0, -2.7256379713943852, -0.9080181026559658,
-0.7739948323247023, -0.2915789426043727, -2.05998290443513,
-0.6156374289747903, -0.0060865174621348985, -0.8296785483640939,
0.0008092646748983969, 0.041613027034688874, 0.003426469079592851,
-0.9578004568876517, -0.006281929608090211, -0.28835884773094056,
-4.2457774110285245, -4.317412094174614, -8.02385959091948,
-3.240512651984099, -27.289862194996896, -26.8177742762254,
-82.45107056053345, -80.6816768350809
]
###########################################################################
# Testing pure-python and fortran versions of Gaussian-neural on different
# number of processes
for global_search in [None, 'SA']:
for fortran in [False, True]:
for extend_variables in [False, True]:
for data_format in ['db', 'json']:
for save_memory in [False]:
for cores in range(1, 7):
string = 'CuOPdbp/0/%s-%s-%s-%s-%s-%i'
label = string % (global_search, fortran,
extend_variables, data_format,
save_memory, cores)
if global_search is 'SA':
gs = \
SimulatedAnnealing(temperature=10, steps=5)
elif global_search is None:
gs = None
print label
calc = Amp(descriptor=Gaussian(cutoff=6.5, Gs=Gs),
regression=NeuralNetwork(
hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='sigmoid',
),
fortran=fortran,
label=label)
calc.train(images=images,
energy_goal=10.**10.,
force_goal=10.**10.,
force_coefficient=0.04,
cores=cores,
data_format=data_format,
save_memory=save_memory,
global_search=gs,
extend_variables=extend_variables)
assert (abs(calc.cost_function - correct_cost) <
10.**(-5.)), \
'The calculated value of cost function is \
wrong!'
assert (abs(calc.energy_per_atom_rmse -
correct_energy_rmse) <
10.**(-10.)), \
'The calculated value of energy per atom RMSE \
is wrong!'
assert (abs(calc.force_rmse - correct_force_rmse) <
10 ** (-7)), \
'The calculated value of force RMSE is wrong!'
for _ in range(len(correct_der_cost_fxn)):
assert(abs(calc.der_variables_cost_function[
_] - correct_der_cost_fxn[_]) <
10 ** (-9)), \
'The calculated value of cost function \
derivative is wrong!'
dblabel = label
secondlabel = '_' + label
calc = Amp(descriptor=Gaussian(cutoff=6.5, Gs=Gs),
regression=NeuralNetwork(
hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='sigmoid',
),
fortran=fortran,
label=secondlabel,
dblabel=dblabel)
calc.train(images=images,
energy_goal=10.**10.,
force_goal=10.**10.,
force_coefficient=0.04,
cores=cores,
data_format=data_format,
save_memory=save_memory,
global_search=gs,
extend_variables=extend_variables)
assert (abs(calc.cost_function - correct_cost) <
10.**(-5.)), \
'The calculated value of cost function is \
wrong!'
assert (abs(calc.energy_per_atom_rmse -
correct_energy_rmse) <
10.**(-10.)), \
'The calculated value of energy per atom RMSE \
is wrong!'
assert (abs(calc.force_rmse - correct_force_rmse) <
10 ** (-7)), \
'The calculated value of force RMSE is wrong!'
for _ in range(len(correct_der_cost_fxn)):
assert(abs(calc.der_variables_cost_function[
_] - correct_der_cost_fxn[_] <
10 ** (-9))), \
'The calculated value of cost function \
def test():
"""Guassian/Neural training.
Checks consistency of pure-python and fortran versions.
"""
images = make_images()
convergence = {'energy_rmse': 10.**10.,
'energy_maxresid': 10.**10.,
'force_rmse': 10.**10.,
'force_maxresid': 10.**10., }
regressor = Regressor(optimizer='BFGS')
count = 0
for fortran in [False, True]:
for cores in range(1, 2):
string = 'consistgauss/%s-%i'
label = string % (fortran, cores)
calc = Amp(descriptor=Gaussian(cutoff=cutoff,
Gs=Gs,
fortran=fortran,),
model=NeuralNetwork(hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation=activation,
fprange=fingerprints_range,
regressor=regressor,),
label=label,
cores=1)
lossfunction = LossFunction(convergence=convergence)
calc.model.lossfunction = lossfunction
calc.train(images=images,)
if count == 0:
ref_loss = calc.model.lossfunction.loss
ref_energy_loss = calc.model.lossfunction.energy_loss
ref_force_loss = calc.model.lossfunction.force_loss
ref_dloss_dparameters = \
calc.model.lossfunction.dloss_dparameters
else:
assert (abs(calc.model.lossfunction.loss -
ref_loss) < 10.**(-10.)), \
'''Loss function value for %r fortran, and %i cores is
not consistent with the value of python version
on single core.''' % (fortran, cores)
assert (abs(calc.model.lossfunction.energy_loss -
ref_energy_loss) <
10.**(-9.)), \
'''Energy rmse value for %r fortran, and %i cores is not
consistent with the value of python version on
single core.''' % (fortran, cores)
assert (abs(calc.model.lossfunction.force_loss -
ref_force_loss) <
10.**(-9.)), \
'''Force rmse value for %r fortran, and %i cores is not
consistent with the value of python version on
single core.''' % (fortran, cores)
for _ in range(len(ref_dloss_dparameters)):
assert (calc.model.lossfunction.dloss_dparameters[_] -
ref_dloss_dparameters[_] < 10.**(-10.))
'''Derivative of the cost function for %r
fortran, and %i
cores is not consistent with the value of
python version on single
core. ''' % (fortran, cores)
count = count + 1
def non_periodic_0th_bfgs_step_test():
# Making the list of periodic image
images = [Atoms(symbols='PdOPd2',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array(
[[1., 0., 0.],
[0., 1., 0.],
[0., 0., 1.]]),
positions=np.array(
[[0., 0., 0.],
[0., 2., 0.],
[0., 0., 3.],
[1., 0., 0.]])),
Atoms(symbols='PdOPd2',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array(
[[1., 0., 0.],
[0., 1., 0.],
[0., 0., 1.]]),
positions=np.array(
[[0., 1., 0.],
[1., 2., 1.],
[-1., 1., 2.],
[1., 3., 2.]])),
Atoms(symbols='PdO',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array(
[[1., 0., 0.],
[0., 1., 0.],
[0., 0., 1.]]),
positions=np.array(
[[2., 1., -1.],
[1., 2., 1.]])),
Atoms(symbols='Pd2O',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array(
[[1., 0., 0.],
[0., 1., 0.],
[0., 0., 1.]]),
positions=np.array(
[[-2., -1., -1.],
[1., 2., 1.],
[3., 4., 4.]])),
Atoms(symbols='Cu',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array(
[[1., 0., 0.],
[0., 1., 0.],
[0., 0., 1.]]),
positions=np.array(
[[0., 0., 0.]]))]
