"eta": 4.0,
"zeta": 8.0
}
idx = list(range(n_samples))
splitter = modsel.KFold(n_splits=3, random_state=42, shuffle=True)
indices = splitter.split(idx)
all_scores = []
for item in indices:
estimator = ARMP_G(iterations=1,
representation='acsf',
representation_params=acsf_params,
batch_size=250,
tensorboard=True)
estimator.set_xyz(xyz)
estimator.set_classes(zs)
estimator.set_properties(ene)
estimator.set_gradients(forces)
estimator.generate_representation(method='fortran')
idx_train = item[0]
idx_test = item[1]
estimator.fit(idx_train)
score = estimator.score(idx_test)
float(split_line[3])
]
zs.append(dic[split_line[0]])
xyz.append(xyz_atom)
counter += 1
for i in range(19):
xyz.append([0, 0, 0])
zs.append(0)
estimator = ARMP_G(iterations=50,
representation='acsf',
representation_params={
"nRs2": 5,
"nRs3": 5,
"nTs": 2
},
tensorboard=False,
store_frequency=1,
batch_size=20)
estimator.load_nn("saved_model")
xyz = np.asarray(xyz)
xyz = np.reshape(xyz, (1, xyz.shape[0], xyz.shape[1]))
zs = np.asarray(zs, dtype=np.int32)
zs = np.reshape(zs, (1, zs.shape[0]))
print(xyz.shape, zs.shape)
ene_2, f_2 = estimator.predict_from_xyz(xyz, zs)
from qml.aglaia.aglaia import ARMP_G
import h5py
from sklearn import model_selection as modsel
data = h5py.File("/Volumes/Transcend/data_sets/CN_isopentane/pruned_dft_with_forces/pruned_isopentane_cn_dft.hdf5", "r")
n_samples = 100
xyz = np.array(data.get("xyz")[:n_samples], dtype=np.float32)
ene = np.array(data.get("ene")[:n_samples], dtype=np.float32)*2625.50
ene = ene - data.get("ene")[0]*2625.50
zs = np.array(data["zs"][:n_samples], dtype=np.int32)
forces = np.array(data.get("forces")[:n_samples], dtype=np.float32)
acsf_params = {"nRs2":5, "nRs3":5, "nTs":5, "rcut":5, "acut":5, "zeta":220.127, "eta":30.8065}
estimator = ARMP_G(iterations=5000, l1_reg=0.0, l2_reg=0.0, learning_rate=0.075,
representation_name='acsf', representation_params=acsf_params, tensorboard=True, store_frequency=2)
estimator.set_xyz(xyz)
estimator.set_classes(zs)
estimator.set_properties(ene)
estimator.set_gradients(forces)
estimator.generate_representation()
idx = np.arange(0, n_samples)
idx_train, idx_test = modsel.train_test_split(idx, test_size=0, random_state=42, shuffle=True)
# estimator.load_nn()
estimator.fit(idx_train)
# print("Done the fitting")
import h5py
from qml.aglaia.aglaia import ARMP_G
from sklearn.base import clone
data = h5py.File("/Volumes/Transcend/data_sets/CN_isobutane_model/pruned_cn_isobutane/train_isopentane_cn_dft.hdf5", "r")
n_samples = 2
xyz = np.array(data.get("xyz")[:n_samples])
ene = np.array(data.get("ene")[:n_samples])*4.184
ene = ene - ene[0]
zs = np.array(data["zs"][:n_samples], dtype = int)
forces = np.array(data.get("forces")[:n_samples])*4.184
acsf_params = {"nRs2":10, "nRs3":10, "nTs": 5, "eta2":4.0, "eta3":4.0, "zeta":8.0}
estimator = ARMP_G(iterations=1, representation='acsf', representation_params=acsf_params, batch_size=2)
estimator.set_xyz(xyz)
estimator.set_classes(zs)
estimator.set_properties(ene)
estimator.set_gradients(forces)
estimator.generate_representation(method='fortran')
pickle.dump(estimator, open('model.pickle', 'wb'))
with open('idx.csv', 'w') as f:
for i in range(n_samples):
f.write('%s\n' % i)
xyz = np.array(data.get("xyz"))
ene = np.array(data.get("ene"))*4.184
ene = ene - ene[0]
zs = np.array(data["zs"], dtype = int)
forces = np.array(data.get("forces"))*4.184
n_samples = 10
