def test_normal_gaussian_nd(mode, factor, **kwargs):
ndim = 3
kwargs["nsteps"] = 8000
# Isotropic.
_test_normal(
moves.GaussianMove(0.5, factor=factor, mode=mode), ndim=ndim, **kwargs
)
# Axis-aligned.
_test_normal(
moves.GaussianMove(0.5 * np.ones(ndim), factor=factor, mode=mode),
ndim=ndim,
**kwargs,
)
with pytest.raises(ValueError):
_test_normal(
moves.GaussianMove(
0.5 * np.ones(ndim - 1), factor=factor, mode=mode
),
ndim=ndim,
**kwargs,
)
# Full matrix.
if mode == "vector":
_test_normal(
moves.GaussianMove(
np.diag(0.5 * np.ones(ndim)), factor=factor, mode=mode
),
ndim=ndim,
**kwargs,
)
with pytest.raises(ValueError):
_test_normal(
moves.GaussianMove(np.diag(0.5 * np.ones(ndim - 1))),
ndim=ndim,
**kwargs,
)
else:
with pytest.raises(ValueError):
_test_normal(
moves.GaussianMove(
np.diag(0.5 * np.ones(ndim)), factor=factor, mode=mode
),
ndim=ndim,
**kwargs,
)
__all__ = ["test_shapes", "test_errors", "test_thin", "test_vectorize"]
all_backends = backends.get_test_backends()
def normal_log_prob(params):
return -0.5 * np.sum(params**2)
@pytest.mark.parametrize(
"backend, moves",
product(
all_backends,
[
None,
moves.GaussianMove(0.5),
[moves.StretchMove(), moves.GaussianMove(0.5)],
[(moves.StretchMove(), 0.3), (moves.GaussianMove(0.5), 0.1)],
],
),
)
def test_shapes(backend, moves, nwalkers=32, ndim=3, nsteps=10, seed=1234):
# Set up the random number generator.
np.random.seed(seed)
with backend() as be:
# Initialize the ensemble, moves and sampler.
coords = np.random.randn(nwalkers, ndim)
sampler = EnsembleSampler(nwalkers,
ndim,
normal_log_prob,
def test_uniform_gaussian(mode, factor, **kwargs):
_test_uniform(moves.GaussianMove(0.5, factor=factor, mode=mode), **kwargs)
def test_normal_gaussian(mode, factor, **kwargs):
_test_normal(moves.GaussianMove(0.5, mode=mode, factor=factor), **kwargs)
