def get_energy_metrics(config):
trainer = AtomsTrainer(config)
trainer.train()
predictions = trainer.predict(images)
pred_energies = np.array(predictions["energy"])
mae = np.mean(np.abs(true_energies - pred_energies))
assert mae < 0.02
def train_model(train_list, test_list, descriptor_set, trial_num, log_filename):
Gs = construct_parameter_set(descriptor_set, log_filename = log_filename)
# elements = ["Cu", "C", "O"]
elements = ["H","O","C"]
config = {
"model": {"name":"bpnn",
"get_forces": False,
"num_layers": 3,
"num_nodes": 50,
#"elementwise":False,
"batchnorm": True},
"optim": {
"gpus":0,
#"force_coefficient": 0.04,
"force_coefficient": 0.0,
"lr": 1e-3,
"batch_size": 256,
"epochs": 5000,
"loss": "mae",
#"scheduler": {"policy": "StepLR", "params": {"step_size": 1000, "gamma": 0.5}},
},
"dataset": {
"raw_data": train_list,
"val_split": 0.2,
"elements": elements,
"fp_scheme": "gaussian",
"fp_params": Gs,
"save_fps": False,
"scaling": {"type": "normalize", "range": (0, 1)}
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": trial_num,
"identifier": "test",
"verbose": True,
"logger": False,
},
}
trainer = AtomsTrainer(config)
trainer.train()
test_model(trainer, train_list, data_type = "train", log_filename = log_filename)
test_model(trainer, test_list, data_type = "test", log_filename = log_filename)
return
def get_force_metrics(config):
trainer = AtomsTrainer(config)
trainer.train()
predictions = trainer.predict(images)
pred_energies = np.array(predictions["energy"])
pred_forces = np.concatenate(np.array(predictions["forces"]))
e_mae = np.mean(np.abs(true_energies - pred_energies))
f_mae = np.mean(np.abs(pred_forces - true_forces))
assert e_mae < 0.06
assert f_mae < 0.06
def get_performance_metrics(config):
trainer = AtomsTrainer(config)
trainer.train()
predictions = trainer.predict(images)
pred_energies = np.array(predictions["energy"])
pred_forces = np.concatenate(np.array(predictions["forces"]))
e_mae = np.mean(np.abs(true_energies - pred_energies))
f_mae = np.mean(np.abs(pred_forces - true_forces))
assert e_mae < 0.01, "%f !< .01" % e_mae
assert f_mae < 0.03, "%f !< .03" % f_mae
return e_mae, f_mae
def train_and_combine(args_list):
"""
method for training trainer on ensemble sets, then create neural net calc,
returns trained calc
"""
training_dataset = args_list[0]
trainer = args_list[1]
trainer.train(raw_data=training_dataset)
check_path = trainer.cp_dir
trainer = AtomsTrainer()
trainer.load_pretrained(checkpoint_path=check_path)
trainer_calc = trainer.get_calc()
return trainer_calc
def train(self, parent_dataset, new_dataset=None):
"""
Uses Dask to parallelize, must have previously set up cluster,
image to use, and pool of workers
"""
ensemble_sets, parent_dataset = non_bootstrap_ensemble(
parent_dataset, n_ensembles=self.n_ensembles)
def train_and_combine(args_list):
"""
method for training trainer on ensemble sets, then create neural net calc,
returns trained calc
"""
training_dataset = args_list[0]
trainer = args_list[1]
trainer.train(raw_data=training_dataset)
check_path = trainer.cp_dir
trainer = AtomsTrainer()
trainer.load_pretrained(checkpoint_path=check_path)
trainer_calc = trainer.get_calc()
return trainer_calc
# split ensemble sets into separate args_lists, clone: trainer,
# base calc and add to args_lists, add: refs to args_lists
args_lists = []
random.seed(self.amptorch_trainer.config["cmd"]["seed"])