for image in images:
image.set_calculator(EMT())
image.get_potential_energy(apply_constraint=False)
image.get_forces(apply_constraint=False)
# Parameters
Gs = {'O': [{'type': 'G2', 'element': 'Pd', 'eta': 0.8},
{'type': 'G4', 'elements': [
'Pd', 'Pd'], 'eta':0.2, 'gamma':0.3, 'zeta':1},
{'type': 'G4', 'elements': ['O', 'Pd'], 'eta':0.3, 'gamma':0.6,
'zeta':0.5}],
'Pd': [{'type': 'G2', 'element': 'Pd', 'eta': 0.2},
{'type': 'G4', 'elements': ['Pd', 'Pd'],
'eta':0.9, 'gamma':0.75, 'zeta':1.5},
{'type': 'G4', 'elements': ['O', 'Pd'], 'eta':0.4,
'gamma':0.3, 'zeta':4}],
'Cu': [{'type': 'G2', 'element': 'Cu', 'eta': 0.8},
{'type': 'G4', 'elements': ['Cu', 'O'],
'eta':0.2, 'gamma':0.3, 'zeta':1},
{'type': 'G4', 'elements': ['Cu', 'Cu'], 'eta':0.3,
'gamma':0.6, 'zeta':0.5}]}
hiddenlayers = {'O': (2,), 'Pd': (2,), 'Cu': (2,)}
weights = OrderedDict([('O', OrderedDict([(1, np.matrix([[-2.0, 6.0],
[3.0, -3.0],
[1.5, -0.9],
[-2.5, -1.5]])),
(2, np.matrix([[5.5],
[3.6],
[1.4]]))])),
('Pd', OrderedDict([(1, np.matrix([[-1.0, 3.0],
[2.0, 4.2],
[1.0, -0.7],
[-3.0, 2.0]])),
(2, np.matrix([[4.0],
[0.5],
[3.0]]))])),
('Cu', OrderedDict([(1, np.matrix([[0.0, 1.0],
[-1.0, -2.0],
[2.5, -1.9],
[-3.5, 0.5]])),
(2, np.matrix([[0.5],
[1.6],
[-1.4]]))]))])
scalings = OrderedDict([('O', OrderedDict([('intercept', -2.3),
('slope', 4.5)])),
('Pd', OrderedDict([('intercept', 1.6),
('slope', 2.5)])),
('Cu', OrderedDict([('intercept', -0.3),
('slope', -0.5)]))])
# Correct values
ref_loss = 7144.8107853579895
ref_energyloss = (24.318837496016506 ** 2.) * 5
ref_forceloss = (144.70282477494519 ** 2.) * 5
ref_dloss_dparameters = np.array([0, 0, 0, 0, 0, 0,
0.01374139170953901,
0.36318423812749656,
0.028312691567496464,
0.6012336354445753,
0.9659002689921986,
-1.289777005924742,
-0.5718960934643078,
-2.642566722179569,
-1.196039924610482, 0, 0,
-2.72563797131018,
-0.9080181024866707,
-0.7739948323226851,
-0.29157894253717415,
-2.0599829042717404,
-0.6156374289895887,
-0.006086517460749253,
-0.829678548408266,
0.0008092646745710161,
0.04161302703491613,
0.0034264690790135606,
-0.957800456897051,
-0.006281929606579444,
-0.2883588477371198,
-4.245777410962108,
-4.3174120941045535,
-8.02385959091948,
-3.240512651984099,
-27.289862194988853,
-26.8177742762544,
-82.45107056051073,
-80.68167683508715])
ref_energy_maxresid = 54.21915548269209
ref_force_maxresid = 791.6736436232306
# Testing pure-python and fortran versions of Gaussian-neural on different
# number of processes
for fortran in [False, True]:
for cores in range(1, 6):
label = 'train-nonperiodic/%s-%i' % (fortran, cores)
print label
calc = Amp(descriptor=Gaussian(cutoff=6.5,
Gs=Gs,
fortran=fortran,),
model=NeuralNetwork(hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='sigmoid',
regressor=regressor,
fortran=fortran,),
label=label,
dblabel=label,
cores=cores)
lossfunction = LossFunction(convergence=convergence)
calc.model.lossfunction = lossfunction
calc.train(images=images,)
diff = abs(calc.model.lossfunction.loss - ref_loss)
assert (diff < 10.**(-10.)), \
'Calculated value of loss function is wrong!'
diff = abs(calc.model.lossfunction.energy_loss - ref_energyloss)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom RMSE is wrong!'
diff = abs(calc.model.lossfunction.force_loss - ref_forceloss)
assert (diff < 10 ** (-10.)), \
'Calculated value of force RMSE is wrong!'
diff = abs(calc.model.lossfunction.energy_maxresid -
ref_energy_maxresid)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom max residual is wrong!'
diff = abs(calc.model.lossfunction.force_maxresid -
ref_force_maxresid)
assert (diff < 10 ** (-10.)), \
'Calculated value of force max residual is wrong!'
for _ in range(len(ref_dloss_dparameters)):
diff = abs(calc.model.lossfunction.dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-12.)), \
"Calculated value of loss function derivative is wrong!"
dblabel = label
secondlabel = '_' + label
calc = Amp(descriptor=Gaussian(cutoff=6.5,
Gs=Gs,
fortran=fortran,),
model=NeuralNetwork(hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='sigmoid',
regressor=regressor,
fortran=fortran,),
label=secondlabel,
dblabel=dblabel,
cores=cores)
lossfunction = LossFunction(convergence=convergence)
calc.model.lossfunction = lossfunction
calc.train(images=images,)
diff = abs(calc.model.lossfunction.loss - ref_loss)
assert (diff < 10.**(-10.)), \
'Calculated value of loss function is wrong!'
diff = abs(calc.model.lossfunction.energy_loss - ref_energyloss)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom RMSE is wrong!'
diff = abs(calc.model.lossfunction.force_loss - ref_forceloss)
assert (diff < 10 ** (-10.)), \
'Calculated value of force RMSE is wrong!'
diff = abs(calc.model.lossfunction.energy_maxresid -
ref_energy_maxresid)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom max residual is wrong!'
diff = abs(calc.model.lossfunction.force_maxresid -
ref_force_maxresid)
assert (diff < 10 ** (-10.)), \
'Calculated value of force max residual is wrong!'
for _ in range(len(ref_dloss_dparameters)):
diff = abs(calc.model.lossfunction.dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-12.)), \
'Calculated value of loss function derivative is wrong!'
#!/usr/bin/env python
from amp import Amp
from amp.descriptor import Gaussian
from amp.regression import NeuralNetwork
from ase.db import connect
from amp import SimulatedAnnealing
db = connect('../../../database/master.db')
images = []
for d in db.select('train_set=True'):
atoms = db.get_atoms(d.id)
del atoms.constraints
images += [atoms]
for n in [2, 3]:
calc = Amp(label='./',
dblabel='../../',
descriptor=Gaussian(cutoff=6.5),
regression=NeuralNetwork(hiddenlayers=(2, n)))
calc.train(images=images,
data_format='db',
cores=4,
energy_goal=1e-2,
force_goal=1e-1,
global_search=SimulatedAnnealing(temperature=70, steps=50),
extend_variables=False)
def non_periodic_0th_bfgs_step_test():
"""Gaussian/Neural training non-periodic standard test.
Compares results to that expected from separate mathematica
calculations.
