def setup(self):
self.cv = paths.FunctionCV("Id", lambda snap: snap.xyz[0][0])
cv_neg = paths.FunctionCV("Neg", lambda snap: -snap.xyz[0][0])
self.stateA = paths.CVDefinedVolume(self.cv, -1.0, 0.0)
self.stateB = paths.CVDefinedVolume(self.cv, 1.0, 2.0)
self.stateC = paths.CVDefinedVolume(self.cv, 3.0, 4.0)
interfacesAB = paths.VolumeInterfaceSet(
self.cv, -1.0, [0.0, 0.2, 0.4]
)
interfacesBC = paths.VolumeInterfaceSet(
self.cv, 1.0, [2.0, 2.2, 2.4]
)
interfacesBA = paths.VolumeInterfaceSet(
cv_neg, -1.0, [-1.0, -0.8, -0.6]
)
network = paths.MISTISNetwork([
(self.stateA, interfacesAB, self.stateB),
(self.stateB, interfacesBC, self.stateC),
(self.stateB, interfacesBA, self.stateA)
])
self.tisAB = network.input_transitions[(self.stateA, self.stateB)]
self.tisBC = network.input_transitions[(self.stateB, self.stateC)]
self.tisBA = network.input_transitions[(self.stateB, self.stateA)]
self.network = network
self.snapA = make_1d_traj([-0.5])[0]
self.noforbid_noextra_AB = paths.FullBootstrapping(
transition=self.tisAB,
snapshot=self.snapA
)
def setup(self):
paths.InterfaceSet._reset()
self.cv = paths.FunctionCV("x", lambda x: x.xyz[0][0])
self.state_A = paths.CVDefinedVolume(self.cv, float("-inf"), 0.0)
self.state_B = paths.CVDefinedVolume(self.cv, 1.0, float("inf"))
pes = paths.engines.toy.LinearSlope([0, 0, 0], 0)
integ = paths.engines.toy.LangevinBAOABIntegrator(0.01, 0.1, 2.5)
topology = paths.engines.toy.Topology(n_spatial=3,
masses=[1.0],
pes=pes)
self.engine = paths.engines.toy.Engine(options={'integ': integ},
topology=topology)
interfaces = paths.VolumeInterfaceSet(self.cv, float("-inf"),
[0.0, 0.1, 0.2])
network = paths.MISTISNetwork([(self.state_A, interfaces, self.state_B)
])
init_traj = make_1d_traj([-0.1, 0.2, 0.5, 0.8, 1.1])
scheme = paths.MoveScheme(network)
scheme.append([
paths.strategies.OneWayShootingStrategy(
selector=paths.UniformSelector(), engine=self.engine),
paths.strategies.PathReversalStrategy(),
paths.strategies.OrganizeByMoveGroupStrategy()
])
init_cond = scheme.initial_conditions_from_trajectories(init_traj)
self.sim = PathSampling(storage=None,
move_scheme=scheme,
sample_set=init_cond)
def make_mistis_network():
cvX = paths.FunctionCV(name="cvX", f=xval)
cvY = paths.FunctionCV(name="cvY", f=yval)
cvXprime = paths.FunctionCV(name="cvXprime", f=xprime)
x_under_min = paths.CVDefinedVolume(cvX, float("-inf"), -0.35)
x_over_max = paths.CVDefinedVolume(cvX, 0.35, float("inf"))
y_under_min = paths.CVDefinedVolume(cvY, float("-inf"), -0.35)
y_over_max = paths.CVDefinedVolume(cvY, 0.35, float("inf"))
stateA = (x_under_min & y_under_min).named("A")
stateB = (x_over_max & y_under_min).named("B")
stateC = (x_under_min & y_over_max).named("C")
interfacesAB = paths.VolumeInterfaceSet(
cvX, float("-inf"), [-0.35, -0.3, -0.27, -0.24, -0.2, -0.1]
)
interfacesAC = paths.VolumeInterfaceSet(
cvY, float("-inf"), [-0.35, -0.3, -0.27, -0.24, -0.2, -0.1, 0.0]
)
interfacesBA = paths.VolumeInterfaceSet(
cvXprime, float("-inf"), [-0.35, -0.3, -0.27, -0.24, -0.2, -0.1]
)
ms_outer = paths.MSOuterTISInterface.from_lambdas(
