def rho_heatmaps(N,
F0,
lambdaF_list,
eta_list,
rep,
outDeg=20,
power=-2.5,
M=10,
tol=0.0005,
window_size=0,
t_max=0):
"""Generates data for the density heatmap."""
blockPrint()
g = ig.Graph.Barabasi(N, outDeg, power=power) # generate the BA graph
g.simplify()
g.delete_vertices(g.vs.select(lambda vertex: vertex.degree() == 0))
N = len(g.vs)
E = len(g.es)
final_polarised = {}
final_rhoF = {}
final_FT_edges = {}
for lambda_F in lambdaF_list:
for eta in eta_list:
avg_pol = 0
avg_rhoF = 0
avg_FT = 0
for i in range(rep):
v_polarised, rhoF, nFT_edges = newsFlow(
g=g,
F0=F0,
eta=eta,
lambda_F=lambda_F,
M=M,
t_max=t_max,
window_size=window_size,
tol=tol,
sfreq=-1)
avg_pol += (v_polarised / N)
avg_rhoF += (rhoF / N)
avg_FT += (nFT_edges / E)
avg_pol /= rep
avg_rhoF /= rep
avg_FT /= rep
final_polarised[(lambda_F, eta)] = avg_pol
final_rhoF[(lambda_F, eta)] = avg_rhoF
final_FT_edges[(lambda_F, eta)] = avg_FT
return final_polarised, final_rhoF, final_FT_edges
def plot_newsFlow(N,
F0,
eta,
lambda_F,
outDeg=20,
power=-2.5,
M=10,
tol=0.0005,
window_size=0,
t_max=0):
"""Visualises the simulation data."""
g = ig.Graph.Barabasi(N, outDeg, power=power) # generate the BA graph
g.simplify()
g.delete_vertices(g.vs.select(lambda vertex: vertex.degree() == 0))
N = len(g.vs)
E = len(g.es)
time_range, list_polarised, list_rhoF, list_FT_edges = newsFlow(
g=g,
F0=F0,
eta=eta,
lambda_F=lambda_F,
M=M,
t_max=t_max,
window_size=window_size,
tol=tol,
sfreq=0)
_ = plt.tight_layout()
_ = plt.figure(figsize=(8, 8))
_ = plt.suptitle(
r'$N =$ %s, $F_0 =$ %s, $M =$ %s, $\eta =$ %s, $\lambda_F =$ %s' %
(N, F0, 2 * M, eta, lambda_F),
y=1.01)
_ = plt.tight_layout()
_ = sub_plot(1, time_range, [i / N for i in list_polarised],
r'$\rho^P(t)$')
_ = plt.ylim((0, 1))
_ = sub_plot(2, time_range, [i / N for i in list_rhoF], r'$\rho^F(t)$')
_ = plt.ylim((0, 1))
_ = sub_plot(3, time_range, [i / E for i in list_FT_edges],
r'$\rho^{FT}(t)$')
_ = plt.ylim((0, 1))
_ = sub_plot(4, time_range, range(len(time_range)), 'Iterations')
_ = plt.show()
def time_evolution_lambda_F_eta(N,
F0,
list_lambdaF,
list_eta,
outDeg=20,
power=-2.5,
M=10,
tol=0.0005,
window_size=0,
t_max=0):
"""Generates data for the time evolution in dependence on lambda_F and eta."""
blockPrint()
g = ig.Graph.Barabasi(N, outDeg, power=power) # generate the BA graph
g.simplify()
g.delete_vertices(g.vs.select(lambda vertex: vertex.degree() == 0))
N = len(g.vs)
E = len(g.es)
results_times = {}
results_polarised = {}
results_rhoF = {}
results_FT_edges = {}
for lambda_F in list_lambdaF:
for eta in list_eta:
time_range, list_polarised, list_rhoF, list_FT_edges = newsFlow(
g=g,
F0=F0,
eta=eta,
lambda_F=lambda_F,
M=M,
t_max=t_max,
window_size=window_size,
tol=tol,
sfreq=0)
results_times[(lambda_F, eta)] = time_range
results_polarised[(lambda_F,
eta)] = [i / N for i in list_polarised]
results_rhoF[(lambda_F, eta)] = [i / N for i in list_rhoF]
results_FT_edges[(lambda_F, eta)] = [i / E for i in list_FT_edges]
return results_times, results_rhoF, results_FT_edges, results_polarised
def get_media_subscriptions(N,
lambda_F,
eta,
sfreq,
rhoF0=0.5,
outDeg=20,
power=-2.5,
M=10,
tol=0.0005,
window_size=0,
t_max=0):
"""Generates media-reach statistics."""