"""
images = [
Atoms(symbols='PdOPd2',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[0., 0., 0.], [0., 2., 0.], [0., 0., 3.],
[1., 0., 0.]])),
Atoms(symbols='PdOPd2',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[0., 1., 0.], [1., 2., 1.], [-1., 1., 2.],
[1., 3., 2.]])),
Atoms(symbols='PdO',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[2., 1., -1.], [1., 2., 1.]])),
Atoms(symbols='Pd2O',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[-2., -1., -1.], [1., 2., 1.], [3., 4.,
4.]])),
Atoms(symbols='Cu',
pbc=np.array([False, False, False], dtype=bool),
cell=np.array([[1., 0., 0.], [0., 1., 0.], [0., 0., 1.]]),
positions=np.array([[0., 0., 0.]]))
]
for image in images:
image.set_calculator(EMT())
image.get_potential_energy(apply_constraint=False)
image.get_forces(apply_constraint=False)
# Parameters
Gs = {
'O': [{
'type': 'G2',
'element': 'Pd',
'eta': 0.8
}, {
'type': 'G4',
'elements': ['Pd', 'Pd'],
'eta': 0.2,
'gamma': 0.3,
'zeta': 1
}, {
'type': 'G4',
'elements': ['O', 'Pd'],
'eta': 0.3,
'gamma': 0.6,
'zeta': 0.5
}],
'Pd': [{
'type': 'G2',
'element': 'Pd',
'eta': 0.2
}, {
'type': 'G4',
'elements': ['Pd', 'Pd'],
'eta': 0.9,
'gamma': 0.75,
'zeta': 1.5
}, {
'type': 'G4',
'elements': ['O', 'Pd'],
'eta': 0.4,
'gamma': 0.3,
'zeta': 4
}],
'Cu': [{
'type': 'G2',
'element': 'Cu',
'eta': 0.8
}, {
'type': 'G4',
'elements': ['Cu', 'O'],
'eta': 0.2,
'gamma': 0.3,
'zeta': 1
}, {
'type': 'G4',
'elements': ['Cu', 'Cu'],
'eta': 0.3,
'gamma': 0.6,
'zeta': 0.5
}]
}
hiddenlayers = {'O': (2, ), 'Pd': (2, ), 'Cu': (2, )}
weights = OrderedDict([
('O',
OrderedDict([(1,
np.matrix([[-2.0, 6.0], [3.0, -3.0], [1.5, -0.9],
[-2.5, -1.5]])),
(2, np.matrix([[5.5], [3.6], [1.4]]))])),
('Pd',
OrderedDict([(1,
np.matrix([[-1.0, 3.0], [2.0, 4.2], [1.0, -0.7],
[-3.0, 2.0]])),
(2, np.matrix([[4.0], [0.5], [3.0]]))])),
('Cu',
OrderedDict([(1,
np.matrix([[0.0, 1.0], [-1.0, -2.0], [2.5, -1.9],
[-3.5, 0.5]])),
(2, np.matrix([[0.5], [1.6], [-1.4]]))]))
])
scalings = OrderedDict([
('O', OrderedDict([('intercept', -2.3), ('slope', 4.5)])),
('Pd', OrderedDict([('intercept', 1.6), ('slope', 2.5)])),
('Cu', OrderedDict([('intercept', -0.3), ('slope', -0.5)]))
])
# Correct values
if aseversion < 12: # EMT values have changed from 3.12.0 version
ref_loss = 7144.8107853579895
ref_energyloss = (24.318837496016506**2.) * 5
ref_forceloss = (144.70282477494519**2.) * 5
ref_dloss_dparameters = np.array([
0, 0, 0, 0, 0, 0, 0.01374139170953901, 0.36318423812749656,
0.028312691567496464, 0.6012336354445753, 0.9659002689921986,
-1.289777005924742, -0.5718960934643078, -2.642566722179569,
-1.196039924610482, 0, 0, -2.72563797131018, -0.9080181024866707,
-0.7739948323226851, -0.29157894253717415, -2.0599829042717404,
-0.6156374289895887, -0.006086517460749253, -0.829678548408266,
0.0008092646745710161, 0.04161302703491613, 0.0034264690790135606,
-0.957800456897051, -0.006281929606579444, -0.2883588477371198,
-4.245777410962108, -4.3174120941045535, -8.02385959091948,
-3.240512651984099, -27.289862194988853, -26.8177742762544,
-82.45107056051073, -80.68167683508715
])
ref_energy_maxresid = 54.21915548269209
ref_force_maxresid = 791.6736436232306
else:
ref_loss = 7144.807220773296
ref_energyloss = (24.318829702548342**2.) * 5
ref_forceloss = (144.70279593472887**2.) * 5
ref_dloss_dparameters = np.array([
0, 0, 0, 0, 0, 0, 0.01374139170953901, 0.36318423812749656,
0.028312691567496464, 0.6012336354445753, 0.9659002689921986,
-1.2897765357544038, -0.5718958286530584, -2.642565840915077,
-1.1960394346870424, 0, 0, -2.7256370964673238,
-0.9080177898160631, -0.7739945904033205, -0.29157882294526083,
-2.0599825024556027, -0.6156371996742152, -0.006086514109432934,
-0.8296782839032163, 0.0008092653341775424, 0.04161306816722683,
0.0034264692325982156, -0.9578001030483714, -0.006281927374160914,
-0.28835874344086, -4.245775886469167, -4.317410633818672,
-8.02385959091948, -3.240512651984099, -27.289853042932705,
-26.81776520493048, -82.45104200076496, -80.68164887277251
])
ref_energy_maxresid = 54.21913802238612
ref_force_maxresid = 791.6734866205463
# Testing pure-python and fortran versions of Gaussian-neural on different
# number of processes
for fortran in [False, True]:
for cores in range(1, 6):
label = 'train-nonperiodic/%s-%i' % (fortran, cores)
print(label)
calc = Amp(descriptor=Gaussian(
cutoff=6.5,
Gs=Gs,
fortran=fortran,
),
model=NeuralNetwork(
hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='sigmoid',
regressor=regressor,
fortran=fortran,
),
label=label,
dblabel=label,
cores=cores)
lossfunction = LossFunction(convergence=convergence)
calc.model.lossfunction = lossfunction
calc.train(images=images, )
diff = abs(calc.model.lossfunction.loss - ref_loss)
print("diff at 204 =", diff)
assert (diff < 10.**(-10.)), \
'Calculated value of loss function is wrong!'
diff = abs(calc.model.lossfunction.energy_loss - ref_energyloss)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom RMSE is wrong!'
diff = abs(calc.model.lossfunction.force_loss - ref_forceloss)
assert (diff < 10 ** (-10.)), \
'Calculated value of force RMSE is wrong!'
diff = abs(calc.model.lossfunction.energy_maxresid -
ref_energy_maxresid)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom max residual is wrong!'
diff = abs(calc.model.lossfunction.force_maxresid -
ref_force_maxresid)
assert (diff < 10 ** (-10.)), \
'Calculated value of force max residual is wrong!'