{iface: 0.0 for iface in [interfacesAB, interfacesBA]}
)
network = paths.MISTISNetwork(
[(stateA, interfacesAB, stateB),
(stateA, interfacesAC, stateC),
(stateB, interfacesBA, stateA)],
ms_outers=ms_outer,
strict_sampling=True
).named("mistis")
return network
def test_fail_without_lambdas(self):
fake_ifaceA = paths.InterfaceSet(cv=self.ifacesA.cv,
volumes=self.ifacesA.volumes,
direction=self.ifacesA.direction)
network = paths.MISTISNetwork([(self.stateA, fake_ifaceA, self.stateB)
])
network.sampling_transitions[0].tcp = self.tcp_A
bias = paths.SRTISBiasFromNetwork(network)
def setup(self):
# set up the trajectories, ensembles, etc. for this test
paths.InterfaceSet._reset()
cv_A = paths.FunctionCV('Id', lambda s: s.xyz[0][0])
cv_B = paths.FunctionCV('1-Id', lambda s: 1.0-s.xyz[0][0])
self.cv_x = cv_A
self.state_A = paths.CVDefinedVolume(cv_A,
float("-inf"), 0.0).named("A")
self.state_B = paths.CVDefinedVolume(cv_B,
float("-inf"), 0.0).named("B")
interfaces_AB = paths.VolumeInterfaceSet(cv_A, float("-inf"),
[0.0, 0.1, 0.2])
interfaces_BA = paths.VolumeInterfaceSet(cv_B, float("-inf"),
[0.0, 0.1, 0.2])
# trajectory that crosses each interface, one state-to-state
self.trajs_AB = [make_tis_traj_fixed_steps(i) for i in [0, 1, 2]]
self.trajs_AB += [make_1d_traj([(-0.5 + i) * 0.1
for i in range(12)])]
self.trajs_BA = [make_tis_traj_fixed_steps(i, reverse=True)
for i in [0, 1, 2]]
self.trajs_BA += [make_1d_traj([1.0 - (-0.5 + i) * 0.1
for i in range(12)])]
# set up mistis
self.mistis = paths.MISTISNetwork([
(self.state_A, interfaces_AB, self.state_B),
(self.state_B, interfaces_BA, self.state_A)
])
mover_stub_mistis = MoverWithSignature(self.mistis.all_ensembles,
self.mistis.all_ensembles)
mistis_ssets = self._make_fake_sampling_sets(self.mistis)
self.mistis_steps = self._make_fake_steps(mistis_ssets,
mover_stub_mistis)
self.mistis_weighted_trajectories = steps_to_weighted_trajectories(
self.mistis_steps,
self.mistis.sampling_ensembles
)
# TODO: set up mstis
self.mstis = paths.MSTISNetwork([
(self.state_A, interfaces_AB),
(self.state_B, interfaces_BA)
])
mover_stub_mstis = MoverWithSignature(self.mstis.all_ensembles,
self.mstis.all_ensembles)
mstis_ssets = self._make_fake_sampling_sets(self.mstis)
self.mstis_steps = self._make_fake_steps(mstis_ssets,
mover_stub_mstis)
self.mstis_weighted_trajectories = steps_to_weighted_trajectories(
self.mstis_steps,
self.mstis.sampling_ensembles
)
def test_bias_from_ms_network(self):
ms_outer = paths.MSOuterTISInterface.from_lambdas({
self.ifacesA: -0.1,
self.ifacesB: 0.1
})
network = paths.MISTISNetwork(
[(self.stateA, self.ifacesA, self.stateB),
(self.stateB, self.ifacesB, self.stateA)],
ms_outers=[ms_outer])
transition_AB = None
transition_BA = None
for t in network.sampling_transitions:
if t.stateA == self.stateA:
t.tcp = self.tcp_A
transition_AB = t
elif t.stateA == self.stateB:
t.tcp = self.tcp_B
transition_BA = t
else:
print([t.stateA, t.stateB])
print([self.stateA, self.stateB])
raise RuntimeError("Weird states in test transition")
bias = paths.SRTISBiasFromNetwork(network)
n_ensembles = len(bias.dataframe.index)
for i in range(n_ensembles):
