if len(sys.argv) >= 8:
kick_seed = int(sys.argv[7])
param = (dt, tSim, N, S, p, num_fact, p_fact,
dzeta, a_pf,
eps,
f_russo, cm, a, U, w, tau_1, tau_2, tau_3_A,
tau_3_B, g_A,
beta, tau, t_0, g, random_seed, p_0, n_p, nSnap,
russo2008_mode, muted_prop)
# Some data are saved using simple text. Others are stored using the
# pkl format, which enables to store any kind of data very easily
key = file_handling.get_key(param)
ksi_i_mu, delta__ksi_i_mu__k, J_i_j_k_l, _ = file_handling.load_network(key)
cue_mask = iteration.get_units_to_cue(cue, kick_seed,
delta__ksi_i_mu__k, muted_prop)
# Time arrays
tS = np.arange(0, tSim, dt)
nT = tS.shape[0]
tSnap = np.linspace(0, tSim, nSnap)
# Debugging
analyseTime = False
analyseDivergence = False
# Plot parameters
lamb = [] # Crossovers
transition_time = []
# simulations = ['ff9fe40ed43a94577c1cc2fea6453bf0']
n_seeds = 1
key = simulations[0]
retrieved = file_handling.load_retrieved_several(n_seeds, key)
crossover = file_handling.load_crossover_several(n_seeds, key)
trans_time = file_handling.load_transition_time(0, key)[0]
(dt, tSim, N, S, p, num_fact, p_fact, dzeta, a_pf, eps, f_russo, cm,
a, U, w, tau_1, tau_2, tau_3_A, tau_3_B, g_A, beta, tau, t_0, g,
random_seed, p_0, n_p, nSnap, russo2008_mode, kick_prop) = \
file_handling.load_parameters(key)
ksi_i_mu, delta__ksi_i_mu__k, J_i_j_k_l, \
C_i_j = file_handling.load_network(key)
pair_crossovers = [[] for pair in range(p**2)]
previous = []
following = []
crossovers = []
C_as = []
C_ad = []
C_ai = []
C_0 = []
for seed in range(n_seeds):
for ind_cue in range(p):
for ind_trans in range(3, len(retrieved[seed][ind_cue]) - 1):
pattA = retrieved[seed][ind_cue][ind_trans]
pattB = retrieved[seed][ind_cue][ind_trans + 1]
import sys
# Local modules
from parameters import dt, tSim, N, S, p, num_fact, p_fact, dzeta, a_pf, eps, \
f_russo, cm, a, U, w, tau_1, tau_2, tau_3_A, tau_3_B, g_A, beta, tau, \
t_0, g, random_seed, p_0, n_p, nSnap, russo2008_mode
import initialisation
import file_handling
import patterns
param = (dt, tSim, N, S, p, num_fact, p_fact, dzeta, a_pf, eps, f_russo, cm, a,
U, w, tau_1, tau_2, tau_3_A, tau_3_B, g_A, beta, tau, t_0, g,
random_seed, p_0, n_p, nSnap, russo2008_mode)
set_name = file_handling.get_pkl_name(param)
if not file_handling.network_exists(set_name):
ksi_i_mu, delta__ksi_i_mu__k = patterns.get_uncorrelated()
J_i_j_k_l = initialisation.hebbian_tensor(delta__ksi_i_mu__k)
file_handling.save_parameters_pkl(param, set_name)
file_handling.save_network(ksi_i_mu, delta__ksi_i_mu__k, J_i_j_k_l,
set_name, param)
ksi_i_mub, delta__ksi_i_mu__kb, J_i_j_k_lb = file_handling.load_network(
set_name)
ksi_i_muc, delta__ksi_i_mu__kc, J_i_j_k_lc = file_handling.load_network(
set_name)
print((ksi_i_mu == ksi_i_mub).all())
print((ksi_i_muc == ksi_i_mub).all())
def sig_fun(r_i_k):
"""Activity of units"""
rMax = np.max(r_i_k)
sig_i_k = np.exp(beta * (r_i_k - rMax + U_i[:, None]))
Z_i = spread_active_inactive_states.dot(
sum_active_inactive_states.dot(sig_i_k))
sig_i_k = sig_i_k / Z_i
return sig_i_k
for ii in range(15):
print(ii)
key = simulations[ii]
[_, durations] = file_handling.load_text(key + '/metrics_cue_%d.txt' % 0)
ksi_i_mu, delta__ksi_i_mu__k, _, C_i_j = file_handling.load_network(key)
coact_pos = file_handling.load_coact_pos(0, key)
coact_neg = file_handling.load_coact_neg(0, key)
(dt, tSim, N, S, p, num_fact, p_fact, dzeta, a_pf, eps, f_russo,
cm, a, U, w, tau_1, tau_2, tau_3_A, tau_3_B, g_A, beta, tau, t_0,
g, random_seed, p_0, n_p, nSnap, russo2008_mode) = \
file_handling.load_parameters(key)
patt_vect = delta__ksi_i_mu__k.astype(float) - a / S
factor = np.sqrt(N * a * (1 - a / S))
patt_vect /= factor
patt_corr = np.dot(np.transpose(patt_vect), patt_vect)
M = np.dot(patt_vect, np.transpose(patt_vect))
# plt.ylim(0, 3200)
plt.legend()
plt.title(r'Latching length')
plt.ylabel(r'$l$')
plt.xlabel(r'$a_{pf}$')
plt.figure('C_as')
n_pairs = int(p * (p - 1) / 2)
n_apf = len(apf_values)
df = np.zeros((len(apf_values) * n_pairs, 2))
for ind_apf in range(len(apf_values)):
apf = apf_values[ind_apf]
key = np.array(simulations)[apf_s == apf][0]
print(apf, key)
ksi_i_mu, _, _ = file_handling.load_network(key)
C1C2C0 = correlations.cross_correlations(ksi_i_mu)
df[ind_apf * n_pairs:(ind_apf + 1) * n_pairs, 0] = apf
df[ind_apf * n_pairs:(ind_apf + 1) * n_pairs, 1] = C1C2C0[:, 0]
sns.violinplot(x=df[:, 0], y=df[:, 1])
plt.xlabel(r'$a_{pf}$')
plt.ylabel(r'$C_{as}$')
plt.figure('C_ad')
n_pairs = int(p * (p - 1) / 2)
n_apf = len(apf_values)
df = np.zeros((len(apf_values) * n_pairs, 2))
for ind_apf in range(len(apf_values)):
apf = apf_values[ind_apf]
key = np.array(simulations)[apf_s == apf][0]
print(apf, key)