randomlist = [random.randint(0, 4294967295) for set in ensemble_sets]
for i in range(len(ensemble_sets)):
ensemble_set = ensemble_sets[i]
copy_config = copy.deepcopy(self.amptorch_trainer.config)
copy_config["cmd"]["seed"] = randomlist[i]
copy_config["cmd"]["identifier"] = copy_config["cmd"][
"identifier"] + str(uuid.uuid4())
trainer_copy = AtomsTrainer(copy_config)
args_lists.append((ensemble_set, trainer_copy))
# map training method, returns array of delta calcs
trained_calcs = []
if self.executor is not None:
futures = []
for args_list in args_lists:
big_future = self.executor.scatter(args_list)
futures.append(
self.executor.submit(train_and_combine, big_future))
trained_calcs = [future.result() for future in futures]
else:
for args_list in args_lists:
trained_calcs.append(train_and_combine(args_list))
# call init to construct ensemble calc from array of delta calcs
self.trained_calcs = trained_calcs
def run_offlineal(cluster, parent_calc, elements, al_learner_params, config, optimizer):
Gs = {
"default": {
"G2": {
"etas": np.logspace(np.log10(0.05), np.log10(5.0), num=4),
"rs_s": [0],
},
"G4": {"etas": [0.005], "zetas": [1.0, 4.0], "gammas": [1.0, -1.0]},
"cutoff": 6,
},
}
images = [cluster]
al_learner_params["atomistic_method"] = Relaxation(
cluster, optimizer, fmax=0.01, steps=100
)
config["dataset"] = {
"raw_data": images,
"val_split": 0,
"elements": elements,
"fp_params": Gs,
"save_fps": False,
"scaling": {"type": "normalize", "range": (-1, 1)},
}
config["cmd"] = {
"debug": False,
"run_dir": "./",
"seed": 2,
"identifier": "cluster",
"verbose": True,
# "logger": True,
"single-threaded": True,
}
trainer = AtomsTrainer(config)
# base_calc = MultiMorse(images, Gs["default"]["cutoff"], combo="mean")
base_calc = EMT()
offlinecalc = FmaxLearner(
al_learner_params, trainer, images, parent_calc, base_calc
)
if os.path.exists("queried_images.db"):
os.remove("queried_images.db")
offlinecalc.learn()
al_iterations = offlinecalc.iterations - 1
file_path = al_learner_params["file_dir"] + al_learner_params["filename"]
final_ml_traj = read("{}_iter_{}.traj".format(file_path, al_iterations), ":")
relaxed_clus = final_ml_traj[-1]
return relaxed_clus, offlinecalc.parent_calls
def test_pretrained():
torch.set_num_threads(1)
trainer = AtomsTrainer(config)
trainer.train()
trained_cpdir = trainer.cp_dir
e_mae_1, f_mae_1 = get_metrics(trainer)
config["optim"]["epochs"] = 100
pretrained_trainer = AtomsTrainer(config)
pretrained_trainer.load_pretrained(trained_cpdir)
e_mae_2, f_mae_2 = get_metrics(pretrained_trainer)
assert e_mae_1 == e_mae_2, "Pretrained energy metrics inconsistent!"
assert f_mae_1 == f_mae_2, "Pretrained force metrics inconsistent!"
pretrained_trainer.train()
e_mae_3, f_mae_3 = get_metrics(pretrained_trainer)
assert e_mae_3 < e_mae_2, "Retrained metrics are larger!"
assert f_mae_3 < f_mae_2, "Retrained metrics are larger!"
def test_pretrained_no_config():
config_1 = copy.deepcopy(config)
trainer = AtomsTrainer(config_1)
trainer.train()
trained_cpdir = trainer.cp_dir
e_mae_1, f_mae_1 = get_metrics(trainer)
trainer_2 = AtomsTrainer()
trainer_2.load_pretrained(trained_cpdir)
e_mae_2, f_mae_2 = get_metrics(trainer_2)
assert e_mae_1 == e_mae_2, "configless - pretrained energy metrics inconsistent!"
assert f_mae_1 == f_mae_2, "configless - pretrained force metrics inconsistent!"