for j in range(i, n_ensembles):
if not np.isnan(bias.dataframe.loc[i, j]):
np.testing.assert_almost_equal(
bias.dataframe.loc[i, j],
old_div(1.0, bias.dataframe.loc[j, i]))
for i in range(len(transition_AB.ensembles) - 1):
ens_to = transition_AB.ensembles[i]
ens_from = transition_AB.ensembles[i + 1]
assert_almost_equal(bias.bias_value(ens_from, ens_to), 0.5)
for i in range(len(transition_BA.ensembles) - 1):
ens_to = transition_BA.ensembles[i]
ens_from = transition_BA.ensembles[i + 1]
assert_almost_equal(bias.bias_value(ens_from, ens_to), 0.2)
for ensA in transition_AB.ensembles:
for ensB in transition_BA.ensembles:
assert_equal(np.isnan(bias.bias_value(ensA, ensB)), True)
assert_equal(np.isnan(bias.bias_value(ensB, ensA)), True)
assert_almost_equal(
bias.bias_value(transition_BA.ensembles[-1], network.ms_outers[0]),
old_div(5.0, 2))
assert_almost_equal(
bias.bias_value(transition_AB.ensembles[-1], network.ms_outers[0]),
old_div(2.0, 2))
raise SkipTest
def setup(self):
self.HAS_TQDM = paths.progress.HAS_TQDM
paths.progress.HAS_TQDM = False
paths.InterfaceSet._reset()
cvA = paths.FunctionCV(name="xA", f=lambda s: s.xyz[0][0])
cvB = paths.FunctionCV(name="xB", f=lambda s: -s.xyz[0][0])
state_A = paths.CVDefinedVolume(cvA, float("-inf"), -0.5).named("A")
state_B = paths.CVDefinedVolume(cvB, float("-inf"), -0.5).named("B")
interfaces_A = paths.VolumeInterfaceSet(cvA, float("-inf"),
[-0.5, -0.3])
network = paths.MISTISNetwork([(state_A, interfaces_A, state_B)])
self.scheme = MoveScheme(network)
self.scheme.append(OneWayShootingStrategy())
self.scheme.append(NearestNeighborRepExStrategy())
self.scheme.append(OrganizeByMoveGroupStrategy())
root_mover = self.scheme.move_decision_tree()
path_sim_mover = paths.PathSimulatorMover(root_mover, None)
null_mover = paths.IdentityPathMover(counts_as_trial=False)
ens_0 = network.sampling_ensembles[0]
ens_1 = network.sampling_ensembles[1]
# acc repex ens1-2
# acc fwd ens1
# acc bkwd ens2
# rej bkwd ens1
# rej repex ens1-2
step_info = [(1, True, path_sim_mover, 'repex', [ens_0, ens_1], None),
(2, True, path_sim_mover, 'shooting', [ens_0], 0),
(3, True, path_sim_mover, 'shooting', [ens_1], 1),
(4, False, path_sim_mover, 'shooting', [ens_0], 1),
(5, False, path_sim_mover, 'repex', [ens_0, ens_1], None)]
self.steps = [_make_acceptance_mock_step(*info) for info in step_info]
self.null_mover_6 = _make_null_mover_step(6, path_sim_mover,
null_mover)
self.null_mover_change_key = [(None, str([path_sim_mover, [None]]))]
acceptance_empty = MoveAcceptanceAnalysis(self.scheme)
acceptance = MoveAcceptanceAnalysis(self.scheme)
acceptance.add_steps(self.steps)
acceptance_null = MoveAcceptanceAnalysis(self.scheme)
acceptance_null.add_steps(self.steps + [self.null_mover_6])
self.analysis = {
'empty': acceptance_empty,
'normal': acceptance,
'with_null': acceptance_null
}
def unidirectional_tis_network():
r"""Fixture for unidirectional TIS with the default (RETIS) scheme.
This has states defined as initial state :math:`x < 0` and final state
:math:`x \ge 10`. The interfaces are at :math:`x=0`, :math:`x=3`, and
:math:`x=6`.
