File size: 3,330 Bytes
31d1d63 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 | import sys
from pathlib import Path
import numpy as np
sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
from scoobdoob.validation import msm_build as mb
from scoobdoob.validation import tpt, bridge
def _toy_dtrajs():
rng = np.random.default_rng(0)
P = np.array(
[
[0.80, 0.20, 0.00, 0.00, 0.00],
[0.15, 0.70, 0.15, 0.00, 0.00],
[0.00, 0.15, 0.70, 0.15, 0.00],
[0.00, 0.00, 0.15, 0.70, 0.15],
[0.00, 0.00, 0.00, 0.20, 0.80],
]
)
traj = [0]
for _ in range(40000):
traj.append(int(rng.choice(5, p=P[traj[-1]])))
return [np.array(traj)], P
def test_connected_and_reversible():
dtrajs, _ = _toy_dtrajs()
model, syms, sym2a, _ = mb.build_msm(dtrajs, lag=1, k=5, reversible=True)
assert model.n_states == 5
assert abs(model.stationary_distribution.sum() - 1) < 1e-10
pi = model.stationary_distribution
P = model.transition_matrix
assert np.allclose(pi[:, None] * P, (pi[:, None] * P).T, atol=1e-8)
def test_committor_solvers_agree():
dtrajs, _ = _toy_dtrajs()
model, *_ = mb.build_msm(dtrajs, lag=1, k=5, reversible=True)
q_dt = np.asarray(model.committor_forward([0], [4]))
q_ls = tpt.forward_committor_linsolve(model.transition_matrix, [0], [4])
assert np.max(np.abs(q_dt - q_ls)) < 1e-9
assert q_dt[0] == 0.0 and q_dt[4] == 1.0
assert np.all(np.diff(q_dt) >= -1e-9)
def test_committor_half_localizes_in_middle():
dtrajs, _ = _toy_dtrajs()
model, *_ = mb.build_msm(dtrajs, lag=1, k=5, reversible=True)
q = np.asarray(model.committor_forward([0], [4]))
half = tpt.committor_half_states(q, tol=0.25)
assert 2 in half
def test_bridge_conditioning_beats_unconditional():
_, P = _toy_dtrajs()
Ph = bridge.doob_kernels(P, S_end=[4], horizon=20)
cond = bridge.sample_bridge_paths(Ph, S_start=[0], n_paths=200, seed=0)
unc = bridge.sample_unconditional_paths(
P, S_start=[0], horizon=20, n_paths=200, seed=1
)
assert bridge.success_with_ci(cond, [4])["rate"] > 0.9
assert bridge.success_with_ci(unc, [4])["rate"] < 0.5
def test_aib9_macrostate_classification():
from scoobdoob.validation import aib9
state_left = np.array(
[
[0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0],
[1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0],
[1.0, 1.0, 1.0, 1.0, 0.5, 0.0, 0.0, 0.0, 0.0],
[0.0, 0.0, 0.0, 0.0, 0.5, 1.0, 1.0, 1.0, 1.0],
]
)
labels = aib9.classify_microstates(state_left)
assert list(labels) == ["R", "L", "I_N", "I_C"]
def test_aib9_featurize_and_handedness_shapes():
from scoobdoob.validation import aib9
ang = [np.linspace(-3, 3, 18).reshape(1, 18).repeat(50, 0)]
feat = aib9.featurize_sincos(ang)
hand = aib9.handedness_indicator(ang)
assert feat[0].shape == (50, 36)
assert hand[0].shape == (50, 9)
assert set(np.unique(hand[0])) <= {0.0, 1.0}
if __name__ == "__main__":
test_connected_and_reversible()
test_committor_solvers_agree()
test_committor_half_localizes_in_middle()
test_bridge_conditioning_beats_unconditional()
test_aib9_macrostate_classification()
test_aib9_featurize_and_handedness_shapes()
print("all validation smoke tests passed")
|