def __init__(self, smooth=False):
# ========================================== #
# Hyper-Parameters values to test
# ========================================== #
self.w_1 = [1, 2]
self.w_2 = [1, 2]
self.w_3 = [1, 2]
self.w_4 = [1, 2]
self.w_5 = [1, 2]
# Binomial smoother
self.binom_smoother = [2, 3, 4, 5]
# Optimizer to choose:
self.optimizer = 'LBFGSB'
# Step size (if LBFGSB)
self.step_size = [0.1, 0.05, 0.01]
self.model = SEIR.SEIR()
self.smoothing = smooth
if self.smoothing:
self.model.smoothing = True
else:
self.model.smoothing = False
self.model.import_dataset()
self.model.fit_type = 'type_1'
def __init__(self, smooth=False):
# ========================================== #
# Hyper-Parameters values to test
# ========================================== #
self.w_1 = [1, 2, 3]
self.w_2 = [1, 2, 3]
self.w_3 = [1, 2, 3]
self.w_4 = [1, 2, 3]
self.w_5 = [1, 2, 3]
self.w_6 = [1, 2, 3]
self.smoothing = smooth
# Binomial smoother
self.b_s_1 = [1, 2, 3, 4, 5, 6, 7, 8]
self.b_s_2 = [1, 2, 3, 4, 5, 6, 7, 8]
self.b_s_3 = [1, 2, 3, 4, 5, 6, 7, 8]
self.b_s_4 = [1, 2, 3, 4, 5, 6, 7, 8]
self.b_s_5 = [1, 2, 3, 4, 5, 6, 7, 8]
self.b_s_6 = [1, 2, 3, 4, 5, 6, 7, 8]
# Optimizer to choose: 0 = Cobyla, 1 = LBFGSB, 2 = Auto
self.optimizer = ['COBYLA', 'LBFGSB', 'AUTO']
# Spare in parts
self.part = [[1, 2], [3, 4], [5, 6], [7, 8]]
self.model = SEIR.SEIR()
if self.smoothing:
self.model.smoothing = True
self.model.import_dataset()
self.model.fit_type = 'type_1'
def step(self):
ratios = [
SEIR.count_exposed(population) / population.population_size
for population in self.populations
]
for population in self.populations:
population.step()
self.cross_influence(ratios)
'AR': 1 / 5,
'YR': 1 / 5,
'death rate': 0.01,
'immunity period': None,
}
def get_metric(df):
return df['Removed'].max()
p_values = np.linspace(0, 1, num=2)
removed_count = []
for p in p_values:
# m_values = [5*x for x in range(1, 50)]
# for m in m_values:
logger.info("Starting simulation for m = " + str(p) + "\n")
graph = nx.powerlaw_cluster_graph(model_parameters['population size'],
100,
p,
seed=5)
logger.info("Initialized graph")
model = SEIR.Population(graph, model_parameters)
model.run(50)
df = model.get_data()
removed_count.append(get_metric(df))
print(removed_count)
plt.plot(p_values, removed_count)
plt.show()
populations = []
model_parameters = {
'population size': 10000,
'initial outbreak size': 100,
'alpha': 0.7,
'spread_chance': 0.005,
'EAY': 1 / 5,
'AR': 1 / 5,
'YR': 1 / 5,
'death rate': 0,
'immunity period': 20
}
# Create the first population
graph = nx.powerlaw_cluster_graph(model_parameters['population size'], 100,
0.01)
logger.info("Initialized graph")
model = SEIR.Population(graph, model_parameters)
populations.append(model)
# Make adjustments on the second population
graph_2 = nx.powerlaw_cluster_graph(model_parameters['population size'], 100,
0.01)
model_2 = SEIR.Population(graph_2, model_parameters, social_distancing_func)
populations.append(model_2)
meta_population = MetaPopulation(populations, cross_influence_matrix)
meta_population.run(120)
meta_population.save_model("Models/meta_population_osc_control_lowimun")
def social_distancing_func_simple(t):
if 15 < t < 25:
return 0.4
else:
return 1
model_steps = 50
graph = nx.powerlaw_cluster_graph(model_parameters['population size'],
100,
0.01,
seed=5)
logger.info("Initialized graph")
model = SEIR.Population(graph, model_parameters, social_distancing_func_simple)
model.run(model_steps)
df = model.datacollector.get_model_vars_dataframe()
fig = df.plot()
#fig.show()
model2 = SEIR.Population(graph, model_parameters)
model2.run(model_steps)
df2 = model2.datacollector.get_model_vars_dataframe()
#fig, axs = plt.subplots(2)
ax = df.plot()
#df2.plot(ax=ax, grid=True, linestyle='--')
ax2 = df2.plot()
ax.show()
ax2.show()
#sdf_values = [social_distancing_func_simple(t) for t in range(0, model_steps)]
#axs[1].plot(sdf_values)
'population size': 10000,
'initial outbreak size': 0,
'alpha': 0.7,
'spread_chance': 0.001,
'EAY': 1 / 10,
'AR': 1 / 10,
'YR': 1 / 10,
'death rate': 0,
'immunity period': None
}
model_parameters = [model_parameters_0, model_parameters_1, model_parameters_2]
for mp in model_parameters:
graph = nx.powerlaw_cluster_graph(mp['population size'], 100, 0.01)
logger.info("Initialized graph")
model = SEIR.Population(graph, mp)
populations.append(model)
meta_population = MetaPopulation.MetaPopulation(populations,
cross_influence_matrix)
meta_population.run(100)
fig, axs = plt.subplots(len(cross_influence_matrix))
dfs = [
pop.datacollector.get_model_vars_dataframe()
for pop in meta_population.populations
]
for ax, df in zip(axs, dfs):
df.plot(ax=ax, grid=True)
ax.set_title("Subpopulation")