def make_hamiltonian(hamiltonian_type, tau_long, ML_param):
if hamiltonian_type == 'noML':
hamiltonian_obj = noML_hamiltonian()
elif hamiltonian_type == 'pairwise_HNN':
net1 = mlp_net(ML_param.layer_list, ML_param.dropout_list)
net2 = mlp_net(ML_param.layer_list, ML_param.dropout_list)
hamiltonian_obj = pairwise_HNN(net1, net2, ML_param.on_off_noML)
hamiltonian_obj.set_tau_long(tau_long)
elif hamiltonian_type == 'fields_HNN':
net1 = mlp_net(ML_param.layer_list, ML_param.dropout_list)
net2 = mlp_net(ML_param.layer_list, ML_param.dropout_list)
hamiltonian_obj = fields_HNN(net1, net2, ML_param.dgrid,
ML_param.ngrid, ML_param.on_off_noML)
hamiltonian_obj.set_tau_long(tau_long)
else:
assert (False), 'invalid hamiltonian type given'
return hamiltonian_obj
def make_hamiltonian(hamiltonian_type, tau_long, ML_param):
pwnet1 = mlp_net(ML_param.pw_layer_list, ML_param.dropout_list)
pwnet2 = mlp_net(ML_param.pw_layer_list, ML_param.dropout_list)
pwhamiltonian_obj = pairwise_HNN(pwnet1, pwnet2, ML_param.on_off_noML)
pwhamiltonian_obj.set_tau_long(tau_long)
fnet1 = mlp_net(ML_param.f_layer_list, ML_param.dropout_list)
fnet2 = mlp_net(ML_param.f_layer_list, ML_param.dropout_list)
fhamiltonian_obj = fields_HNN(fnet1, fnet2, ML_param.dgrid, ML_param.ngrid,
ML_param.on_off_noML)
fhamiltonian_obj.set_tau_long(tau_long)
if hamiltonian_type == 'noML':
return noML_hamiltonian()
elif hamiltonian_type == 'pairwise_HNN':
return pwhamiltonian_obj
elif hamiltonian_type == 'fields_HNN':
return fhamiltonian_obj
elif hamiltonian_type == 'fields_pairwise_HNN':
fpwhamiltonian_obj = fields_pairwise_HNN(fhamiltonian_obj,
pwhamiltonian_obj)
return fpwhamiltonian_obj
else:
assert (False), 'invalid hamiltonian type given'
import json
import torch
import math
if __name__ == '__main__':
'''python check_gen_by_mc2.py jsonfile1.dict'''
argv = sys.argv
json_file = argv[1]
with open(json_file) as f:
data = json.load(f)
phase_space = phase_space()
hamiltonian_obj = noML_hamiltonian()
filelist = data['filelist']
init_e1 = {}
ke1 = {}
p_dist1 = {}
init_u1 = {}
u1 = {}
f1_magnitude1 = {}
rho = [0.10, 0.14, 0.2, 0.27, 0.38]
for i in range(len(filelist)):
infile = torch.load(filelist[i])
# io varaiables
filename = MC_parameters.filename
# seed setting
np.random.seed(seed)
random.seed(seed)
torch.manual_seed(seed)
torch.set_default_dtype(torch.float64)
_ = system_logs(mydevice)
system_logs.print_start_logs()
data_io_obj = data_io()
phase_space = phase_space.phase_space()
noMLhamiltonian = noML_hamiltonian()
metropolis_mc = metropolis_mc(system_logs)
q_hist, U, ACCRatio, spec = metropolis_mc.step(noMLhamiltonian,
phase_space)
# take q every interval
q_hist = q_hist[:, 0::interval, :, :]
# shape is [mc nsamples, mc interval step, nparticle, DIM]
q_hist = torch.reshape(q_hist, (-1, q_hist.shape[2], q_hist.shape[3]))
# shape is [(mc nsamples x mc step), nparticle, DIM]
# momentum sampler
momentum_sampler = momentum_sampler(q_hist.shape[0])
p_hist = momentum_sampler.momentum_samples()