def plotFluxPathways(self,
statetype='macro',
mode='net_flux',
fraction=1.0):
""" Plot flux pathways between source and sink state.
The flux is in units of transition events per lag time used to construct the Markov model.
Parameters
----------
statetype : {'macro','coarse','micro'}
What type of states to plot
mode : {'net_flux', 'gross_flux'}
Type of fluxes to plot
fraction : float
Fraction of fluxes for which to report pathways. Doesn't change the plot, only the text output.
"""
# Make mode a radio button with interactive plot
from pyemma import msm
from pyemma.plots import plot_flux
from matplotlib import pylab as plt
self._intergrityCheck()
plt.figure()
if statetype == 'micro':
tpt = msm.tpt(self.model.msm, [self.sourcemicro], [self.sinkmicro])
fig, pos = plot_flux(tpt, attribute_to_plot=mode)
elif statetype == 'macro' or statetype == 'coarse':
metastable_sets = []
for i in range(self.model.macronum):
metastable_sets.append(
np.where(self.model.macro_ofmicro == i)[0])
tpt = msm.tpt(self.model.msm, metastable_sets[self.source],
metastable_sets[self.sink])
#from IPython.core.debugger import Tracer
#Tracer()()
newsets, tpt = tpt.coarse_grain(metastable_sets)
setmap = []
# getting the mapping of new sets to old sets
for set1 in newsets:
for idx2, set2 in enumerate(metastable_sets):
if set(set1) == set(set2):
setmap.append(idx2)
continue
setmap = np.array(setmap)
fig, pos = plot_flux(tpt,
attribute_to_plot=mode,
state_labels=setmap)
fig.show()
paths, pathfluxes = tpt.pathways(fraction=fraction)
cumflux = 0
print("Path flux\t\t%path\t%of total\tpath")
for i in range(len(paths)):
cumflux += pathfluxes[i]
print('{}\t{:3.1f}%\t{:3.1f}%\t\t{}'.format(
pathfluxes[i], 100.0 * pathfluxes[i] / tpt.total_flux,
100.0 * cumflux / tpt.total_flux, setmap[paths[i]]))
def fluxPathways(self):
import pyemma.plots as mplt
import pyemma.msm as msm
import matplotlib.pylab as plt
self._intergrityCheck()
plt.ion()
tpt = msm.tpt(self.model.coarsemsm, [self.source], [self.sink])
mplt.plot_flux(tpt[1], show_committor=False)
plt.show()
def plotFluxPathways(self, statetype='macro', mode='net_flux', fraction=1.0):
""" Plot flux pathways between source and sink state.
Parameters
----------
statetype : {'macro','coarse','micro'}
What type of states to plot
mode : {'net_flux', 'gross_flux'}
Type of fluxes to plot
fraction : float
Fraction of fluxes for which to report pathways. Doesn't change the plot, only the text output.
"""
# Make mode a radio button with interactive plot
from pyemma import msm
from pyemma.plots import plot_flux
from matplotlib import pylab as plt
self._intergrityCheck()
plt.figure()
if statetype == 'micro':
tpt = msm.tpt(self.model.msm, [self.sourcemicro], [self.sinkmicro])
fig, pos = plot_flux(tpt, attribute_to_plot=mode)
elif statetype == 'macro' or statetype == 'coarse':
metastable_sets = []
for i in range(self.model.macronum):
metastable_sets.append(np.where(self.model.macro_ofmicro == i)[0])
tpt = msm.tpt(self.model.msm, metastable_sets[self.source], metastable_sets[self.sink])
#from IPython.core.debugger import Tracer
#Tracer()()
newsets, tpt = tpt.coarse_grain(metastable_sets)
setmap = []
# getting the mapping of new sets to old sets
for set1 in newsets:
for idx2, set2 in enumerate(metastable_sets):
if set(set1) == set(set2):
setmap.append(idx2)
continue
setmap = np.array(setmap)
fig, pos = plot_flux(tpt, attribute_to_plot=mode, state_labels=setmap)
fig.show()
paths, pathfluxes = tpt.pathways(fraction=fraction)
cumflux = 0
print("Path flux\t\t%path\t%of total\tpath")
for i in range(len(paths)):
cumflux += pathfluxes[i]
print('{}\t{:3.1f}%\t{:3.1f}%\t\t{}'.format(
pathfluxes[i], 100.0*pathfluxes[i]/tpt.total_flux,
100.0*cumflux/tpt.total_flux, setmap[paths[i]]))
B = [4]
tpt = msm.tpt(M, A, B)
# get tpt gross flux
F = tpt.gross_flux
print('**Flux matrix**:')
print(F)
print('**forward committor**:')
print(tpt.committor)
print('**backward committor**:')
print(tpt.backward_committor)
# we position states along the y-axis according to the commitor
tptpos = np.array([tpt.committor, [0, 0, 0.5, -0.5, 0]]).transpose()
print('\n**Gross flux illustration**: ')
pl = mplt.plot_flux(tpt,
pos=tptpos,
arrow_label_format="%10.4f",
attribute_to_plot='gross_flux')
"""Note that in the gross flux above we had fluxes between states 1 and 3.