for _ in range(len(ref_dloss_dparameters)):
diff = abs(calc.model.lossfunction.dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-12.)), \
"Calculated value of loss function derivative is wrong!"
dblabel = label
secondlabel = '_' + label
calc = Amp(descriptor=Gaussian(
cutoff=6.5,
Gs=Gs,
fortran=fortran,
),
model=NeuralNetwork(
hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='sigmoid',
regressor=regressor,
fortran=fortran,
),
label=secondlabel,
dblabel=dblabel,
cores=cores)
lossfunction = LossFunction(convergence=convergence)
calc.model.lossfunction = lossfunction
calc.train(images=images, )
diff = abs(calc.model.lossfunction.loss - ref_loss)
assert (diff < 10.**(-10.)), \
'Calculated value of loss function is wrong!'
diff = abs(calc.model.lossfunction.energy_loss - ref_energyloss)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom RMSE is wrong!'
diff = abs(calc.model.lossfunction.force_loss - ref_forceloss)
assert (diff < 10 ** (-10.)), \
'Calculated value of force RMSE is wrong!'
diff = abs(calc.model.lossfunction.energy_maxresid -
ref_energy_maxresid)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom max residual is wrong!'
diff = abs(calc.model.lossfunction.force_maxresid -
ref_force_maxresid)
assert (diff < 10 ** (-10.)), \
'Calculated value of force max residual is wrong!'
for _ in range(len(ref_dloss_dparameters)):
diff = abs(calc.model.lossfunction.dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-12.)), \
'Calculated value of loss function derivative is wrong!'
def test():
pwd = os.getcwd()
os.mkdir(os.path.join(pwd, 'CuOPdnone'))
os.mkdir(os.path.join(pwd, '_CuOPdnone'))
###########################################################################
# Parameters
weights = OrderedDict([(1,
np.array([[1., 2.5], [0., 1.5], [0.,
-1.5], [3., 9.],
[1., -2.5], [2., 3.], [2.,
2.5], [3., 0.],
[-3.5, 1.], [5., 3.], [-2., 2.5],
[-4., 4.], [0., 0.]])),
(2, np.array([[1.], [2.], [0.]])),
(3, np.array([[3.5], [0.]]))])
scalings = OrderedDict([('intercept', 3.), ('slope', 2.)])
images = generate_images()
###########################################################################
# Testing pure-python and fortran versions of Gaussian-neural on different
# number of processes
for global_search in [None, 'SA']:
for fortran in [False, True]:
for extend_variables in [False, True]:
for data_format in ['db', 'json']:
for save_memory in [False]:
for cores in range(1, 7):
string = 'CuOPdnone/%s-%s-%s-%s-%s-%i'
label = string % (global_search, fortran,
extend_variables, data_format,
save_memory, cores)
if global_search is 'SA':
gs = \
SimulatedAnnealing(temperature=10, steps=5)
elif global_search is None:
gs = None
print label
calc = Amp(
descriptor=None,
regression=NeuralNetwork(
hiddenlayers=(2, 1),
activation='tanh',
weights=weights,
scalings=scalings,
),
fortran=fortran,
label=label,
)
calc.train(images=images,
energy_goal=10.**10.,
force_goal=10.**10.,
force_coefficient=0.04,
cores=cores,
data_format=data_format,
save_memory=save_memory,
global_search=gs,
extend_variables=extend_variables)
# Check for consistency between the two models
assert (abs(calc.cost_function - cost_function) <
10.**(-5.)), \
'The calculated value of cost function is \
wrong!'
assert (abs(calc.energy_per_atom_rmse -
energy_rmse) <
10.**(-5.)), \
'The calculated value of energy per atom RMSE \
is wrong!'
assert (abs(calc.force_rmse - force_rmse) <
10 ** (-5)), \
'The calculated value of force RMSE is wrong!'
dblabel = label
secondlabel = '_' + label
calc = Amp(descriptor=None,
regression=NeuralNetwork(
hiddenlayers=(2, 1),
activation='tanh',
weights=weights,
scalings=scalings,
),
fortran=fortran,
label=secondlabel,
dblabel=dblabel)
calc.train(images=images,
energy_goal=10.**10.,
force_goal=10.**10.,
force_coefficient=0.04,
cores=cores,
data_format=data_format,
save_memory=save_memory,
global_search=gs,
extend_variables=extend_variables)
# Check for consistency between the two models
assert (abs(calc.cost_function - cost_function) <
10.**(-5.)), \
'The calculated value of cost function is \
wrong!'
assert (abs(calc.energy_per_atom_rmse -
energy_rmse) <
10.**(-5.)), \
'The calculated value of energy per atom RMSE \
is wrong!'
assert (abs(calc.force_rmse - force_rmse) <
10 ** (-5)), \
'The calculated value of force RMSE is wrong!'
#!/usr/bin/env python
from amp import Amp
from amp.descriptor import *
from amp.regression import *
calc = Amp(label="./",
descriptor=Behler(cutoff=6.0),
regression=NeuralNetwork(hiddenlayers=(2, 3)))
calc.train("../train.db", # The training data
cores=1,
global_search=None, # not found the simulated annealing feature useful
extend_variables=False) # feature does not work properly and will crash
def train_rnn(baseframe=100,
tframe=8,
total_step=10,
traj='ethane.traj',
convergence={
'energy_rmse': 0.25,
'force_rmse': 0.5
},
elements=['C', 'H', 'O'],
hiddenlayers=(64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64),
optim='ADAM',
cores=4):
"""Gaussian/tflow train test."""
p = ple()
label = 'amp'
all_images = Trajectory(traj)
nimg, mean_e = get_mean_energy(all_images)
G = make_symmetry_functions(elements=elements,
type='G2',
etas=np.logspace(np.log10(0.05),
np.log10(5.),
num=4))
G += make_symmetry_functions(elements=elements,
type='G5',
etas=[0.005],
zetas=[1., 4.],
gammas=[+1., -1.])