def module_evaluate(learning_rate, num_nodes, num_layers):
learning_rate = float(learning_rate)
num_nodes = int(num_nodes)
num_layers = int(num_layers)
input_filename = "../data/water_dft.traj"
# split input if there's no split
if (os.path.exists("../data/train.traj") is
False) or (os.path.exists("../data/test.traj") is False):
print("Creating train_test split. ")
train_ratio = 0.9
training_list, test_list = split_train_test(
input_filename,
train_ratio,
save=True,
filenames=["../data/train.traj", "../data/test.traj"])
else:
print("Reading train_test split. ")
training_list, test_list = load_training_data("../data/train.traj",
"../data/test.traj")
sigmas = np.logspace(np.log10(0.02), np.log10(1.0), num=5)
MCSHs = {
"MCSHs": {
"0": {
"groups": [1],
"sigmas": sigmas
},
"1": {
"groups": [1],
"sigmas": sigmas
},
"2": {
"groups": [1, 2],
"sigmas": sigmas
},
"3": {
"groups": [1, 2, 3],
"sigmas": sigmas
},
"4": {
"groups": [1, 2, 3, 4],
"sigmas": sigmas
},
"5": {
"groups": [1, 2, 3, 4, 5],
"sigmas": sigmas
},
# "6": {"groups": [1, 2, 3, 4, 5, 6, 7], "sigmas": sigmas},
},
"atom_gaussians": {
"H": "../MCSH_potentials/H_pseudodensity_2.g",
"O": "../MCSH_potentials/O_pseudodensity_4.g",
},
"cutoff": 8,
}
elements = ["H", "O"]
config = {
"model": {
"get_forces": True,
"num_layers": num_layers,
"num_nodes": num_nodes
},
"optim": {
"device": "cpu",
"force_coefficient": 0.2,
"lr": learning_rate,
"batch_size": 8,
"epochs": 200,
},
"dataset": {
"raw_data": training_list,
# "val_split": 0.1,
"elements": elements,
"fp_scheme": "gmp",
"fp_params": MCSHs,
"save_fps": True,
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": False,
"logger": False,
},
}
trainer = AtomsTrainer(config)
trainer.train()
predictions = trainer.predict(test_list)
true_energies = np.array(
[image.get_potential_energy() for image in test_list])
pred_energies = np.array(predictions["energy"])
mae_result = np.mean(np.abs(true_energies - pred_energies))
return mae_result
def run_oal(atomistic_method, images, elements, dbname, parent_calc):
Gs = {
"default": {
"G2": {
"etas": np.logspace(np.log10(0.05), np.log10(5.0), num=4),
"rs_s": [0],
},
"G4": {
"etas": [0.005],
"zetas": [1.0, 4.0],
"gammas": [1.0, -1.0]
},
"cutoff": 6,
},
}
learner_params = {
"max_iterations": 10,
"samples_to_retrain": 1,
"filename": "relax_example",
"file_dir": "./",
"uncertain_tol": 5.0,
"fmax_verify_threshold": 0.05, # eV/AA
"relative_variance": True,
"n_ensembles": 10,
"use_dask": True,
}
config = {
"model": {
"get_forces": True,
"num_layers": 3,
"num_nodes": 5
},
"optim": {
"device": "cpu",
"force_coefficient": 4.0,
"lr": 1,
"batch_size": 10,
"epochs": 100,
"optimizer": torch.optim.LBFGS,
"optimizer_args": {
"optimizer__line_search_fn": "strong_wolfe"
},
},
"dataset": {
"raw_data": images,
"val_split": 0,
"elements": elements,
"fp_params": Gs,
"save_fps": False,
"scaling": {
"type": "standardize"
},
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": False,
# "logger": True,
"single-threaded": True,
},
}
trainer = AtomsTrainer(config)
onlinecalc = OnlineLearner(
learner_params,
trainer,
images,
parent_calc,
)
if os.path.exists("dft_calls.db"):
os.remove("dft_calls.db")
atomistic_method.run(onlinecalc, filename=dbname)
return onlinecalc, atomistic_method
MCSHs = MCSHs_dict[MCSHs_index]
MCSHs = {
"MCSHs": MCSHs,
"atom_gaussians": potential_files,
"cutoff": cutoff_distance
}
elements = [
