reader = coor.source(trajnames, features=feat)
# Estimate Markov state model
#tica_lag = 20
#keep_dims = 23
#keep_dims = 23
if not noplots:
print("Plotting tica timescales vs lagtime...")
plot_tica_stuff()
#tica_lag = 50 # lagtime where TICA timescales are converged
#keep_dims = 5 # num dims where cumulative variance reaches ~0.8
tica = coor.tica(lag=lagtime, stride=1)
coor.pipeline([reader, tica])
Y = tica.get_output(dimensions=range(keep_dims))
#np.save(msm_savedir + "/tica_ti.npy", tica.timescales)
#print("Saving tica coordinates...")
##if not os.path.exists(msm_savedir + "/run_1_TIC_1.npy"):
#for i in range(keep_dims):
# for n in range(len(Y)):
# # save TIC with indices of corresponding traj
# idx1, idx2 = traj_idxs[n]
# tic_saveas = msm_savedir + "/run_{}_{}_TIC_{}.npy".format(idx1, idx2, i+1)
# if not os.path.exists(tic_saveas) or resave_tic:
# np.save(tic_saveas, Y[n][:,i])
raise SystemExit
def plot_tica_stuff():
# calculate TICA at different lagtimes
#tica_lags = np.array(range(1, 11) + [12, 15, 20, 25, 50, 75, 100, 150, 200])
tica_lags = np.array([1, 5, 10, 25, 50, 100, 200, 500, 1000])
all_cumvar = []
all_tica_ti = []
for i in range(len(tica_lags)):
tica = coor.tica(lag=tica_lags[i], stride=1)
coor.pipeline([reader, tica])
all_cumvar.append(tica.cumvar)
all_tica_ti.append(tica.timescales)
all_cumvar = np.array(all_cumvar)
all_tica_ti = np.array(all_tica_ti)
# times vs lag
plt.figure()
for i in range(20):
plt.plot(tica_lags, all_tica_ti[:, i])
plt.fill_between(tica_lags, tica_lags, color='gray', lw=2)
#ymin, ymax = plt.ylim()
#plt.ylim(ymin, ymax)
plt.grid(True, alpha=1, color='k', ls='--')
plt.xlabel(r"Lag time $\tau$")
plt.ylabel(r"TICA $t_i(\tau)$")
plt.title(f_str)
plt.savefig(msm_savedir + "/tica_its_vs_lag.pdf")
plt.savefig(msm_savedir + "/tica_its_vs_lag.png")
# cumulative variance
plt.figure()
for i in range(len(tica_lags)):
plt.plot(np.arange(1,
len(all_cumvar[i]) + 1),
all_cumvar[i],
label=str(tica_lags[i]))
plt.legend(loc=4)
plt.grid(True, alpha=1, color='k', ls='--')
#ymin, ymax = plt.ylim()
plt.ylim(0, 1)
plt.xlabel("Index")
plt.ylabel("Kinetic Variance")
plt.title(f_str)
plt.savefig(msm_savedir + "/tica_cumvar.pdf")
plt.savefig(msm_savedir + "/tica_cumvar.png")
# times vs index
plt.figure()
for i in range(len(tica_lags)):
plt.plot(all_tica_ti[i, :20], 'o', label=str(tica_lags[i]))
plt.legend()
plt.grid(True, alpha=1, color='k', ls='--')
#ymin, ymax = plt.ylim()
#plt.ylim(ymin, ymax)
plt.xlabel("Index")
plt.ylabel(r"TICA $t_i$")
plt.title(f_str)
plt.savefig(msm_savedir + "/tica_its.pdf")
plt.savefig(msm_savedir + "/tica_its.png")
print("There are %d frames total in %d trajectories." % (coordinates_source.n_frames_total(), coordinates_source.number_of_trajectories()))
################################################################################
# Do tICA
################################################################################
print('tICA...')
running_tica = coor.tica(lag=100, dim=100)
################################################################################
# Cluster
################################################################################
print('Clustering...')