"""
paths.InterfaceSet._reset()
state_A = paths.CVDefinedVolume(DEFAULT_CV, float("-inf"), 0)
state_B = paths.CVDefinedVolume(DEFAULT_CV, 10, float("inf"))
interfaces = paths.VolumeInterfaceSet(DEFAULT_CV, float("-inf"), [0, 3, 6])
network = paths.MISTISNetwork([(state_A, interfaces, state_B)])
return network
def test_bad_network(self):
# raises error if more than one transition shares a minus ensemble
# (flux cannot be calculated with multiple interface set minus move)
state_C = paths.CVDefinedVolume(self.cv_x, 0.5, 0.7)
trans_AB = self.mistis.transitions[(self.state_A, self.state_B)]
trans_BA = self.mistis.transitions[(self.state_B, self.state_A)]
interfaces_AB = trans_AB.interfaces
interfaces_BA = trans_BA.interfaces
interfaces_AC = trans_AB.interfaces
bad_mistis = paths.MISTISNetwork([
(self.state_A, interfaces_AB, self.state_B),
(self.state_B, interfaces_BA, self.state_A),
(self.state_A, interfaces_AC, state_C)
])
scheme = paths.DefaultScheme(bad_mistis)
scheme.build_move_decision_tree()
minus_flux = MinusMoveFlux(scheme)
def setup(self):
paths.InterfaceSet._reset()
self.cv_inc = paths.FunctionCV(name="inc", f=lambda s: s.xyz[0][0])
self.cv_dec = paths.FunctionCV(name="dec",
f=lambda s: 1.0 - s.xyz[0][0])
self.lambdas = [0.0, 0.1, 0.2, 0.3]
self.interfaces_inc = paths.VolumeInterfaceSet(cv=self.cv_inc,
minvals=float("-inf"),
maxvals=self.lambdas)
self.interfaces_dec = paths.VolumeInterfaceSet(cv=self.cv_dec,
minvals=float("-inf"),
maxvals=self.lambdas)
self.stateA = paths.CVDefinedVolume(self.cv_inc, float("-inf"),
0.0).named("A")
self.stateB = paths.CVDefinedVolume(self.cv_dec, float("-inf"),
0.0).named("B")
self.network = paths.MISTISNetwork([
(self.stateA, self.interfaces_inc, self.stateB),
(self.stateB, self.interfaces_dec, self.stateA)
])
self.volumes = [
self.interfaces_inc.new_interface(0.5),
self.interfaces_dec.new_interface(0.4)
]
self.ms_outer_explicit = paths.MSOuterTISInterface(
interface_sets=[self.interfaces_inc, self.interfaces_dec],
volumes=self.volumes,
lambdas=[0.5, 0.4])
self.ms_outer = paths.MSOuterTISInterface.from_lambdas({
self.interfaces_inc:
0.5,
self.interfaces_dec:
0.4
})
# TODO: temporary hack until networks working; remove after
self.post_network = paths.MSOuterTISInterface.from_lambdas({
t.interfaces: {
self.cv_inc: 0.5,
self.cv_dec: 0.4
}[t.interfaces.cv]
for t in self.network.sampling_transitions
})
def setup(self):
# create the network
paths.InterfaceSet._reset()
xval = paths.FunctionCV(name="xA", f=lambda s: s.xyz[0][0])
self.stateA = paths.CVDefinedVolume(xval, -1.0, -0.5).named("A")
self.stateB = paths.CVDefinedVolume(xval, 0.5, float("inf")).named("B")
ifacesA = paths.VolumeInterfaceSet(xval, float(-1.0),
[-0.5, -0.4, -0.3, -0.2])
self.network = paths.MISTISNetwork([(self.stateA, ifacesA, self.stateB)
])
transition = self.network.transitions[(self.stateA, self.stateB)]
ensembles = transition.ensembles
self.xval = xval
self.ifacesA = ifacesA
# create the biases
bias_table = {}
bias_table[ensembles[0]] = 1.0
bias_table[ensembles[1]] = 0.5
bias_table[ensembles[2]] = 0.2
self.bias = BiasEnsembleTable.ratios_from_dictionary(bias_table)
# samples, moves, changes
traj = make_1d_traj(
[-0.55, -0.45, -0.35, -0.25, -0.15, -0.26, -0.36, -0.46, -0.56])