idx = list(range(n_samples))
splitter = modsel.KFold(n_splits=3, random_state=42, shuffle=True)
indices = splitter.split(idx)
acsf_params = {"radial_rs":np.arange(0,10, 0.6), "angular_rs":np.arange(0.5, 10.5, 0.6), "theta_s": np.arange(0, 3.14, 0.5), "eta":4.0, "zeta":8.0}
estimator = ARMP_G(iterations=2, representation_name='acsf', representation_params=acsf_params, batch_size=512,
tensorboard=True, tensorboard_subdir=os.getcwd(), hidden_layer_sizes=(50,30,10),
l1_reg=1.273530531541467e-10, l2_reg=0.06740360714223229, learning_rate=0.01, store_frequency=10)
# estimator.set_xyz(xyz[:n_samples])
# estimator.set_classes(zs[:n_samples])
# estimator.set_properties(ene[:n_samples])
# estimator.set_gradients(forces[:n_samples])
all_scores = []
counter = 0
for item in indices:
dolly = clone(estimator)
dolly.set_xyz(xyz[:n_samples])
import matplotlib.pyplot as plt
data = h5py.File(
"/Volumes/Transcend/data_sets/CN_isobutane_model/pruned_dft_with_forces/pruned_isopentane_cn_dft.hdf5",
"r")
n_samples = 100
xyz = np.array(data.get("xyz")[-n_samples:])
ene = np.array(data.get("ene")[-n_samples:]) * 2625.50
ene = ene - data.get("ene")[0] * 2625.50
zs = np.array(data["zs"][-n_samples:], dtype=np.int32)
forces = np.array(data.get("forces")[-n_samples:])
idx = list(range(n_samples))
# Creating the estimator
estimator = ARMP_G(iterations=50, batch_size=10)
estimator.set_xyz(xyz)
estimator.set_properties(ene)
estimator.set_classes(zs)
estimator.set_gradients(forces)
estimator.load_nn()
estimator.predict(idx)
estimator.fit(idx)
estimator.score(idx)
"r")
n_samples = 10
xyz = np.array(data.get("xyz")[:n_samples])
ene = np.array(data.get("ene")[:n_samples]) * 4.184
zs = np.array(data["zs"][:n_samples], dtype=np.int32)
forces = np.array(data.get("forces")[:n_samples]) * 4.184
## ------------- ** Setting up the estimator ** ---------------
estimator = ARMP_G(iterations=10,
representation_name='acsf',
representation_params={
"nRs2": 5,
"nRs3": 5,
"nTs": 2
},
tensorboard=False,
store_frequency=1,
batch_size=20)
estimator.set_xyz(xyz[:n_samples])
estimator.set_classes(zs[:n_samples])
estimator.set_properties(ene[:n_samples])
estimator.set_gradients(forces[:n_samples])
idx = np.arange(0, n_samples)
estimator.fit(idx)
print("\n Done the fitting")
import numpy as np
import matplotlib.pyplot as plt
data = h5py.File("/Volumes/Transcend/data_sets/CN_isopentane/pruned_dft_with_forces/pruned_isopentane_cn_dft.hdf5", "r")
n_samples = 10
xyz = np.array(data.get("xyz")[:n_samples])
ene = np.array(data.get("ene")[:n_samples])*2625.50
ene = ene - data.get("ene")[0]*2625.50
zs = np.array(data["zs"][:n_samples], dtype=np.int32)
forces = np.array(data.get("forces")[:n_samples])
# Creating the estimator
estimator = ARMP_G(iterations=10, batch_size=30, hidden_layer_sizes=(40, 20, 10), l2_reg=0.0, tensorboard=False,
store_frequency=5, learning_rate=0.001, representation_params={'nRs2':7, 'nRs3':7, 'nTs':7,
'zeta':220.127, 'eta':30.8065})
estimator.set_xyz(xyz)
estimator.set_properties(ene)
estimator.set_classes(zs)
estimator.set_gradients(forces)
idx = list(range(n_samples))
estimator.load_nn("saved_model")
estimator.fit(idx)
pred_ene, pred_forces = estimator.predict(idx)
score = estimator.score(idx)
data = h5py.File(
"/Volumes/Transcend/data_sets/CN_isobutane_model/pruned_cn_isobutane/train_isopentane_cn_dft.hdf5",
"r")
n_samples = 12
idx = list(range(n_samples))
