def get_stiminfo(stim_path):
stim = Stimulus2D.load(stim_path)
assert len(stim.operations) <= 2
if len(stim.operations) == 2:
assert stim.operations[0][0] == "flip"
assert stim.operations[1][0] == "rotate"
flipped = True
theta = stim.operations[1][1]
elif len(stim.operations) == 1:
if stim.operations[0][0] == "rotate":
flipped = False
theta = stim.operations[0][1]
elif stim.operations[0][0] == "flip":
flipped = True
theta = 0
elif len(stim.operations) == 0:
flipped = False
theta = 0
info = {
'v0': stim._v.tolist(),
'v1': stim.vertices.tolist(),
'flipped': flipped,
'theta': theta,
'stimulus': stim_path.namebase.split("_")[0],
}
return info
def simulate(task):
stim_path = task["stim_path"]
model_name = task["model"]
seed = task["seed"]
data_path = path.Path(task["data_path"])
model_opts = task["model_opts"]
X = Stimulus2D.load(stim_path)
Xa = X.copy_from_initial().vertices
Xb = X.copy_from_vertices().vertices
try:
model_class = getattr(m, model_name)
except AttributeError:
raise ValueError("unhandled model: %s" % model_name)
np.random.seed(seed)
if not data_path.exists():
data_path.makedirs()
for iopt, opts in model_opts.iteritems():
logger.info("Task '%s', part %s", task['task_name'], iopt)
dest = data_path.joinpath("part_%s" % iopt)
model = model_class(Xa, Xb, **opts)
try:
model.sample()
except SystemExit:
raise
except:
print traceback.format_exc()
if dest.exists():
dest.rmtree_p()
model.save(dest)
def pytest_generate_tests(metafunc):
if hasattr(metafunc.function, "full"):
theta = sorted(range(0, 360, 45) + [39])
flipped = [True, False]
elif hasattr(metafunc.function, "once"):
theta = [0]
flipped = [False]
elif hasattr(metafunc.function, "smallrot"):
theta = [20, 340]
flipped = [True, False]
else:
theta = [39]
flipped = [True, False]
all_argnames = ['theta', 'flipped', 'Xa', 'Xb', 'model']
argnames = [x for x in metafunc.fixturenames if x in all_argnames]
if len(argnames) == 0:
return
argvalues = []
for t, f in product(theta, flipped):
# create stimulus
X = Stimulus2D.random(8)
if f:
X.flip([0, 1])
X.rotate(t)
Xa = X.copy_from_initial()
Xb = X.copy_from_vertices()
# create model
model = metafunc.cls.cls(
Xa.vertices, Xb.vertices,
S_sigma=S_sigma,
step=step,
prior=prior)
args = dict(theta=t, flipped=f, Xa=Xa, Xb=Xb, model=model)
argvalues.append([args[x] for x in argnames])
metafunc.parametrize(argnames, argvalues)
def X0():
X = Stimulus2D.random(8)
return X