s0 = paths.Sample(replica=0, ensemble=ensembles[0], trajectory=traj)
s1 = paths.Sample(replica=1, ensemble=ensembles[1], trajectory=traj)
s2 = paths.Sample(replica=2, ensemble=ensembles[2], trajectory=traj)
self.sample_set = paths.SampleSet([s0, s1, s2])
self.sample_set.sanity_check()
move_01 = paths.EnsembleHopMover(ensembles[0], ensembles[1])
move_02 = paths.EnsembleHopMover(ensembles[0], ensembles[2])
move_12 = paths.EnsembleHopMover(ensembles[1], ensembles[2])
move_21 = paths.EnsembleHopMover(ensembles[2], ensembles[1])
move_20 = paths.EnsembleHopMover(ensembles[2], ensembles[0])
move_10 = paths.EnsembleHopMover(ensembles[1], ensembles[0])
# NOTE: all changes here are accepted
self.change_01 = move_01.move(self.sample_set)
self.change_02 = move_02.move(self.sample_set)
self.change_12 = move_12.move(self.sample_set)
self.change_21 = move_21.move(self.sample_set)
self.change_20 = move_20.move(self.sample_set)
self.change_10 = move_10.move(self.sample_set)
# convenience lists for changes going outward vs. inward
self.out_changes = [self.change_01, self.change_02, self.change_12]
self.in_changes = [self.change_10, self.change_20, self.change_21]
def test_bias_from_network(self):
# force the TCP in
network = paths.MISTISNetwork([(self.stateA, self.ifacesA, self.stateB)
])
network.sampling_transitions[0].tcp = self.tcp_A
bias = paths.SRTISBiasFromNetwork(network)
transition = list(network.transitions.values())[0] # only one
# check reciprocal of symmetric partners
for i in range(4):
for j in range(i, 4):
assert_equal(bias.dataframe.loc[i, j],
old_div(1.0, bias.dataframe.loc[j, i]))
for i in range(len(transition.ensembles) - 1):
ens_to = transition.ensembles[i]
ens_from = transition.ensembles[i + 1]
assert_equal(bias.bias_value(ens_from, ens_to), 0.5)
for i in range(len(transition.ensembles) - 2):
ens_to = transition.ensembles[i]
ens_from = transition.ensembles[i + 2]
assert_equal(bias.bias_value(ens_from, ens_to), 0.25)
def test_fail_without_tcp(self):
network = paths.MISTISNetwork([(self.stateA, self.ifacesA, self.stateB)
])
bias = paths.SRTISBiasFromNetwork(network)
def test_add_biases(self):
# this is where we combine multiple biases into one
ifacesA = self.ifacesA[:-1]
xval2 = paths.FunctionCV(name="xB", f=lambda s: 0.5 - s.xyz[0][0])
ifacesB = paths.VolumeInterfaceSet(xval2, float("-inf"),
[0.0, 0.1, 0.2])
xval3 = paths.FunctionCV(name="xC", f=lambda s: s.xyz[0][0] - 2.0)
stateC = paths.CVDefinedVolume(self.xval, -3.0, 2.0)
ifacesC = paths.VolumeInterfaceSet(xval3, -1.0, [0.0, 0.1, 0.2, 0.3])
network = paths.MISTISNetwork(
[(self.stateA, ifacesA, self.stateB),
(self.stateB, ifacesB, self.stateA),
(stateC, ifacesC, self.stateA)],
ms_outers=paths.MSOuterTISInterface.from_lambdas({
ifacesA: -0.2,
ifacesB: 0.3
}))
ens_A = network.transitions[(self.stateA, self.stateB)].ensembles
ens_B = network.transitions[(self.stateB, self.stateA)].ensembles
ens_C = network.transitions[(stateC, self.stateA)].ensembles
ms_outer = list(network.special_ensembles['ms_outer'].keys())[0]
dict_A = {ens_A[0]: 1.0, ens_A[1]: 0.5, ens_A[2]: 0.2, ms_outer: 0.1}
dict_B = {ens_B[0]: 1.0, ens_B[1]: 0.6, ens_B[2]: 0.3, ms_outer: 0.15}
dict_C = {ens_C[0]: 1.0, ens_C[1]: 0.8, ens_C[2]: 0.2}
bias_A = BiasEnsembleTable.ratios_from_dictionary(dict_A)