'Pt', 'Al', 'V', 'Pd', 'Fe', 'Sn', 'Ge', 'Bi', 'Ir', 'Re', 'Cd', 'Cr',
'Ag', 'Hf', 'Ru', 'Ti', 'Cs', 'Os', 'N', 'As', 'O', 'S', 'Mo', 'Ta', 'Zn',
'Y', 'Mn', 'Na', 'Rh', 'Hg', 'C', 'Co', 'Nb', 'Sc', 'Sr', 'H', 'Au', 'Ga',
'Tl', 'K', 'Se', 'B', 'Pb', 'Ca', 'Cl', 'Cu', 'Zr', 'Rb', 'P', 'W', 'Tc',
'Te', 'Ni', 'Sb', 'Si', 'In'
]
trainer = AtomsTrainer()
trainer.load_pretrained(checkpoint_name, gpu2cpu=True)
trainer.config["dataset"]["save_fps"] = False
result_dirname = "./test_result_val_id/sigma{}_MCSH{}_nodes{}_layers{}_cutoff{}_numtraining{}_results".format(
sigmas_index, MCSHs_index, num_nodes, num_layers, cutoff_distance,
num_training)
test_mae = predict_data(trainer,
test_images,
set_index,
result_dirname=result_dirname,
image_type="test")
"elements": elements,
"fp_scheme": "mcsh",
"fp_params": MCSHs,
"save_fps": True,
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": True,
"logger": False,
},
}
trainer = AtomsTrainer(config)
trainer.train()
dataset = trainer.train_dataset
position_array = []
fp_array = []
energy_array = []
for i, data in enumerate(dataset):
position = images[i].get_positions()
position_array.append(np.asarray(position))
fingerprint = data.fingerprint.numpy()
fp_array.append(fingerprint)
energy = images[i].get_potential_energy()
energy_array.append(energy)
"val_split": 0,
"elements": elements,
"fp_params": Gs,
"save_fps": True,
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": True,
"logger": False,
},
}
trainer = AtomsTrainer(config)
# building base morse calculator as base calculator
cutoff = Gs["default"]["cutoff"]
base_calc = MultiMorse(images, cutoff, combo="mean")
# define learner_params OfflineActiveLearner
learner_params = {
"atomistic_method":
Relaxation(initial_geometry=slab.copy(),
optimizer=BFGS,
fmax=0.01,
steps=100),
"max_iterations":
def run_offline_al(atomistic_method, images, dbname, parent_calc):
Gs = {
"default": {
"G2": {
"etas": np.logspace(np.log10(0.05), np.log10(5.0), num=4),
"rs_s": [0] * 4,
},
"G4": {"etas": [0.005], "zetas": [1.0, 4.0], "gammas": [1.0, -1.0]},
"cutoff": 6,
},
}
elements = np.unique(images[0].get_chemical_symbols())
learner_params = {
"atomistic_method": atomistic_method,
"max_iterations": 10,
"force_tolerance": 0.01,
"samples_to_retrain": 2,
"filename": "relax_example",
"file_dir": "./",
"query_method": "random",
"use_dask": False,
"max_evA": 0.05,
}
config = {
"model": {
"get_forces": True,
"num_layers": 3,
"num_nodes": 20,
},
"optim": {
"device": "cpu",
"force_coefficient": 30,
"lr": 1,
"batch_size": 10,
"epochs": 100, # was 100
"loss": "mse",
"metric": "mae",
"optimizer": torch.optim.LBFGS,
"optimizer_args": {"optimizer__line_search_fn": "strong_wolfe"},
},
"dataset": {
"raw_data": images,
"val_split": 0,
"elements": elements,
"fp_params": Gs,
"save_fps": False,
"scaling": {"type": "normalize", "range": (-1, 1)},
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": True,
# "logger": True,
"single-threaded": True,
},
}
trainer = AtomsTrainer(config)
cutoff = Gs["default"]["cutoff"]
base_calc = MultiMorse(images, cutoff, combo="mean")
learner = FmaxLearner(
learner_params,
images,
trainer,
parent_calc,
base_calc,
)
learner.learn()
trained_calc = learner.trained_calc
al_iterations = learner.iterations - 1
file_path = learner_params["file_dir"] + learner_params["filename"]
final_ml_traj = ase.io.read("{}_iter_{}.traj".format(file_path, al_iterations), ":")
if os.path.exists("dft_calls.db"):
