def first_estimations(distributions, strats, cluster, cluster_index):
"""
"""
epoch = 1
for s in strats:
i = 0
while i < settings.N:
#n, z = get_rand()
n, z = launch_simulation_with_strategy(s, cluster)
distributions[str(s[0])].get_posterior(n , z)
plot.plot_distributions(distributions, cluster_index, epoch)
i += 1
epoch += 1
def simulate_pairs(distributions, strats, cluster, cluster_index, d1, d2, epoch):
"""
"""
r = 0
while r < settings.N:
epoch += 1
s = strats[int(d1)]
n, z = launch_simulation_with_strategy(s, cluster)
distributions[d1].get_posterior(n , z)
s = strats[int(d2)]
n, z = launch_simulation_with_strategy(s, cluster)
distributions[d2].get_posterior(n , z)
plot.plot_distributions(distributions, cluster_index, epoch)
fig.savefig(
os.path.join(fig_dir,
'new2_JSD_embedding_all_strains_tbins=%d.png' % (tbins)))
#%% Plot Distributions and distances
#for i, strain in enumerate(strains):
for i, s0 in enumerate(strains):
fig = plt.figure(300 + i, figsize=(21, 14))
plt.clf()
ax = plt.subplot(3, 1, 1)
rs = plt.cm.Reds(0.9)
fplt.plot_distributions(
dat_bin[s0],
percentiles_colors=['black', rs, 'red', rs, 'black'],
percentiles=[5, 25, 50, 75, 95],
cmap=plt.cm.viridis)
plt.xlim(0, tbins)
plt.ylim(0, 1)
plt.title('%s %s' % (s0, feat))
plt.subplot(3, 1, 2, sharex=ax)
plt.plot(ent_t_g[s0])
plt.title('JS divergence to average distribution')
plt.xlim(0, len(ent_t_g[s0]))
plt.subplot(3, 1, 3, sharex=ax)
plt.plot(ent_t_s_ref[s0])
plt.title('JS divergence to average stage distributions')
plt.xlim(0, len(ent_t_g[s0]))
plt.tight_layout()
z = random.randint(1, 20)
return n, z
#Global variables
t_begin = time.time()
storage = db.Database(settings.DATABASE)
cluster_dict = storage.build_cluster_dictionnary()
strats = storage.get_detailled_strategies()
l_team = settings.TEAM_PATH + " " + settings.TEAM_PARAM
#For each cluster
for c in cluster_dict:
write_log("cluster " + str(c) + ":\n")
distributions = init_distributions(strats)
plot.plot_distributions(distributions, 0, 0)
cluster = cluster_dict[c]
first_estimations(distributions, strats, cluster, c)
strat_pairs = build_strategies_pairs(strats)
strat_comparisons = init_strat_comparisons(strats)
p = 0
#For each possible pair
while p < len(strat_pairs):
if len(strat_pairs) > 0:
cp = strat_pairs[p]
write_log("\tpair " + str(cp) + ":")
d1 = distributions[str(cp[0])]
d2 = distributions[str(cp[1])]
write_log("[" + str(d1.n) + "," + str(d1.z) + "]-")
write_log("[" + str(d2.n) + "," + str(d2.z) + "]--")
conclusion = fe.formation_evaluator(d1.prior, d2.prior)
#!/usr/bin/env python3 #-*- coding: utf-8 -*- from scipy.stats import beta import numpy as np import corner_kick_distribution as ckd import formation_evaluator as fe import plot test = ckd.CK_Distribution() x = np.linspace(0, 1, 10000) p1 = test.get_posterior(40, 4) p2 = test.get_posterior(45, 3) p3 = test.get_posterior(36, 2) distributions = [p1, p2, p3] plot.plot_distributions(distributions)