G = {element: G for element in elements} # Gs=G
if not isfile('amp.amp'):
print('\nset up calculator ...\n')
calc = Amp(descriptor=Gaussian(mode='atom-centered', Gs=G),
model=NeuralNetwork(hiddenlayers=hiddenlayers,
convergenceCriteria=convergence,
activation='tanh',
energy_coefficient=1.0,
force_coefficient=None,
optimizationMethod=optim,
parameters={'energyMeanScale': mean_e},
maxTrainingEpochs=100000),
label=label,
cores=cores) # 'l-BFGS-b' or 'ADAM'
trained_images = [all_images[j] for j in range(0, baseframe)]
calc.train(overwrite=True, images=trained_images)
del calc
else:
calc = Amp.load('amp.amp')
calc.model.parameters['convergence'] = convergence
calc.model.lossfunction = LossFunction(convergence=convergence)
trained_images = [all_images[j] for j in range(0, baseframe)]
calc.train(overwrite=True, images=trained_images)
del calc
tstep = int((nimg - baseframe) / tframe)
if total_step > tstep:
total_step = tstep
print('Max train cycle of %d is allowed.' % total_step)
edfts, eamps, eamps_ = [], [], []
dolabel = True
basestep = int(baseframe / tframe)
for step in range(basestep, total_step + basestep):
new_images = [
all_images[j]
for j in range(0 + step * tframe, tframe + step * tframe)
]
trained_images.extend(new_images)
x, edft, eamp, eamp_ = [], [], [], []
ii = step * tframe
# ----- test -----
calc1 = Amp.load('amp.amp')
for i, image in enumerate(new_images):
x.append(ii)
eamp_.append(calc1.get_potential_energy(image))
eamps_.append(eamp_[-1])
edft.append(image.get_potential_energy())
edfts.append(edft[-1])
ii += 1
del calc1
# ----- train -----
calc = Amp.load('amp.amp')
calc.model.lossfunction = LossFunction(convergence=convergence)
# calc.model.convergenceCriteria=convergence
calc.train(overwrite=True, images=trained_images)
del calc
# ----- test -----
calc2 = Amp.load('amp.amp')
print('\n---- current training result ---- \n')
for i, image in enumerate(new_images):
eamp.append(calc2.get_potential_energy(image))
eamps.append(eamp[-1])
print("energy(AMP) = %f energy(DFT) = %f" % (eamp[-1], edft[i]))
# print("forces = %s" % str(calc2.get_forces(image)))
del calc2
plot_energies(edfts, eamps, eamp_=None)
system('epstopdf energies.eps')
p.scatter(x, edft, eamp, eamp_, dolabel=dolabel)
if dolabel:
dolabel = False
p.plot()
system('epstopdf energies_scatter.eps')
def train_images(images, HL, E_conv):
Hidden_Layer=tuple(HL)
calc = Amp(descriptor=Gaussian(), model=NeuralNetwork(hiddenlayers=Hidden_Layer))
calc.model.lossfunction = LossFunction(convergence={'energy_rmse': E_conv})
#calc.model.lossfunction = LossFunction(force_coefficient=-0.1)
calc.train(images=images, overwrite=True)
def test():
images = generate_data(2)
regressor = Regressor(optimizer='BFGS')
calc = Amp(descriptor=Gaussian(),
model=NeuralNetwork(hiddenlayers=(3, 3),
regressor=regressor,),
cores=1)
step = 0
for d in [None, 0.00001]:
for fortran in [True, False]:
for cores in [1, 2]:
step += 1
label = \
'numeric_analytic_test/analytic-%s-%i' % (fortran, cores) \
if d is None \
else 'numeric_analytic_test/numeric-%s-%i' \
% (fortran, cores)
print(label)
loss = LossFunction(convergence={'energy_rmse': 10 ** 10,
'force_rmse': 10 ** 10},
d=d)
calc.set_label(label)
calc.dblabel = 'numeric_analytic_test/analytic-True-1'
calc.model.lossfunction = loss
calc.descriptor.fortran = fortran
calc.model.fortran = fortran
calc.cores = cores
calc.train(images=images,)
if step == 1:
ref_energies = []
ref_forces = []
for image in images:
ref_energies += [calc.get_potential_energy(image)]
ref_forces += [calc.get_forces(image)]
ref_dloss_dparameters = \
calc.model.lossfunction.dloss_dparameters
else:
energies = []
forces = []
for image in images:
energies += [calc.get_potential_energy(image)]
forces += [calc.get_forces(image)]
dloss_dparameters = \
calc.model.lossfunction.dloss_dparameters
for image_no in range(2):
diff = abs(energies[image_no] - ref_energies[image_no])
assert (diff < 10.**(-13.)), \
'The calculated value of energy of image %i is ' \
'wrong!' % (image_no + 1)
for atom_no in range(6):
for i in range(3):
diff = abs(forces[image_no][atom_no][i] -
ref_forces[image_no][atom_no][i])
assert (diff < 10.**(-10.)), \
'The calculated %i force of atom %i of ' \
'image %i is wrong!' \
% (i, atom_no, image_no + 1)
# Checks analytical and numerical dloss_dparameters
for _ in range(len(ref_dloss_dparameters)):
diff = abs(dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-10.)), \
'The calculated value of loss function ' \
'derivative is wrong!'
# Checks analytical and numerical forces
forces = []
for image in images:
image.set_calculator(calc)
forces += [calc.calculate_numerical_forces(image, d=d)]
for atom_no in range(6):
for i in range(3):
diff = abs(forces[image_no][atom_no][i] -
ref_forces[image_no][atom_no][i])
assert (diff < 10.**(-9.)), \
'The calculated %i force of atom %i of ' \
'image %i is wrong!' % (i, atom_no, image_no + 1)
#!/usr/bin/env python
from amp import Amp
from amp.descriptor import *
from amp.regression import *
calc = Amp(label="./",
descriptor=Behler(cutoff=6.0),
regression=NeuralNetwork(hiddenlayers=(2, 3)))
calc.train(
"../train.db", # The training data
cores=1,
global_search=None, # not found the simulated annealing feature useful
extend_variables=False) # feature does not work properly and will crash
def periodic_0th_bfgs_step_test():
"""Gaussian/Neural training periodic standard test.
Compares results to that expected from separate mathematica
calculations.
"""
# Making the list of images
images = [
Atoms(symbols='PdOPd',
pbc=np.array([True, False, False], dtype=bool),
cell=np.array([[2., 0., 0.], [0., 2., 0.], [0., 0., 2.]]),
positions=np.array([[0.5, 1., 0.5], [1., 0.5, 1.],
[1.5, 1.5, 1.5]])),
Atoms(symbols='PdO',
pbc=np.array([True, True, False], dtype=bool),
cell=np.array([[2., 0., 0.], [0., 2., 0.], [0., 0., 2.]]),
positions=np.array([[0.5, 1., 0.5], [1., 0.5, 1.]])),
Atoms(symbols='Cu',
pbc=np.array([True, True, False], dtype=bool),
cell=np.array([[1.8, 0., 0.], [0., 1.8, 0.], [0., 0., 1.8]]),
positions=np.array([[0., 0., 0.]]))