blockPrint()
g = ig.Graph.Barabasi(N, outDeg, power=power) # generate the BA graph
g.simplify()
g.delete_vertices(g.vs.select(lambda vertex: vertex.degree() == 0))
N = len(g.vs)
F0 = int(N * rhoF0)
list_rhoF, T_Fmedia, F_Fmedia, T_Tmedia, F_Tmedia, Tmedia_T, Tmedia_F, Fmedia_T, Fmedia_F = newsFlow(
g=g,
F0=F0,
eta=eta,
lambda_F=lambda_F,
M=M,
t_max=t_max,
window_size=window_size,
tol=tol,
sfreq=sfreq)
df = pd.DataFrame(columns=['t_step', 'type', '#F_subscribers'
]) # store the results as a dataframe
for t_step in Tmedia_T.keys():
TF = Tmedia_F[t_step]
TT = Tmedia_T[t_step]
temp = pd.DataFrame({
't_step': [round(t_step) for i in range(len(TF))],
'type': [True for i in range(len(TF))],
'#F_subscribers':
[TF[i] / (TF[i] + TT[i]) for i in range(len(TF))]
})
df = pd.concat([df, temp])
for t_step in Fmedia_F.keys():
FF = Fmedia_F[t_step]
FT = Fmedia_T[t_step]
temp = pd.DataFrame({
't_step': [round(t_step) for i in range(len(FF))],
'type': [False for i in range(len(FF))],
'#F_subscribers':
[FF[i] / (FF[i] + FT[i]) for i in range(len(FF))]
})
df = pd.concat([df, temp])
df['#F_subscribers'] = df['#F_subscribers'].astype(float)
return df, lambda_F, eta, Tmedia_T, Tmedia_F, Fmedia_T, Fmedia_F, M, N, list_rhoF
def rho_eta(list_N,
list_rhoF0,
rep,
list_eta,
lambda_F=1,
outDeg=20,
power=-2.5,
M=10,
tol=0.0005,
window_size=0,
t_max=0):
"""Generates data for the evolution of the densities in dependence on eta."""
blockPrint()
final_rhoF_eta = {}
final_rhoFT_eta = {}
final_rhoP_eta = {}
for N in list_N:
g = ig.Graph.Barabasi(N, outDeg, power=power) # generate the BA graph
g.simplify()
g.delete_vertices(g.vs.select(lambda vertex: vertex.degree() == 0))
N = len(g.vs)
E = len(g.es)
for rhoF0 in list_rhoF0:
F0 = int(N * rhoF0)
final_rhoF_eta[(N, rhoF0)] = {}
final_rhoFT_eta[(N, rhoF0)] = {}
final_rhoP_eta[(N, rhoF0)] = {}
for eta in list_eta:
avg_rhoF = 0
avg_FT = 0
avg_pol = 0
for i in range(rep):
v_polarised, rhoF, nFT_edges = newsFlow(
g=g,
F0=F0,
eta=eta,
lambda_F=lambda_F,
M=M,
t_max=t_max,
window_size=window_size,
tol=tol,
sfreq=-1)
avg_rhoF += rhoF / N
avg_FT += nFT_edges / E
avg_pol += v_polarised / N
avg_rhoF /= rep
avg_FT /= rep
avg_pol /= rep
final_rhoF_eta[(N, rhoF0)][eta] = avg_rhoF
final_rhoFT_eta[(N, rhoF0)][eta] = avg_FT
final_rhoP_eta[(N, rhoF0)][eta] = avg_pol
return final_rhoF_eta, final_rhoFT_eta, final_rhoP_eta, lambda_F