def fit_all_models(self):
functions = {}
inputs_list = [
self.B, self.F,
np.hstack((self.D, self.C)),
np.hstack((self.B, self.C)),
np.hstack((self.A, self.C, self.E)),
np.hstack((self.B, self.C, self.D)),
np.hstack((self.D, self.E, self.C, self.A)),
np.hstack((self.B, self.E, self.C, self.A)),
np.hstack((self.A, self.B, self.C, self.D, self.E)),
np.hstack((self.A, self.B, self.C, self.D, self.E, self.F))
]
output_list = [
self.C, self.Y, self.Y, self.Y, self.Y, self.Y, self.Y, self.Y,
self.Y, self.Y
]
name_list = [
'gp_C', 'gp_A', 'gp_D_C', 'gp_B_C', 'gp_A_C_E', 'gp_B_C_D',
'gp_D_E_C_A', 'gp_B_E_C_A', 'gp_A_B_C_D_E', 'gp_A_B_C_D_E_F'
]
parameter_list = [[1., 1., 0.0001, False], [1., 1., 10., False],
[1., 1., 1., False], [1., 1., 1., False],
[1., 1., 10., False], [1., 1., 1., False],
[1., 1., 10., False], [1., 1., 10., False],
[1., 1., 10., False], [1., 1., 10., False]]
## Fit all conditional models
for i in range(len(inputs_list)):
X = inputs_list[i]
Y = output_list[i]
functions[name_list[i]] = fit_single_GP_model(
X, Y, parameter_list[i])
return functions
def refit_models(self, observational_samples):
Y = np.asarray(observational_samples['Y'])[:, np.newaxis]
N = np.asarray(observational_samples['N'])[:, np.newaxis]
CO = np.asarray(observational_samples['CO'])[:, np.newaxis]
T = np.asarray(observational_samples['T'])[:, np.newaxis]
D = np.asarray(observational_samples['D'])[:, np.newaxis]
P = np.asarray(observational_samples['P'])[:, np.newaxis]
O = np.asarray(observational_samples['O'])[:, np.newaxis]
S = np.asarray(observational_samples['S'])[:, np.newaxis]
L = np.asarray(observational_samples['L'])[:, np.newaxis]
TE = np.asarray(observational_samples['TE'])[:, np.newaxis]
C = np.asarray(observational_samples['C'])[:, np.newaxis]
functions = {}
inputs_list = [
N,
np.hstack((O, S, T, D, TE)),
np.hstack((C, N, L, TE)),
np.hstack((T, S)),
np.hstack((D, S)),
np.hstack((N, O, S, T, D, TE)),
np.hstack((N, T, S)),
np.hstack((N, D, S)),
np.hstack((O, C, N, L, TE, S, T, D)),
np.hstack((T, C, S, TE, L, N)),
np.hstack((T, D, S)),
np.hstack((C, D, S, TE, L, N)),
np.hstack((N, C, T, S, N, L, TE)),
np.hstack((N, T, D, S)),
np.hstack((C, T, D, S, N, L, TE))
]
output_list = [Y, Y, Y, Y, Y, Y, Y, Y, Y, Y, Y, Y, Y, Y, Y]
name_list = [
'gp_N', 'gp_O_S_T_D_TE', 'gp_C_N_L_TE', 'gp_T_S', 'gp_D_S',
'gp_N_O_S_T_D_TE', 'gp_N_T_S', 'gp_N_D_S', 'gp_O_C_N_L_TE_S_T_D',
'gp_T_C_S_TE_L_N', 'gp_T_D_S', 'gp_C_D_S_TE_L_N',
'gp_N_C_T_S_N_L_TE', 'gp_N_T_D_S', 'gp_C_T_D_S_N_L_TE'
]
parameter_list = [[1., 1., 10., False], [1., 1., 1., True],
[1., 1., 1., True], [1., 1., 1., True],
[1., 1., 10., True], [1., 1., 1., False],
[1., 1., 1., False], [1., 1., 1., False],
[1., 1., 1., False], [1., 1., 1., False],
[1., 1., 1., False], [1., 1., 1., False],
[1., 1., 1., False], [1., 1., 1., False],
[1., 1., 1., False]]
## Fit all conditional models
for i in range(len(inputs_list)):
X = inputs_list[i]