]
for image in images:
image.set_calculator(EMT())
image.get_potential_energy(apply_constraint=False)
image.get_forces(apply_constraint=False)
# Parameters
Gs = {
'O': [{
'type': 'G2',
'element': 'Pd',
'eta': 0.8
}, {
'type': 'G4',
'elements': ['O', 'Pd'],
'eta': 0.3,
'gamma': 0.6,
'zeta': 0.5
}],
'Pd': [{
'type': 'G2',
'element': 'Pd',
'eta': 0.2
}, {
'type': 'G4',
'elements': ['Pd', 'Pd'],
'eta': 0.9,
'gamma': 0.75,
'zeta': 1.5
}],
'Cu': [{
'type': 'G2',
'element': 'Cu',
'eta': 0.8
}, {
'type': 'G4',
'elements': ['Cu', 'Cu'],
'eta': 0.3,
'gamma': 0.6,
'zeta': 0.5
}]
}
hiddenlayers = {'O': (2, ), 'Pd': (2, ), 'Cu': (2, )}
weights = OrderedDict([
('O',
OrderedDict([(1, np.matrix([[-2.0, 6.0], [3.0, -3.0], [1.5, -0.9]])),
(2, np.matrix([[5.5], [3.6], [1.4]]))])),
('Pd',
OrderedDict([(1, np.matrix([[-1.0, 3.0], [2.0, 4.2], [1.0, -0.7]])),
(2, np.matrix([[4.0], [0.5], [3.0]]))])),
('Cu',
OrderedDict([(1, np.matrix([[0.0, 1.0], [-1.0, -2.0], [2.5, -1.9]])),
(2, np.matrix([[0.5], [1.6], [-1.4]]))]))
])
scalings = OrderedDict([
('O', OrderedDict([('intercept', -2.3), ('slope', 4.5)])),
('Pd', OrderedDict([('intercept', 1.6), ('slope', 2.5)])),
('Cu', OrderedDict([('intercept', -0.3), ('slope', -0.5)]))
])
# Correct values
if aseversion < 12: # EMT values have changed from 3.12.0 version
ref_loss = 8004.292841411172
ref_energyloss = (43.7360019403031**2.) * 3
ref_forceloss = (137.40994760947325**2.) * 3
ref_dloss_dparameters = np.array([
0.08141668748130322, 0.03231235582925534, 0.04388650395738586,
0.017417514465922313, 0.028431276597563077, 0.011283700608814465,
0.0941695726576061, -0.12322258890990219, 0.12679918754154568,
63.53960075374332, 0.01624770019548904, -86.6263955859162,
-0.01777752828707744, 86.22415217526024, 0.017745913074496918,
104.58358033298292, -96.73280209888215, -99.09843648905876,
-8.302880631972338, -1.2590007162074357, 8.302877346883133,
1.25875988418134, -8.302866610678247, -1.2563833805675353,
28.324298392680998, 28.093155094726413, -29.37874455931869,
-11.247473567044866, 11.119951466664787, -87.08582317481387,
-20.939485239182346, -125.73267675705365, -35.138524407482116
])
else:
ref_loss = 8004.287750978173
ref_energyloss = (43.73598563177581**2.) * 3
ref_forceloss = (137.409923023214**2.) * 3
ref_dloss_dparameters = np.array([
0.08141663280688925, 0.03231233413027478, 0.043886474485922956,
0.01741750276939638, 0.02843125750487539, 0.011283693031378718,
0.09416950941914284, -0.12322250616122936, 0.1267991023910503,
63.53958764057119, 0.016247696749304368, -86.62637753054923,
-0.01777752451341436, 86.22413420485914, 0.01774590930723711,
104.58353326982777, -96.73275667196937, -99.09839026204304,
-8.302877823431269, -1.2590002903842232, 8.302874538343092,
1.2587594584335775, -8.302863802141216, -1.2563829555383859,
28.32428881173613, 28.093145591893936, -29.37873462156934,
-11.24746601393696, 11.11994399919284, -87.08579155328007,
-20.93947792122797, -125.73262989900473, -35.13850819392253
])
# Testing pure-python and fortran versions of Gaussian-neural on different
# number of processes
for fortran in [False, True]:
for cores in range(1, 4):
label = 'train-periodic/%s-%i' % (fortran, cores)
print(label)
calc = Amp(descriptor=Gaussian(
cutoff=4.,
Gs=Gs,
fortran=fortran,
),
model=NeuralNetwork(
hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='tanh',
regressor=regressor,
fortran=fortran,
),
label=label,
dblabel=label,
cores=cores)
lossfunction = LossFunction(convergence=convergence)
calc.model.lossfunction = lossfunction
calc.train(images=images, )
diff = abs(calc.model.lossfunction.loss - ref_loss)
print("diff at 414 =", diff)
assert (diff < 10.**(-10.)), \
'Calculated value of loss function is wrong!'
diff = abs(calc.model.lossfunction.energy_loss - ref_energyloss)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom RMSE is wrong!'
diff = abs(calc.model.lossfunction.force_loss - ref_forceloss)
assert (diff < 10 ** (-9.)), \
'Calculated value of force RMSE is wrong!'
for _ in range(len(ref_dloss_dparameters)):
diff = abs(calc.model.lossfunction.dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-10.)), \
'Calculated value of loss function derivative is wrong!'
dblabel = label
secondlabel = '_' + label
calc = Amp(descriptor=Gaussian(cutoff=4., Gs=Gs, fortran=fortran),
model=NeuralNetwork(
hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='tanh',
regressor=regressor,
fortran=fortran,
),
label=secondlabel,
dblabel=dblabel,
cores=cores)
lossfunction = LossFunction(convergence=convergence)
calc.model.lossfunction = lossfunction
calc.train(images=images, )
diff = abs(calc.model.lossfunction.loss - ref_loss)
assert (diff < 10.**(-10.)), \
'Calculated value of loss function is wrong!'
diff = abs(calc.model.lossfunction.energy_loss - ref_energyloss)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom RMSE is wrong!'
diff = abs(calc.model.lossfunction.force_loss - ref_forceloss)
assert (diff < 10 ** (-9.)), \
'Calculated value of force RMSE is wrong!'
for _ in range(len(ref_dloss_dparameters)):
diff = abs(calc.model.lossfunction.dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-10.)), \
'Calculated value of loss function derivative is wrong!'
def test():
images = make_training_images()
for descriptor in [None, Gaussian()]:
for global_search in [
None, SimulatedAnnealing(temperature=10, steps=5)
]:
for data_format in ['json', 'db']:
for save_memory in [
False,
]:
for fortran in [False, True]:
for cores in range(1, 4):
print(descriptor, global_search, data_format,
save_memory, fortran, cores)
pwd = os.getcwd()
testdir = 'read_write_test'
os.mkdir(testdir)
os.chdir(testdir)
regression = NeuralNetwork(hiddenlayers=(
5,
5,
))
calc = Amp(
label='calc',
descriptor=descriptor,
regression=regression,
fortran=fortran,
)
# Should start with new variables
calc.train(
images,
energy_goal=0.01,
force_goal=10.,
global_search=global_search,
extend_variables=True,
data_format=data_format,
save_memory=save_memory,
cores=cores,
)
# Test that we cannot overwrite. (Strange code
# here because we *want* it to raise an
# exception...)
try:
calc.train(
images,
energy_goal=0.01,
force_goal=10.,
global_search=global_search,
extend_variables=True,
data_format=data_format,
save_memory=save_memory,
cores=cores,
)
except IOError:
pass
else:
raise RuntimeError(
'Code allowed to overwrite!')
