def reset(self):
genericmodel.__init__(self, self.id, [], [], {}, {})
self.alpha = []
if 'sigma' not in self.params.keys():
self.params['sigma'] = 0.1
if 'lamda' not in self.params.keys():
self.params['lamda'] = 0.0001
if not 'two_body_scaling' in self.params:
self.params['two_body_scaling'] = np.sqrt(8)
if not 'three_body_scaling' in self.params:
self.params['three_body_scaling'] = 1.6
if not 'two_body_width' in self.params:
self.params['two_body_width'] = 0.2
if not 'three_body_width' in self.params:
self.params['three_body_width'] = np.pi
if not 'two_body_power' in self.params:
self.params['two_body_power'] = 4.0
if not 'three_body_power' in self.params:
self.params['three_body_power'] = 2.0
if not 'cut_start' in self.params:
self.params['cut_start'] = 1.0
if not 'alchemy_period_width' in self.params:
self.params['alchemy_period_width'] = 1.6
if not 'alchemy_group_width' in self.params:
self.params['alchemy_group_width'] = 1.6
def reset(self):
genericmodel.__init__(self, self.id, [], [], {}, {})
self.alpha = []
if 'sigma' not in self.params.keys():
self.params['sigma'] = 0.1
if 'lamda' not in self.params.keys():
self.params['lamda'] = 0.0001
def __init__(self, id='KRRmodel', x=[], y=[], params={}, model_args={}):
genericmodel.__init__(self, id, x, y, params, model_args)
self.alpha = []
if 'sigma' not in self.params.keys():
self.params['sigma'] = 0.1
if 'lamda' not in self.params.keys():
self.params['lamda'] = 0.0001
def __init__(self,
mol_order,
id='TFMmodel',
x=[],
y=[],
r=[],
files=[],
params={},
model_args={}):
torch.cuda.empty_cache()
genericmodel.__init__(self, id, x, y, params, model_args)
self.r = r
self.molfiles = files
self.mol_order = mol_order
self.N_atypes = [5, 10, 15]
self.N_btypes = [30, 33, 33]
self.N_ttypes = [23, 79]
self.N_qtypes = [40]
if len(x) > 0:
with gzip.open(self.train_x[0], "rb") as f:
train_dataset = TensorDataset(*pickle.load(f))
self.N_atypes = [
int(train_dataset.tensors[1][:, :, i].max()) for i in range(3)
] # Atom hierarchy has 3 levels
self.N_btypes = [
int(train_dataset.tensors[3][:, :, i].max()) for i in range(3)
] # Bond hierarchy has 3 levels
self.N_ttypes = [
int(train_dataset.tensors[5][:, :, i].max()) for i in range(2)
] # Triplet hierarchy has 2 levels
self.N_qtypes = [
int(train_dataset.tensors[7][:, :, i].max()) for i in range(1)
] # Quad hierarchy has only 1 level
def __init__(self, id='FCHLmodel', x=[], y=[], params={}, model_args={}):
genericmodel.__init__(self, id, x, y, params, model_args)
if not 'cutoff' in self.params:
self.params['cutoff'] = 5.0
if not 'two_body_scaling' in self.params:
self.params['two_body_scaling'] = np.sqrt(8)
if not 'three_body_scaling' in self.params:
self.params['three_body_scaling'] = 1.6
if not 'two_body_width' in self.params:
self.params['two_body_width'] = 0.2
if not 'three_body_width' in self.params:
self.params['three_body_width'] = np.pi
if not 'two_body_power' in self.params:
self.params['two_body_power'] = 4.0
if not 'three_body_power' in self.params:
self.params['three_body_power'] = 2.0
if not 'cut_start' in self.params:
self.params['cut_start'] = 1.0
if not 'alchemy_period_width' in self.params:
self.params['alchemy_period_width'] = 1.6
if not 'alchemy_group_width' in self.params:
self.params['alchemy_group_width'] =1.6
def __init__(self, id='FCHLmodel', x=[], y=[], params={}, model_args={}):
genericmodel.__init__(self, id, x, y, params, model_args)
self.check_params()
def __init__(self, id='NNmodel', x=[], y=[], params={}, model_args={}):
genericmodel.__init__(self, id, x, y, params, model_args)
def __init__(self, id='KRRmodel', x=[], y=[], params={}, model_args={}):
genericmodel.__init__(self, id, x, y, params, model_args)
self.alpha = []