import h5py
from MakeFigure import *
# Read data from file
filename = 'iceberg_mcmc.h5'
hdf5file = h5py.File(filename, 'r')
t1 = hdf5file['data_mcmc/theta1'].value
t2 = hdf5file['data_mcmc/theta2'].value
acc_rate = hdf5file['data_mcmc/acc_rate'].value
print("Acceptance rate = ", acc_rate)
# Mixing graphs for theta1 and theta2
fig = MakeFigure(450, 1)
ax = plt.gca()
ax.set_title('Water Coefficient Mixing', fontsize=12)
ax.set_xlabel('Iterations', fontsize=10)
ax.set_ylabel('Water Coefficient ($Cw$)', fontsize=10)
ax.plot(t1)
fig = MakeFigure(450, 1)
ax = plt.gca()
ax.set_title('Air Coefficient Mixing', fontsize=12)
ax.set_xlabel('Iterations', fontsize=10)
ax.set_ylabel('Air Coefficient ($Ca$)', fontsize=10)
ax.plot(t2)
# Scatter Plot
fig = MakeFigure(450, 1)
ax = plt.gca()
theta1 = np.linspace(1.3, 2, t1_size)
theta2 = np.linspace(1.3, 2, t2_size)
startTime = datetime.now()
print(startTime)
log_prior = np.empty(shape=(t2_size, t1_size))
log_likelihood = np.empty(shape=(t2_size, t1_size))
log_posterior = np.empty(shape=(t2_size, t1_size))
for i in range(t2_size):
for j in range(t1_size):
log_prior[i][j] = post.log_prior([theta1[j], theta2[i]])
log_likelihood[i][j] = post.log_likelihood([theta1[j], theta2[i]])
log_posterior[i][j] = log_prior[i][j] + log_likelihood[i][j]
print(i, datetime.now() - startTime)
fig = MakeFigure(700, 0.75)
ax = plt.gca()
ax.contourf(theta1, theta2, np.exp(np.array(log_prior)))
ax.set_title('Prior', fontsize=16)
ax.set_xlabel('Water Coefficient ($Cw$)', fontsize=12)
ax.set_ylabel('Air Coefficient ($Ca$)', fontsize=12)
fig = MakeFigure(700, 1)
ax = plt.gca()
cont = ax.contourf(theta1, theta2, np.exp(np.array(log_likelihood)))
plt.colorbar(cont)
ax.set_title('Likelihood', fontsize=16)
ax.set_xlabel('Water Coefficient ($Cw$)', fontsize=12)
ax.set_ylabel('Air Coefficient ($Ca$)', fontsize=12)
fig = MakeFigure(700, 1)
projection_cluster_assignment1 = get_cluster_assignment(s1_density_clean, projection1, "s1_decision_graph.png", interactive, dir_name)
projection_cluster_assignment2 = get_cluster_assignment(s2_density_clean, projection2, "s2_decision_graph.png", interactive, dir_name)
clusters1, outliers1 = get_top_clusters(5, dihedral_angle1, projection_cluster_assignment1)
clusters2, outliers2 = get_top_clusters(5, dihedral_angle2, projection_cluster_assignment2)
projection1_clean = projection1[np.argwhere(projection_cluster_assignment1 != 0).flatten()]
projection2_clean = projection2[np.argwhere(projection_cluster_assignment2 != 0).flatten()]
print("Calculating NIP scores...")
h1_clean, h2_clean = create_histogram(projection1_clean, projection2_clean)
NIP_clean = calc_NIP(h1_clean, h2_clean)
NIP_ttl = calc_NIP(h1, h2)
print("Plotting...")
MakeFigure(clusters1, res_name, u1_num_frame, NIP_ttl[0], NIP_clean[0], "s1_rama.png", dir_name)
MakeFigure(clusters2, res_name, u2_num_frame, NIP_ttl[1], NIP_clean[1], "s2_rama.png", dir_name)
plt.close('all')
if time_procedure:
end = time.perf_counter()
print("=" * 80)
print(f"This analysis procedure finishes in {end - start:0.4f} seconds")
if debug:
covar = pca.get_covariance()
np.savetxt("covar.dat", covar, fmt="%10.6f")
np.savetxt("dihedral.trr", trig_trr_ttl, fmt="%10.6f")
np.savetxt("projection.txt", projection, fmt="%10.6f")
np.savetxt("s1_projection.txt", projection1, fmt="%10.6f")
np.savetxt("s2_projection.txt", projection2, fmt="%10.6f")
ObsData = ForwardModel(tobs, theta, state0)
xobs = [x[0] for x in ObsData]
yobs = [y[1] for y in ObsData]
cov = np.diag([sig2] * T)
x_noise = np.random.multivariate_normal([0.0] * T, cov)
y_noise = np.random.multivariate_normal([0.0] * T, cov)
x_data = xobs + x_noise
y_data = yobs + y_noise
WriteData(hdf5file, 'data/xobs', x_data)
WriteData(hdf5file, 'data/yobs', y_data)
time = np.linspace(t0, tf, 1000)
TrueData = ForwardModel(time, theta, state0)
xtrue = [x[0] for x in TrueData]
ytrue = [y[1] for y in TrueData]
fig = MakeFigure(425, 0.9)
ax = plt.gca()
ax.plot(xtrue, ytrue, color='#111111')
ax.plot(x_data,
y_data,
'o',
markerfacecolor='#000cff',
markeredgecolor='#000cff',
markersize=8)
ax.set_xlabel('Longitude', fontsize=30, color='#969696')
ax.set_ylabel('Latitude', fontsize=30, color='#969696')
plt.show()