# Test that we can manually overwrite.
# Should start with existing variables
calc.train(
images,
energy_goal=0.01,
force_goal=10.,
global_search=global_search,
extend_variables=True,
data_format=data_format,
save_memory=save_memory,
overwrite=True,
cores=cores,
)
label = 'testdir'
if not os.path.exists(label):
os.mkdir(label)
# New directory calculator.
calc = Amp(
label='testdir/calc',
descriptor=descriptor,
regression=regression,
fortran=fortran,
)
# Should start with new variables
calc.train(
images,
energy_goal=0.01,
force_goal=10.,
global_search=global_search,
extend_variables=True,
data_format=data_format,
save_memory=save_memory,
cores=cores,
)
# Open existing, save under new name.
calc = Amp(
load='calc',
label='calc2',
descriptor=descriptor,
regression=regression,
fortran=fortran,
)
# Should start with existing variables
calc.train(
images,
energy_goal=0.01,
force_goal=10.,
global_search=global_search,
extend_variables=True,
data_format=data_format,
save_memory=save_memory,
cores=cores,
)
label = 'calc_new'
if not os.path.exists(label):
os.mkdir(label)
# Change label and re-train
calc.set_label('calc_new/calc')
# Should start with existing variables
calc.train(
images,
energy_goal=0.01,
force_goal=10.,
global_search=global_search,
extend_variables=True,
data_format=data_format,
save_memory=save_memory,
cores=cores,
)
# Open existing without specifying new name.
calc = Amp(
load='calc',
descriptor=descriptor,
regression=regression,
fortran=fortran,
)
# Should start with existing variables
calc.train(
images,
energy_goal=0.01,
force_goal=10.,
global_search=global_search,
extend_variables=True,
data_format=data_format,
save_memory=save_memory,
cores=cores,
)
os.chdir(pwd)
shutil.rmtree(testdir, ignore_errors=True)
del calc, regression
def periodic_0th_bfgs_step_test():
###########################################################################
# Making the list of images
images = [
Atoms(symbols='PdOPd',
pbc=np.array([True, False, False], dtype=bool),
cell=np.array([[2., 0., 0.], [0., 2., 0.], [0., 0., 2.]]),
positions=np.array([[0.5, 1., 0.5], [1., 0.5, 1.],
[1.5, 1.5, 1.5]])),
Atoms(symbols='PdO',
pbc=np.array([True, True, False], dtype=bool),
cell=np.array([[2., 0., 0.], [0., 2., 0.], [0., 0., 2.]]),
positions=np.array([[0.5, 1., 0.5], [1., 0.5, 1.]])),
Atoms(symbols='Cu',
pbc=np.array([True, True, False], dtype=bool),
cell=np.array([[1.8, 0., 0.], [0., 1.8, 0.], [0., 0., 1.8]]),
positions=np.array([[0., 0., 0.]]))
]
for image in images:
image.set_calculator(EMT())
image.get_potential_energy(apply_constraint=False)
image.get_forces(apply_constraint=False)
###########################################################################
# Parameters
Gs = {
'O': [{
'type': 'G2',
'element': 'Pd',
'eta': 0.8
}, {
'type': 'G4',
'elements': ['O', 'Pd'],
'eta': 0.3,
'gamma': 0.6,
'zeta': 0.5
}],
'Pd': [{
'type': 'G2',
'element': 'Pd',
'eta': 0.2
}, {
'type': 'G4',
'elements': ['Pd', 'Pd'],
'eta': 0.9,
'gamma': 0.75,
'zeta': 1.5
}],
'Cu': [{
'type': 'G2',
'element': 'Cu',
'eta': 0.8
}, {
'type': 'G4',
'elements': ['Cu', 'Cu'],
'eta': 0.3,
'gamma': 0.6,
'zeta': 0.5
}]
}
hiddenlayers = {'O': (2), 'Pd': (2), 'Cu': (2)}
weights = OrderedDict([
('O',
OrderedDict([(1, np.matrix([[-2.0, 6.0], [3.0, -3.0], [1.5, -0.9]])),
(2, np.matrix([[5.5], [3.6], [1.4]]))])),
('Pd',
OrderedDict([(1, np.matrix([[-1.0, 3.0], [2.0, 4.2], [1.0, -0.7]])),
(2, np.matrix([[4.0], [0.5], [3.0]]))])),
('Cu',
OrderedDict([(1, np.matrix([[0.0, 1.0], [-1.0, -2.0], [2.5, -1.9]])),
(2, np.matrix([[0.5], [1.6], [-1.4]]))]))
])
scalings = OrderedDict([
('O', OrderedDict([('intercept', -2.3), ('slope', 4.5)])),
('Pd', OrderedDict([('intercept', 1.6), ('slope', 2.5)])),
('Cu', OrderedDict([('intercept', -0.3), ('slope', -0.5)]))
])
###########################################################################
# Correct values
correct_cost = 8004.292841472513
correct_energy_rmse = 43.736001940333836
correct_force_rmse = 137.4099476110887
correct_der_cost_fxn = [
0.0814166874813534, 0.03231235582927526, 0.04388650395741291,
0.017417514465933048, 0.0284312765975806, 0.011283700608821421,
0.09416957265766414, -0.12322258890997816, 0.12679918754162384,
63.5396007548815, 0.016247700195771732, -86.62639558745185,
-0.017777528287386473, 86.22415217678898, 0.017745913074805372,
104.58358033260711, -96.7328020983672, -99.09843648854351,
-8.302880631971407, -1.2590007162073242, 8.3028773468822,
1.258759884181224, -8.302866610677315, -1.2563833805673688,
28.324298392677846, 28.09315509472324, -29.378744559315365,
-11.247473567051799, 11.119951466671642, -87.08582317485761,
-20.93948523898559, -125.73267675714658, -35.13852440758523
]
###########################################################################
# Testing pure-python and fortran versions of Gaussian-neural on different
# number of processes
for global_search in [None, 'SA']:
for fortran in [False, True]:
for extend_variables in [False, True]:
for data_format in ['db', 'json']:
for save_memory in [False]:
for cores in range(1, 5):
string = 'CuOPdbp/2/%s-%s-%s-%s-%s-%i'
label = string % (global_search, fortran,
extend_variables, data_format,
save_memory, cores)
if global_search is 'SA':
gs = \
SimulatedAnnealing(temperature=10, steps=5)
elif global_search is None:
gs = None
print label
calc = Amp(descriptor=Gaussian(cutoff=4., Gs=Gs),
regression=NeuralNetwork(
hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='tanh',
),
fortran=fortran,
label=label)
calc.train(images=images,
energy_goal=10.**10.,
force_goal=10.**10,
force_coefficient=0.04,
cores=cores,
data_format=data_format,
save_memory=save_memory,
global_search=gs,
extend_variables=extend_variables)
assert (abs(calc.cost_function - correct_cost) <
10.**(-7.)), \
'The calculated value of cost function is \
wrong!'
assert (abs(calc.energy_per_atom_rmse -
correct_energy_rmse) < 10.**(-10.)), \
'The calculated value of energy per atom RMSE \
is wrong!'
assert (abs(calc.force_rmse - correct_force_rmse) <
10 ** (-8.)), \
'The calculated value of force RMSE is wrong!'
for _ in range(len(correct_der_cost_fxn)):
assert(abs(calc.der_variables_cost_function[
_] -
correct_der_cost_fxn[_]) < 10 ** (-8)), \
'The calculated value of cost function \
derivative is wrong!'
dblabel = label
secondlabel = '_' + label
calc = Amp(descriptor=Gaussian(cutoff=4., Gs=Gs),
regression=NeuralNetwork(
hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='tanh',
),
fortran=fortran,
label=secondlabel,
dblabel=dblabel)
calc.train(images=images,
energy_goal=10.**10.,
force_goal=10.**10,
force_coefficient=0.04,
cores=cores,
data_format=data_format,
save_memory=save_memory,
global_search=gs,
extend_variables=extend_variables)
assert (abs(calc.cost_function - correct_cost) <
10.**(-7.)), \
'The calculated value of cost function is \
wrong!'
