import matplotlib.patches as mpatches
path_input = '/home/brainlab/Desktop/Rudas/Data/Ising/HCP/data/hcp_mean/J_ij.csv'
path_output = '/home/brainlab/Desktop/Rudas/Data/Ising/HCP/data/hcp_mean/results/'
# Ising Parameters
temperature_parameters = (
0.05, 5, 5
) # Temperature parameters (initial tempeture, final tempeture, number of steps)
no_simulations = 150 # Number of simulation after thermalization
thermalize_time = 0.3 #
J = to_normalize(np.loadtxt(path_input, delimiter=','))
ts = np.linspace(temperature_parameters[0],
temperature_parameters[1],
num=temperature_parameters[2])
colors = ['red', 'green', 'black', 'blue', 'purple']
f = plt.figure(figsize=(18, 10)) # plot the calculated values
simulated_fc, critical_temperature, E, M, S, H = generalized_ising(
J,
temperature_parameters=temperature_parameters,
no_simulations=no_simulations,
thermalize_time=thermalize_time)
makedir(path_output)
save_results(temperature_parameters, J, E, M, S, H, simulated_fc,
critical_temperature, path_output)
n_time_points = 4000 # Number of simulation after thermalization
thermalize_time = 0.2 #
main_path = '/home/brainlab/Desktop/Popiel/Ising_HCP/'
parcels = ['Aud', 'CinguloOperc', 'CinguloParietal', 'DMN', 'Dorsal', 'FrontoParietal', 'Retrosplenial', 'SMhand',
'SMmouth', 'Ventral', 'Visual']
for parcel in parcels:
print('Running', parcel)
parcel_path = main_path + parcel + '/'
#results_path = sub_path + 'results/'
save_path = parcel_path + '/results/'
makedir(save_path)
if not file_exists(save_path + 'phi.csv'):
Jij = to_normalize(load_matrix(parcel_path + 'Jij_avg.csv'))
start_time = time.time()
simulated_fc, critical_temperature, E, M, S, H, spin_mean, tc = generalized_ising(Jij,
temperature_parameters=temperature_parameters,
n_time_points=n_time_points,
thermalize_time=thermalize_time,
phi_variables=True,
return_tc=True)
print('It took ',time.time()-start_time, 'seconds to fit the generalized Ising model')
makedir(save_path)
parcels = [
'Aud', 'CinguloOperc', 'CinguloParietal', 'DMN', 'Dorsal',
'FrontoParietal', 'Retrosplenial', 'SMhand', 'SMmouth', 'Ventral',
'Visual'
]
for parcel in parcels:
print('Running', parcel)
parcel_path = main_path + parcel + '/'
#results_path = sub_path + 'results/'
save_path = parcel_path + '/results/avg/'
save_path_phi = main_path + parcel + '/phi/'
makedir(save_path_phi)
makedir(save_path)
Jij = to_normalize(load_matrix(parcel_path + 'Jij_avg.csv'))
start_time = time.time()
simulated_fc, critical_temperature, E, M, S, H, spin_mean, tc = generalized_ising(
Jij,
temperature_parameters=temperature_parameters,
n_time_points=n_time_points,
thermalize_time=thermalize_time,
phi_variables=True,
return_tc=True)
def phi_sim_save(size,
output_directory,
count,
temperature_parameters=(-1, 5, 50),
no_simulations=500,
thermalize_time=0.3):
import numpy as np
from projects.generalize_ising_model.core import generalized_ising
import time
from projects.generalize_ising_model.tools.utils import to_save_results, makedir, to_generate_randon_graph, save_graph
from projects.generalize_ising_model.phi_project.utils import to_estimate_tpm_from_ising_model, to_calculate_mean_phi, \
to_save_phi
makedir(output_directory + '/phi/')
makedir(output_directory + '/ising/')
size = int(size)
output_path_phi = output_directory + '/' + 'phi/' + str(count) + '/'
output_path_ising = output_directory + '/' + 'ising/' + str(count) + '/'
if makedir(output_path_ising) and makedir(output_path_phi):
J = save_graph(
output_path_phi + 'Jij_' + str(count) + '.csv',
to_generate_randon_graph(size, isolate=False, weighted=True))
# Ising Parameters
temperature_parameters = temperature_parameters # Temperature parameters (initial tempeture, final tempeture, number of steps)
no_simulations = no_simulations # Number of simulation after thermalization
thermalize_time = thermalize_time #
start_time = time.time()
print('Fitting Generalized Ising model for a ', size, ' by', size,
' random matrix.')
simulated_fc, critical_temperature, E, M, S, H, spin_mean = generalized_ising(
J,
temperature_parameters=temperature_parameters,
no_simulations=no_simulations,
thermalize_time=thermalize_time,
temperature_distribution='log')
to_save_results(temperature_parameters,
J,
E,
M,
S,
H,
simulated_fc,
critical_temperature,
output_path_ising,
temperature_distribution='log')
print('It took ',
time.time() - start_time,
'seconds to fit the generalized ising model')
print('Computing Phi for random ' + str(size) + ' by ' + str(size) +
' matrix')
ts = np.logspace(temperature_parameters[0],
np.log10(temperature_parameters[1]),
temperature_parameters[2])
phi_temperature, phi_sum, phi_sus = [], [], []
cont = 0
start_time = time.time()
for t in ts:
# print('Temperature: ' + str(t))
tpm, fpm = to_estimate_tpm_from_ising_model(J, t)
phi_, phiSum, phiSus = to_calculate_mean_phi(
fpm, spin_mean[:, cont], t)
phi_temperature.append(phi_)
phi_sum.append(phiSum)
phi_sus.append(phiSus)
cont += 1
to_save_phi(ts, phi_temperature, phi_sum, phi_sus, S,
critical_temperature, size, output_path_phi)
print('It takes ',
time.time() - start_time, 'seconds to compute phi for a ', size,
'by', size, ' random matrix.')
else:
print(str(count) + ' is already done!')
print('Extracting groups from: ' + network_name)
for num in n_groups:
print('Number of groups: ' + str(num))
network_centroids = network['Centroid (MNI)'].tolist()
centroid_list_splited = [[np.float(value) for value in centroid.split(' ')] for centroid in network_centroids]
network_centroids_nd = np.asarray(centroid_list_splited)
if num <= network_centroids_nd.shape[0]:
kmeans = KMeans(n_clusters=num)
kmeans.fit(network_centroids_nd)
y_kmeans = kmeans.predict(network_centroids_nd)
indexs = network['ParcelID'].tolist()
new_img_data = np.zeros(data_img.shape)
for index, group in zip(indexs, y_kmeans):
temporal_img_data = np.zeros(data_img.shape)
temporal_img_data[np.where(data_img == index)] = 1
temporal_img_data = ndim.binary_fill_holes(temporal_img_data)
temporal_img_data = ndim.binary_dilation(temporal_img_data)
temporal_img_data = ndim.binary_fill_holes(temporal_img_data)
new_img_data[temporal_img_data] = group + 1
makedir(output_path + network_name + '/')
nib.save(nib.Nifti1Image(new_img_data.astype(np.float32), affine_img), output_path + network_name + '/' + network_name + '_parcellation_' + str(num) + '.nii')