Y = output_list[i]
functions[name_list[i]] = fit_single_GP_model(
X, Y, parameter_list[i])
return functions
def fit_all_models(self):
functions = {}
inputs_list = [
self.N,
np.hstack((self.O, self.S, self.T, self.D, self.TE)),
np.hstack((self.C, self.N, self.L, self.TE)),
np.hstack((self.T, self.S)),
np.hstack((self.D, self.S)),
np.hstack((self.N, self.O, self.S, self.T, self.D, self.TE)),
np.hstack((self.N, self.T, self.S)),
np.hstack((self.N, self.D, self.S)),
np.hstack((self.O, self.C, self.N, self.L, self.TE, self.S, self.T,
self.D)),
np.hstack((self.T, self.C, self.S, self.TE, self.L, self.N)),
np.hstack((self.T, self.D, self.S)),
np.hstack((self.C, self.D, self.S, self.TE, self.L, self.N)),
np.hstack(
(self.N, self.C, self.T, self.S, self.N, self.L, self.TE)),
np.hstack((self.N, self.T, self.D, self.S)),
np.hstack(
(self.C, self.T, self.D, self.S, self.N, self.L, self.TE))
]
output_list = [
self.Y, self.Y, self.Y, self.Y, self.Y, self.Y, self.Y, self.Y,
self.Y, self.Y, self.Y, self.Y, self.Y, self.Y, self.Y
]
name_list = [
'gp_N', 'gp_O_S_T_D_TE', 'gp_C_N_L_TE', 'gp_T_S', 'gp_D_S',
'gp_N_O_S_T_D_TE', 'gp_N_T_S', 'gp_N_D_S', 'gp_O_C_N_L_TE_S_T_D',
'gp_T_C_S_TE_L_N', 'gp_T_D_S', 'gp_C_D_S_TE_L_N',
'gp_N_C_T_S_N_L_TE', 'gp_N_T_D_S', 'gp_C_T_D_S_N_L_TE'
]
parameter_list = [[1., 1., 10., False], [1., 1., 1., True],
[1., 1., 1., True], [1., 1., 1., True],
[1., 1., 10., True], [1., 1., 1., False],
[1., 1., 1., False], [1., 1., 1., False],
[1., 1., 1., False], [1., 1., 1., False],
[1., 1., 1., False], [1., 1., 1., False],
[1., 1., 1., False], [1., 1., 1., False],
[1., 1., 1., False]]
## Fit all conditional models
for i in range(len(inputs_list)):
X = inputs_list[i]
Y = output_list[i]
functions[name_list[i]] = fit_single_GP_model(
X, Y, parameter_list[i])
return functions
def refit_models(self, observational_samples):
A = np.asarray(observational_samples['A'])[:, np.newaxis]
B = np.asarray(observational_samples['B'])[:, np.newaxis]
C = np.asarray(observational_samples['C'])[:, np.newaxis]
D = np.asarray(observational_samples['D'])[:, np.newaxis]
E = np.asarray(observational_samples['E'])[:, np.newaxis]
F = np.asarray(observational_samples['F'])[:, np.newaxis]
Y = np.asarray(observational_samples['Y'])[:, np.newaxis]
functions = {}
inputs_list = [
B,
np.hstack((A, C, E)),
np.hstack((D, C)),
np.hstack((B, C)),
np.hstack((B, C, D)),
np.hstack((D, E, C, A)),
np.hstack((B, E, C, A))
]
output_list = [C, Y, Y, Y, Y, Y, Y, Y]
name_list = [
'gp_C', 'gp_A_C_E', 'gp_D_C', 'gp_B_C', 'gp_B_C_D', 'gp_D_E_C_A',
'gp_B_E_C_A'
]
parameter_list = [[1., 1., 10., False], [1., 1., 10., False],
[1., 1., 1., False], [1., 1., 10., False],
[1., 1., 10., False], [1., 1., 10., False],
[1., 1., 10., False]]
## Fit all conditional models
for i in range(len(inputs_list)):
X = inputs_list[i]
Y = output_list[i]
functions[name_list[i]] = fit_single_GP_model(
X, Y, parameter_list[i])
return functions