splitter = modsel.KFold(n_splits=3, random_state=42, shuffle=True)
indices = splitter.split(idx)
xyz = np.array(data.get("xyz")[:n_samples])
ene = np.array(data.get("ene")[:n_samples]) * 4.184
zs = np.array(data["zs"][:n_samples])
forces = np.array(data.get("forces")[:n_samples]) * 4.184
estimator = ARMP_G(representation_params={"nRs2": 5, "nRs3": 5, "nTs": 2})
estimator.load_nn("saved_model")
estimator.set_xyz(xyz[:n_samples])
estimator.set_classes(zs[:n_samples])
estimator.set_properties(ene[:n_samples])
estimator.set_gradients(forces[:n_samples])
# estimator.load_representations_and_dgdr("descrpt_and_grad.hdf5")
idx = np.arange(0, n_samples)
# estimator.fit(idx)
ene_1, f_1 = estimator.predict(idx)
from qml.aglaia.aglaia import ARMP_G
import matplotlib.pyplot as plt
current_dir = os.path.dirname(os.path.realpath(__file__))
data = np.load(current_dir+"/../test/data/CN_isopentane_forces.npz")
xyz = data["arr_0"]
zs = data["arr_1"]
ene = data["arr_2"]
forces = data["arr_3"]
print(xyz.shape, zs.shape, ene.shape, forces.shape)
## ------------- ** Setting up the estimator ** ---------------
estimator = ARMP_G(iterations=1, representation='acsf', representation_params={"radial_rs": np.arange(0,10, 2), "angular_rs": np.arange(0.5, 10.5, 2),
"theta_s": np.arange(0, 3.14, 2.5)}, tensorboard=False, store_frequency=1, batch_size=10)
estimator.set_xyz(xyz[:2])
estimator.set_classes(zs[:2])
estimator.set_properties(ene[:2])
estimator.set_gradients(forces[:2])
estimator.generate_representation()
print(estimator.representation.shape[0])
idx = np.arange(0, 2)
estimator.fit(idx)
ene_1, f_1 = estimator.predict(idx)
ene_2, f_2 = estimator.predict_from_xyz(xyz[:2], zs[:2])
from qml.aglaia.aglaia import ARMP_G
import numpy as np
acsf_params = {
"nRs2": 10,
"nRs3": 10,
"nTs": 5,
"eta2": 4.0,
"eta3": 4.0,
"zeta": 8.0
}
estimator = ARMP_G(iterations=5,
representation='acsf',
representation_params=acsf_params,
batch_size=5,
l1_reg=0.1,
l2_reg=0.3)
parameters = estimator.get_params()
print(parameters["l1_reg"], parameters["l2_reg"])
print(parameters["representation_params"])
forces = np.array(data.get("forces")[:n_samples]) * 2625.5
acsf_param = {
"nRs2": 5,
"nRs3": 5,
"nTs": 5,
"rcut": 5,
"acut": 5,
"zeta": 220.127,
"eta": 30.8065
}
estimator = ARMP_G(iterations=100,
representation_name='acsf',
representation_params=acsf_param,
tensorboard=True,
hidden_layer_sizes=(50, 30, 10),
l1_reg=0,
l2_reg=0,
store_frequency=20,
learning_rate=0.005)
# estimator.load_nn()
estimator.set_xyz(xyz)
estimator.set_classes(zs)
# estimator.generate_representation(xyz, zs, method='fortran')
estimator.set_properties(ene)
estimator.set_gradients(forces)
estimator.generate_representation(method='fortran')
print("Generated all representations.")
current_dir = os.path.dirname(os.path.realpath(__file__))
data = np.load(current_dir + "/../../qml/test/data/CN_isopentane_forces.npz")
xyz = data["arr_0"]
zs = data["arr_1"]
ene = data["arr_2"]
forces = data["arr_3"]
## ------------- ** Setting up the estimator ** ---------------
estimator = ARMP_G(iterations=5,
representation_name='acsf',
representation_params={
"nRs2": 5,
"nRs3": 5,
"nTs": 2
},
tensorboard=False,
store_frequency=1,
batch_size=20,
method='fortran')
n_samples = 10
estimator.set_xyz(xyz[:n_samples])
estimator.set_classes(zs[:n_samples])
estimator.set_properties(ene[:n_samples])
estimator.set_gradients(forces[:n_samples])
idx = np.arange(0, n_samples)