bias_B = BiasEnsembleTable.ratios_from_dictionary(dict_B)
bias_C = BiasEnsembleTable.ratios_from_dictionary(dict_C)
bias_AB = bias_A + bias_B
# check the ensembles_to_ids
assert_equal(len(bias_AB.ensembles_to_ids), 7)
for ens in ens_A:
assert_in(bias_AB.ensembles_to_ids[ens], [0, 1, 2])
for ens in ens_B:
assert_in(bias_AB.ensembles_to_ids[ens], [3, 4, 5])
assert_equal(bias_AB.ensembles_to_ids[ms_outer], 6)
# check values
df_A = bias_A.dataframe
df_B = bias_B.dataframe
df_AB = bias_AB.dataframe
col_A_msouter = bias_A.ensembles_to_ids[ms_outer]
col_B_msouter = bias_B.ensembles_to_ids[ms_outer]
col_AB_msouter = bias_AB.ensembles_to_ids[ms_outer]
for ens1 in ens_A:
idx_A = bias_A.ensembles_to_ids[ens1]
idx_AB = bias_AB.ensembles_to_ids[ens1]
for ens2 in ens_A:
col_A = bias_A.ensembles_to_ids[ens2]
col_AB = bias_AB.ensembles_to_ids[ens2]
val_A = df_A.loc[idx_A, col_A]
val_AB = df_AB.loc[idx_AB, col_AB]
assert_equal(val_A, val_AB)
for ens2 in ens_B:
col_AB = bias_AB.ensembles_to_ids[ens2]
assert_equal(np.isnan(df_AB.loc[idx_AB, col_AB]), True)
assert_equal(df_A.loc[idx_A, col_A_msouter],
df_AB.loc[idx_AB, col_AB_msouter])
assert_equal(df_A.loc[col_A_msouter, idx_A],
df_AB.loc[col_AB_msouter, idx_AB])
for ens1 in ens_B:
idx_B = bias_B.ensembles_to_ids[ens1]
idx_AB = bias_AB.ensembles_to_ids[ens1]
for ens2 in ens_B:
col_B = bias_B.ensembles_to_ids[ens2]
col_AB = bias_AB.ensembles_to_ids[ens2]
val_B = df_B.loc[idx_B, col_B]
val_AB = df_AB.loc[idx_AB, col_AB]
assert_equal(val_B, val_AB)
for ens2 in ens_A:
col_AB = bias_AB.ensembles_to_ids[ens2]
assert_equal(np.isnan(df_AB.loc[idx_AB, col_AB]), True)
assert_equal(df_B.loc[idx_B, col_B_msouter],
df_AB.loc[idx_AB, col_AB_msouter])
assert_equal(df_B.loc[col_B_msouter, idx_B],
df_AB.loc[col_AB_msouter, idx_AB])
# just to make sure no errors raise when there are NaNs in table
bias_ABC = bias_A + bias_B + bias_C
def tis_network(cv_and_states):
cv, state_A, state_B = cv_and_states
interfaces = paths.VolumeInterfaceSet(cv, float("-inf"), [0.0, 0.1, 0.2])
network = paths.MISTISNetwork([(state_A, interfaces, state_B)])
return network
min_unbound = 0.90 # nanometers, minimum unbound state separation distance
print('Creating interfaces...')
ninterfaces = 100
bound = paths.CVDefinedVolume(cv, lambda_min=0.0, lambda_max=max_bound)
unbound = paths.CVDefinedVolume(cv,
lambda_min=min_unbound,
lambda_max=float("inf"))
interfaces = paths.VolumeInterfaceSet(cv,
minvals=0.0,
maxvals=np.linspace(
max_bound, min_unbound - 0.01,
ninterfaces))
print('Creating network...')
mistis = paths.MISTISNetwork([(bound, interfaces, unbound)])
initial_trajectory_method = 'bootstrap'
if initial_trajectory_method == 'high-temperature':
# We are starting in the bound state, so
# generate high-temperature trajectory that reaches the unbound state
print('Generating high-temperature trajectory...')
#ensemble = not (paths.ExitsXEnsemble(bound) & paths.EntersXEnsemble(unbound))
unbinding_ensemble = paths.AllOutXEnsemble(unbound)
bridging_ensemble = paths.AllOutXEnsemble(bound) & paths.AllOutXEnsemble(
unbound)
initial_trajectories = list()
minus_trajectories = list()
tmp_network = paths.TPSNetwork(bound, unbound)
attempt = 0
while (len(initial_trajectories) == 0) or (len(minus_trajectories) == 0):