assert (abs(calc.energy_per_atom_rmse -
correct_energy_rmse) < 10.**(-10.)), \
'The calculated value of energy per atom RMSE \
is wrong!'
assert (abs(calc.force_rmse - correct_force_rmse) <
10 ** (-8.)), \
'The calculated value of force RMSE is wrong!'
for _ in range(len(correct_der_cost_fxn)):
assert(abs(calc.der_variables_cost_function[
_] -
correct_der_cost_fxn[_] < 10 ** (-8))), \
'The calculated value of cost function \
def periodic_0th_bfgs_step_test():
# Making the list of images
images = [Atoms(symbols='PdOPd',
pbc=np.array([True, False, False], dtype=bool),
cell=np.array(
[[2., 0., 0.],
[0., 2., 0.],
[0., 0., 2.]]),
positions=np.array(
[[0.5, 1., 0.5],
[1., 0.5, 1.],
[1.5, 1.5, 1.5]])),
Atoms(symbols='PdO',
pbc=np.array([True, True, False], dtype=bool),
cell=np.array(
[[2., 0., 0.],
[0., 2., 0.],
[0., 0., 2.]]),
positions=np.array(
[[0.5, 1., 0.5],
[1., 0.5, 1.]])),
Atoms(symbols='Cu',
pbc=np.array([True, True, False], dtype=bool),
cell=np.array(
[[1.8, 0., 0.],
[0., 1.8, 0.],
[0., 0., 1.8]]),
positions=np.array(
[[0., 0., 0.]]))]
for image in images:
image.set_calculator(EMT())
image.get_potential_energy(apply_constraint=False)
image.get_forces(apply_constraint=False)
# Parameters
Gs = {'O': [{'type': 'G2', 'element': 'Pd', 'eta': 0.8},
{'type': 'G4', 'elements': ['O', 'Pd'], 'eta':0.3, 'gamma':0.6,
'zeta':0.5}],
'Pd': [{'type': 'G2', 'element': 'Pd', 'eta': 0.2},
{'type': 'G4', 'elements': ['Pd', 'Pd'],
'eta':0.9, 'gamma':0.75, 'zeta':1.5}],
'Cu': [{'type': 'G2', 'element': 'Cu', 'eta': 0.8},
{'type': 'G4', 'elements': ['Cu', 'Cu'], 'eta':0.3,
'gamma':0.6, 'zeta':0.5}]}
hiddenlayers = {'O': (2,), 'Pd': (2,), 'Cu': (2,)}
weights = OrderedDict([('O', OrderedDict([(1, np.matrix([[-2.0, 6.0],
[3.0, -3.0],
[1.5, -0.9]])),
(2, np.matrix([[5.5],
[3.6],
[1.4]]))])),
('Pd', OrderedDict([(1, np.matrix([[-1.0, 3.0],
[2.0, 4.2],
[1.0, -0.7]])),
(2, np.matrix([[4.0],
[0.5],
[3.0]]))])),
('Cu', OrderedDict([(1, np.matrix([[0.0, 1.0],
[-1.0, -2.0],
[2.5, -1.9]])),
(2, np.matrix([[0.5],
[1.6],
[-1.4]]))]))])
scalings = OrderedDict([('O', OrderedDict([('intercept', -2.3),
('slope', 4.5)])),
('Pd', OrderedDict([('intercept', 1.6),
('slope', 2.5)])),
('Cu', OrderedDict([('intercept', -0.3),
('slope', -0.5)]))])
# Correct values
ref_loss = 8004.292841411172
ref_energyloss = (43.7360019403031 ** 2.) * 3
ref_forceloss = (137.40994760947325 ** 2.) * 3
ref_dloss_dparameters = np.array([0.08141668748130322,
0.03231235582925534,
0.04388650395738586,
0.017417514465922313,
0.028431276597563077,
0.011283700608814465,
0.0941695726576061,
-0.12322258890990219,
0.12679918754154568,
63.53960075374332,
0.01624770019548904,
-86.6263955859162,
-0.01777752828707744,
86.22415217526024,
0.017745913074496918,
104.58358033298292,
-96.73280209888215,
-99.09843648905876,
-8.302880631972338,
-1.2590007162074357,
8.302877346883133,
1.25875988418134,
-8.302866610678247,
-1.2563833805675353,
28.324298392680998,
28.093155094726413,
-29.37874455931869,
-11.247473567044866,
11.119951466664787,
-87.08582317481387,
-20.939485239182346,
-125.73267675705365,
-35.138524407482116])
# Testing pure-python and fortran versions of Gaussian-neural on different
# number of processes
for fortran in [False, True]:
for cores in range(1, 4):
label = 'train-periodic/%s-%i' % (fortran, cores)
print label
calc = Amp(descriptor=Gaussian(cutoff=4.,
Gs=Gs,
fortran=fortran,),
model=NeuralNetwork(hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='tanh',
regressor=regressor,
fortran=fortran,),
label=label,
dblabel=label,
cores=cores)
lossfunction = LossFunction(convergence=convergence)
calc.model.lossfunction = lossfunction
calc.train(images=images,)
diff = abs(calc.model.lossfunction.loss - ref_loss)
assert (diff < 10.**(-10.)), \
'Calculated value of loss function is wrong!'
diff = abs(calc.model.lossfunction.energy_loss - ref_energyloss)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom RMSE is wrong!'
diff = abs(calc.model.lossfunction.force_loss - ref_forceloss)
assert (diff < 10 ** (-9.)), \
'Calculated value of force RMSE is wrong!'
for _ in range(len(ref_dloss_dparameters)):
diff = abs(calc.model.lossfunction.dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-10.)), \
'Calculated value of loss function derivative is wrong!'
dblabel = label
secondlabel = '_' + label
calc = Amp(descriptor=Gaussian(cutoff=4.,
Gs=Gs,
fortran=fortran),
model=NeuralNetwork(hiddenlayers=hiddenlayers,
weights=weights,
scalings=scalings,
activation='tanh',
regressor=regressor,
fortran=fortran,),
label=secondlabel,
dblabel=dblabel,
cores=cores)
lossfunction = LossFunction(convergence=convergence)
calc.model.lossfunction = lossfunction
calc.train(images=images,)
diff = abs(calc.model.lossfunction.loss - ref_loss)
assert (diff < 10.**(-10.)), \
'Calculated value of loss function is wrong!'
diff = abs(calc.model.lossfunction.energy_loss - ref_energyloss)
assert (diff < 10.**(-10.)), \
'Calculated value of energy per atom RMSE is wrong!'
diff = abs(calc.model.lossfunction.force_loss - ref_forceloss)
assert (diff < 10 ** (-9.)), \
'Calculated value of force RMSE is wrong!'
for _ in range(len(ref_dloss_dparameters)):
diff = abs(calc.model.lossfunction.dloss_dparameters[_] -
ref_dloss_dparameters[_])
assert(diff < 10 ** (-10.)), \
'Calculated value of loss function derivative is wrong!'