os.remove("dft_calls.db")
# atomistic_method.run(learner, filename=dbname)
return learner, trained_calc, final_ml_traj
"fp_params": Gs,
"save_fps": True,
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": True,
"logger": False,
},
}
config["dataset"]["cutoff_params"] = cosine_cutoff_params
torch.set_num_threads(1)
cosine_trainer = AtomsTrainer(config)
cosine_trainer.train()
predictions = cosine_trainer.predict(images)
true_energies = np.array([image.get_potential_energy() for image in images])
cosine_pred_energies = np.array(predictions["energy"])
image.set_calculator(AMPtorch(cosine_trainer))
image.get_potential_energy()
config["dataset"]["cutoff_params"] = polynomial_cutoff_params
torch.set_num_threads(1)
polynomial_trainer = AtomsTrainer(config)
polynomial_trainer.train()
def offline_neb(images, parent_calc, iter=4, intermediate_images=3):
torch.set_num_threads(1)
parent_calc = parent_calc
Gs = {
"default": {
"G2": {
"etas": np.logspace(np.log10(0.05), np.log10(5.0), num=4),
"rs_s": [0],
},
"G4": {
"etas": [0.005],
"zetas": [1.0, 4.0],
"gammas": [1.0, -1.0]
},
"cutoff": 5.0,
},
}
elements = ["Cu", "C"]
config = {
"model": {
"get_forces": True,
"num_layers": 3,
"num_nodes": 20,
"activation": Tanh,
},
"optim": {
"device": "cpu",
"force_coefficient": 27,
"lr": 1e-2,
"batch_size": 1000,
"epochs": 300,
"loss": "mse",
"metric": "mse",
"optimizer": torch.optim.LBFGS,
"optimizer_args": {
"optimizer__line_search_fn": "strong_wolfe"
},
"scheduler": {
"policy": torch.optim.lr_scheduler.CosineAnnealingWarmRestarts,
"params": {
"T_0": 10,
"T_mult": 2
},
},
},
"dataset": {
"raw_data": images,
"val_split": 0,
"elements": elements,
"fp_params": Gs,
"save_fps": True,
"scaling": {
"type": "normalize",
"range": (-1, 1)
},
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": True,
"logger": False,
"dtype": torch.DoubleTensor,
},
}
trainer = AtomsTrainer(config)
# building base morse calculator as base calculator
cutoff = Gs["default"]["cutoff"]
neb_images = images.copy()
base_calc = MultiMorse(neb_images, cutoff, combo="mean")
# base_calc = Dummy(images)
# define learner_params OfflineActiveLearner
learner_params = {
"atomistic_method":
NEBcalc(
starting_images=neb_images,
intermediate_samples=intermediate_images,
),
"max_iterations":
iter,
"samples_to_retrain":
intermediate_images,
"filename":
"example",
"file_dir":
"./",
"use_dask":
False,
# "max_evA": 0.01,
}
learner = NEBLearner(learner_params, images, trainer, parent_calc,
base_calc)
learner.learn()
return learner
def load_trainer(checkpoint_path):
trainer = AtomsTrainer()
# loading the pretrained model
trainer.load_pretrained(checkpoint_path)
return trainer
"batch_size": batch_size,
"epochs": epochs,
"loss": "mae",
},
"dataset": {
"lmdb_path": lmdb_paths,
"val_split": val_split,
"val_split_mode": "inorder",
"cache": "partial"
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": True,
# Weights and Biases used for logging - an account(free) is required
"logger": False,
},
}
trainer = AtomsTrainer(config)
if os.path.isdir(checkpoint_name):
trainer.load_pretrained(checkpoint_name)
else:
print(
"**** WARNING: checkpoint not found: {} ****".format(checkpoint_name))
print("training")
trainer.train()
print("end training")
def calculate_energy(x0):
OH_bond_length = x0[0]
bond_angle = x0[1]
images = read("../data/water.traj", index=":")
images = [images[0]]
sigmas = np.logspace(np.log10(0.02), np.log10(1.0), num=5)