These fluxes just represent that some reactive trajectories leave 1, go to 3,
go then back to 1 and then go on to the target state 4 directly or via state 2.
This detour via state 3 is usually of no interest, and their contribution is
already contained in the main paths 0->1 and 1->{2,4}. Therefore we remove all n
onproductive recrossings by taking the difference flux between pairs of states that
have fluxes in both directions. This gives us the net flux."""
# get tpt net flux
Fp = tpt.net_flux
# or: tpt.flux (it's the same!)
print('**Net-Flux matrix**:')
print(Fp)
# visualize
def plotFluxPathways(
self, statetype="macro", mode="net_flux", fraction=1.0, plot=True, save=None
):
"""Plot flux pathways between source and sink state.
The flux is in units of transition events per lag time used to construct the Markov model.
Parameters
----------
statetype : {'macro','coarse','micro'}
What type of states to plot
mode : {'net_flux', 'gross_flux'}
Type of fluxes to plot
fraction : float
Fraction of fluxes for which to report pathways. Doesn't change the plot, only the text output.
plot : bool
If set it False the plot will not show up in a figure
save : str
If a path is passed to save, the plot will be saved to the specified file
"""
# Make mode a radio button with interactive plot
from pyemma import msm
from pyemma.plots import plot_flux
from matplotlib import pylab as plt
self._intergrityCheck()
plt.figure()
if statetype == "micro":
tpt = msm.tpt(self.model.msm, [self.sourcemicro], [self.sinkmicro])
fig, pos = plot_flux(tpt, attribute_to_plot=mode)
elif statetype == "macro" or statetype == "coarse":
metastable_sets = []
for i in range(self.model.macronum):
metastable_sets.append(np.where(self.model.macro_ofmicro == i)[0])
tpt = msm.tpt(
self.model.msm, metastable_sets[self.source], metastable_sets[self.sink]
)
newsets, tpt = tpt.coarse_grain(metastable_sets)
setmap = []
# getting the mapping of new sets to old sets
for set1 in newsets:
for idx2, set2 in enumerate(metastable_sets):
if set(set1) == set(set2):
setmap.append(idx2)
continue
setmap = np.array(setmap)
fig, pos = plot_flux(tpt, attribute_to_plot=mode, state_labels=setmap)
if save is not None:
fig.savefig(save, dpi=300, bbox_inches="tight", pad_inches=0.2)
if plot:
fig.show()
paths, pathfluxes = tpt.pathways(fraction=fraction)
cumflux = 0
print("Path flux\t\t%path\t%of total\tpath")
for i in range(len(paths)):
cumflux += pathfluxes[i]
print(
"{}\t{:3.1f}%\t{:3.1f}%\t\t{}".format(
pathfluxes[i],
100.0 * pathfluxes[i] / tpt.total_flux,
100.0 * cumflux / tpt.total_flux,
setmap[paths[i]],
)
)
def plotTPT(TPT_object, state_labels='auto', outfile=None):
""" Plot the flux network from a Reactive Flux object"""
flux_figure = mplt.plot_flux(TPT_object, state_labels=state_labels)
if not outfile is None:
flux_figure.savefig(outfile)
return flux_figure