clustering = coor.cluster_kmeans(k=100, stride=50)
coor.pipeline([coordinates_source,running_tica,clustering])
dtrajs = clustering.dtrajs
# Save discrete trajectories.
clustering.save_dtrajs(output_format='npy', extension='.npy')
################################################################################
# Make tics plot
################################################################################
tics = running_tica.get_output()[0]
z,x,y = np.histogram2d(tics[:,0],tics[:,1], bins=50)
F = -np.log(z+1)
extent = [x[0], x[-1], y[0], y[-1]]
# Define coordinates source
################################################################################
trajectory_files = glob(os.path.join(source_directory, '*0.h5'))
coordinates_source = coor.source(trajectory_files, featurizer)
print("There are %d frames total in %d trajectories." %
(coordinates_source.n_frames_total(),
coordinates_source.number_of_trajectories()))
################################################################################
# Do tICA
################################################################################
print('tICA...')
running_tica = coor.tica(lag=1600, dim=100)
coor.pipeline([coordinates_source, running_tica])
################################################################################
# Make eigenvalues plot
################################################################################
plt.clf()
eigenvalues = (running_tica.eigenvalues)**2
sum_eigenvalues = np.sum(eigenvalues[0:2])
print "This is the sum of the first two eigenvalues: %s." % sum_eigenvalues
plt.plot(eigenvalues)
plt.xlim(0, 4)
plt.ylim(0, 1.2)
coordinates_source.number_of_trajectories()))
################################################################################
# Do tICA
################################################################################
print('tICA...')
running_tica = coor.tica(lag=100, dim=100)
################################################################################
# Cluster
################################################################################
print('Clustering...')
clustering = coor.cluster_kmeans(k=100, stride=50)
coor.pipeline([coordinates_source, running_tica, clustering])
dtrajs = clustering.dtrajs
# Save discrete trajectories.
clustering.save_dtrajs(output_format='npy', extension='.npy')
################################################################################
# Make tics plot
################################################################################
tics = running_tica.get_output()[0]
z, x, y = np.histogram2d(tics[:, 0], tics[:, 1], bins=50)
F = -np.log(z + 1)
extent = [x[0], x[-1], y[0], y[-1]]
################################################################################
# Define coordinates source
################################################################################
trajectory_files = glob(os.path.join(source_directory, '*0.h5'))
coordinates_source = coor.source(trajectory_files,featurizer)
print("There are %d frames total in %d trajectories." % (coordinates_source.n_frames_total(), coordinates_source.number_of_trajectories()))
################################################################################
# Do tICA
################################################################################
print('tICA...')
running_tica = coor.tica(lag=1600, dim=100)
coor.pipeline([coordinates_source,running_tica])
################################################################################
# Make eigenvalues plot
################################################################################
plt.clf()
eigenvalues = (running_tica.eigenvalues)**2
sum_eigenvalues = np.sum(eigenvalues[0:2])
print "This is the sum of the first two eigenvalues: %s." % sum_eigenvalues
plt.plot(eigenvalues)
plt.xlim(0,4)
plt.ylim(0,1.2)
feat.add_distances(pair_idxs)
if "invdists" in feature_set:
feat.add_inverse_distances(pair_idxs)
if "rho" in feature_set:
feat.add_custom_feature(
CustomFeature(local_density_feature,
len(all_pair_idxs),
fun_args=(all_pair_idxs, r0, widths)))
if "rg" in feature_set:
feat.add_custom_feature(CustomFeature(rg_feature, 1))
reader = coor.source(trajnames, features=feat)
transform = coor.tica(lag=lagtime, stride=stride)
cluster_pipe = [reader, transform, cluster]
pipeline = coor.pipeline(cluster_pipe)
dtrajs = cluster.dtrajs
#lags = [1,2,5,10,20,50,100,200,300,400,500,600,700,800,900,1000]
lags = [10, 25, 50, 100, 200, 500, 1000]
its = msm.its(dtrajs, lags=lags)
T = 300
save_markov_state_models(T, its.models)
# save name should have n_clusters
saveas = "msm_its_vs_lag_{}".format(n_clusters)
mplt.plot_implied_timescales(its, ylog=False)
#plt.title("T = " + str(T))
plt.savefig(msm_savedir + "/" + saveas + ".pdf")
plt.savefig(msm_savedir + "/" + saveas + ".png")
