Sz_array = np.zeros([N_iter, L], dtype=np.complex)
ent_array = np.zeros([N_iter, L - 1], dtype=np.double)
num_iter_array = np.zeros([N_iter], dtype=np.int)
################# INITIALIZATION ######################
product_state = [np.array([1., 0.]).reshape([2, 1, 1]) for i in range(L)]
for dep_idx in range(depth):
my_circuit.append([qTEBD.random_2site_U(2) for i in range(L - 1)])
current_depth = dep_idx + 1
mps_of_layer = qTEBD.circuit_2_mps(my_circuit, product_state)
mps_of_last_layer = [A.copy() for A in mps_of_layer[current_depth]]
Sz_array[0, :] = mps_func.expectation_values_1_site(
mps_of_layer[-1], Sz_list)
ent_array[0, :] = mps_func.get_entanglement(mps_of_last_layer)
fidelity_reached = np.abs(mps_func.overlap(target_mps,
mps_of_last_layer))**2
error_list.append(1. - fidelity_reached)
stop_crit = 1e-1
assert np.isclose(mps_func.overlap(target_mps, target_mps), 1.)
for idx in range(0, N_iter):
#################################
#### variational optimzation ####
#################################
mps_of_last_layer, my_circuit = qTEBD.var_circuit(
target_mps, mps_of_last_layer, my_circuit, product_state)
#################
#### Measure ####
#################
mps_func.right_canonicalize(trunc_mps, no_trunc=True)
print("trunc_error : ", trunc_error)
target_mps_ = mps_func.plr_2_lpr(target_mps)
trunc_mps_ = mps_func.plr_2_lpr(trunc_mps)
trunc_error_ = mps_func.MPS_compression_variational(trunc_mps_,
target_mps_,
verbose=1)
print("trunc_error (var) : ", trunc_error_)
trunc_mps = mps_func.lpr_2_plr(trunc_mps_)
mps_func.left_canonicalize(trunc_mps, no_trunc=True)
E_list.append(np.sum(mps_func.expectation_values(trunc_mps, H_list)))
Sz_array[T_idx, :] = mps_func.expectation_values_1_site(
trunc_mps, Sz_list)
ent_array[T_idx, :] = mps_func.get_entanglement(trunc_mps)
# print("ENTANGLEMENT !!!!! : ", mps_func.get_entanglement(trunc_mps))
fidelity_reached = np.abs(mps_func.overlap(target_mps, trunc_mps))**2
error_list.append(1. - fidelity_reached)
t_list.append(T)
# trunc_mps_dir_path = 'data_tebd_dt%e/1d_%s_g%.4f_h%.4f/L%d/trunc_wf_chi%d_%s/' % (data_dt, Hamiltonian, g, h, L, new_chi, data_order)
# if not os.path.exists(trunc_mps_dir_path):
# os.makedirs(trunc_mps_dir_path)
# filename = trunc_mps_dir_path + 'T%.1f.pkl' % T
# pickle.dump(trunc_mps, open(filename, 'wb'))
num_data = len(t_list)
assert num_data > 0
Sz_array = Sz_array[:num_data, :]
mps_func.right_canonicalize(trunc_mps, no_trunc=True)
trunc_mps, trunc_error = mps_func.left_canonicalize(trunc_mps, no_trunc=False, chi=new_chi)
mps_func.right_canonicalize(trunc_mps, no_trunc=True)
print("trunc_error : ", trunc_error)
target_mps_ = mps_func.plr_2_lpr(target_mps)
trunc_mps_ = mps_func.plr_2_lpr(trunc_mps)
trunc_error_ = mps_func.MPS_compression_variational(trunc_mps_, target_mps_, verbose=1)
print("trunc_error (var) : ", trunc_error_)
trunc_mps = mps_func.lpr_2_plr(trunc_mps_)
mps_func.left_canonicalize(trunc_mps, no_trunc=True)
E_list.append(np.sum(mps_func.expectation_values(trunc_mps, H_list)))
Sz_array[T_idx, :] = mps_func.expectation_values_1_site(trunc_mps, Sz_list)
ent_array[T_idx, :] = mps_func.get_entanglement(trunc_mps)
print("ENTANGLEMENT !!!!! : ", mps_func.get_entanglement(trunc_mps))
fidelity_reached = np.abs(mps_func.overlap(target_mps, trunc_mps))**2
error_list.append(1. - fidelity_reached)
t_list.append(T)
trunc_mps_dir_path = '../2_time_evolution/data_tebd/1d_%s_g%.4f_h%.4f/L%d/trunc_wf_chi%d_1st/' % (Hamiltonian, g, h, L, new_chi)
if not os.path.exists(trunc_mps_dir_path):
os.makedirs(trunc_mps_dir_path)
filename = trunc_mps_dir_path + 'T%.1f.pkl' % T
pickle.dump(trunc_mps, open(filename, 'wb'))
num_data = len(t_list)
assert num_data > 0
Sz_array = Sz_array[:num_data, :]