MCSHs = {
"MCSHs": {
"0": {"groups": [1], "sigmas": sigmas},
"1": {"groups": [1], "sigmas": sigmas},
"2": {"groups": [1, 2], "sigmas": sigmas},
"3": {"groups": [1, 2, 3], "sigmas": sigmas},
"4": {"groups": [1, 2, 3, 4], "sigmas": sigmas},
"5": {"groups": [1, 2, 3, 4, 5], "sigmas": sigmas},
# "6": {"groups": [1, 2, 3, 4, 5, 6, 7], "sigmas": sigmas},
},
"atom_gaussians": {
"H": "../MCSH_potentials/H_pseudodensity_2.g",
"O": "../MCSH_potentials/O_pseudodensity_4.g",
},
"cutoff": 8,
}
elements = ["H", "O"]
config = {
"model": {"get_forces": True, "num_layers": 3, "num_nodes": 20},
"optim": {
"device": "cpu",
"force_coefficient": 0.2,
"lr": 1e-3,
"batch_size": 8,
"epochs": 500,
},
"dataset": {
"raw_data": images,
# "val_split": 0.1,
"elements": elements,
"fp_scheme": "mcsh",
"fp_params": MCSHs,
"save_fps": True,
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": True,
"logger": False,
},
}
trainer = AtomsTrainer(config)
# loading the pretrained model
trainer.load_pretrained("../checkpoints/2021-03-22-14-02-20-test")
calculator = AMPtorch(trainer)
image = molecule('H2O')
image.set_distance(0, 2, OH_bond_length, fix=0)
image.set_angle(1, 0, 2, bond_angle)
image.set_cell([10, 10, 10])
image.center()
image.set_calculator(calculator)
return image.get_potential_energy()
"batch_size": 10,
"epochs": 100,
},
"dataset": {
"raw_data": images,
"val_split": 0,
"elements": elements,
"fp_scheme": "mcsh",
"fp_params": MCSHs,
"save_fps": True,
},
"cmd": {
"debug": False,
"run_dir": "./",
"seed": 1,
"identifier": "test",
"verbose": True,
"logger": False,
},
}
trainer = AtomsTrainer(config)
trainer.train()
predictions = trainer.predict(images[:10])
true_energies = np.array([image.get_potential_energy() for image in images])
pred_energies = np.array(predictions["energy"])
print("Energy MSE:", np.mean((true_energies - pred_energies)**2))
def objective_function(rank, scratch_dir, params):
train_images = Trajectory('train.traj')
test_images = Trajectory('test.traj')
elements = np.unique([atom.symbol for atom in train_images[0]])
cutoff = 6.0
cosine_cutoff_params = {'cutoff_func': 'cosine'}
gds = GaussianDescriptorSet(elements, cutoff, cosine_cutoff_params)
g2_etas = [0.25, 2.5, 0.25, 2.5]
g2_rs_s = [0.0, 0.0, 3.0, 3.0]
gds.batch_add_descriptors(2, g2_etas, g2_rs_s, [])
g4_etas = [0.005, 0.005, 0.01, 0.01]
g4_zetas = [1.0, 4.0, 4.0, 16.0]
g4_gammas = [1.0, 1.0, -1.0, -1.0]
gds.batch_add_descriptors(4, g4_etas, g4_zetas, g4_gammas)
amptorch_config = {
'model': {
'get_forces': True,
'num_layers': params['num_layers'],
'num_nodes': params['num_nodes'],
'batchnorm': False,
},
'optim': {
'force_coefficient': 0.04,
'lr': 1e-2,
'batch_size': 32,
'epochs': 100,
'loss': 'mse',
'metric': 'mae',
'gpus': 0,
},
'dataset': {
'raw_data': train_images,
'val_split': 0.1,
'fp_params': gds,
'save_fps': True,
'scaling': {
'type': 'normalize',
'range': (0, 1)
},
},
'cmd': {
'debug': False,
'run_dir': scratch_dir,
'seed': 1,
'identifier': 'rank{}'.format(rank),
'verbose': False,
'logger': False,
},
}
mse = None
with NoLogging():
# train on train_images.traj
torch.set_num_threads(1)
trainer = AtomsTrainer(amptorch_config)
trainer.train()
# evaluate on test_images.traj
predictions = trainer.predict(test_images)
true_energies = np.array(
[image.get_potential_energy() for image in test_images])
pred_energies = np.array(predictions['energy'])
mse = np.mean((true_energies - pred_energies)**2)
return mse
