def test_grating():
gm = movie.GratingMovie(row_size=120, col_size=240, frame_rate=1000.0)
# Default grating movie
mv = gm.create_movie()
assert (mv.data.shape == (1001, 60, 120))
assert (mv.row_range[0] == 0.0)
assert (mv.row_range[-1] == 120.0)
assert (mv.col_range[0] == 0.0)
assert (mv.col_range[-1] == 240.0)
assert (mv.t_range[0] == 0)
assert (mv.t_range[-1] == 1.0)
# 2 second gratings with 0.5 second gray screen
mv = gm.create_movie(t_max=2.0, gray_screen_dur=0.5)
assert (mv.data.shape == (2001, 60, 120))
assert (mv.t_range[0] == 0.0)
assert (mv.t_range[-1] == 2.0)
assert (np.sum(mv.data[0:500, :, :]) == 0) # check 0.5 second grey screen
assert (np.sum(mv.data[501, :, :]) > 0) # movie starts
# disable aliasing
mv = gm.create_movie(cpd=0.1)
assert (mv.data.shape == (1001, 120, 240))
assert (mv.t_range[0] == 0.0)
assert (mv.t_range[-1] == 1.0)
assert (mv.row_range[-1] == 120)
assert (mv.col_range[-1] == 240)
def plot_sfilter_params():
gm = movie.GratingMovie(200, 200)
mv = gm.create_movie(t_max=2.0)
fig, axes = plt.subplots(2, 2, figsize=(7, 7))
rotations = [0.0, 45.0]
sigmas = [(30.0, 20.0), (20.0, 30.0)]
for r, sigma in enumerate(sigmas):
for c, rot in enumerate(rotations):
gsf = GaussianSpatialFilter(translate=(0, 0),
sigma=sigma,
rotation=rot)
axes[r, c].imshow(gsf.get_kernel(mv.row_range,
mv.col_range).full(),
extent=(0, 200, 0, 200),
origin='lower')
if r == 0:
axes[r, c].title.set_text('spatial_rotation={}'.format(rot))
if c == 0:
axes[r, c].set_ylabel('spatial_size={}'.format(sigma))
plt.show()
def test_offunit(cell_type, expected_val):
gm = movie.GratingMovie(row_size=120, col_size=240, frame_rate=24.0)
mv = gm.create_movie(t_max=1.0)
cell = default_cell_loader(MockNode(), ('lgnmodel', cell_type), dynamics_params)
assert(isinstance(cell, OffUnit))
lgn = LGNModel([cell])
results = lgn.evaluate(mv, downsample=5) # Does the filtering + non-linearity on movie object m
times = np.array(results[0][0], dtype=np.float64)
rates = np.array(results[0][1], dtype=np.float64)
assert(np.allclose(times, [0.0, 0.2083, 0.4167, 0.6250, 0.8333], atol=1.0e-3))
assert(np.allclose(rates, expected_val, atol=1.0e-3))
def plot_tfilter_params():
gm = movie.GratingMovie(120, 240)
mv = gm.create_movie(t_max=2.0)
#
# tf = TemporalFilterCosineBump(weights=[33.328, -20.10059],
# kpeaks=[59.0, 120.0], # [9.67, 20.03],
# delays=[0.0, 0.0])
#
# tf.get_kernel(t_range=mv.t_range, threshold=0.0, reverse=True)
# tf.imshow(t_range=mv.t_range, reverse=True)
fig, axes = plt.subplots(3, 3, figsize=(10, 7))
ri = ci = 0
weights = [(30.0, -20.0), (30.0, -1.0), (15.0, -20.0)]
kpeaks = [(3.0, 5.0), (3.0, 30.0), (20.0, 40.0)]
delays = [(0.0, 0.0), (0.0, 60.0), (20.0, 60.0)]
# weights control the amplitude of the peaks
for ci, w in enumerate(weights):
tf = TemporalFilterCosineBump(weights=w,
kpeaks=[9.67, 20.03],
delays=[0.0, 1.0])
linear_kernel = tf.get_kernel(t_range=mv.t_range, reverse=True)
axes[ri, ci].plot(linear_kernel.t_range[linear_kernel.t_inds],
linear_kernel.kernel)
axes[ri, ci].set_ylim([-3.5, 10.0])
axes[ri, ci].text(0.05,
0.90,
'weights={}'.format(w),
horizontalalignment='left',
verticalalignment='top',
transform=axes[ri, ci].transAxes)
axes[0, 0].set_ylabel('effect of weights')
ri += 1
# kpeaks parameters controll the spread of both peaks, the second peak must have a bigger spread
for ci, kp in enumerate(kpeaks):
tf = TemporalFilterCosineBump(weights=[30.0, -20.0],
kpeaks=kp,
delays=[0.0, 1.0])
linear_kernel = tf.get_kernel(t_range=mv.t_range, reverse=True)
axes[ri, ci].plot(linear_kernel.t_range[linear_kernel.t_inds],
linear_kernel.kernel)
axes[ri, ci].set_xlim([-0.15, 0.005])
axes[ri, ci].text(0.05,
0.90,
'kpeaks={}'.format(kp),
horizontalalignment='left',
verticalalignment='top',
transform=axes[ri, ci].transAxes)
axes[1, 0].set_ylabel('effects of kpeaks')
ri += 1
for ci, d in enumerate(delays):
tf = TemporalFilterCosineBump(weights=[30.0, -20.0],
kpeaks=[9.67, 20.03],
delays=d)
linear_kernel = tf.get_kernel(t_range=mv.t_range, reverse=True)
axes[ri, ci].plot(linear_kernel.t_range[linear_kernel.t_inds],
linear_kernel.kernel)
axes[ri, ci].set_xlim([-0.125, 0.001])
axes[ri, ci].text(0.05,
0.90,
'delays={}'.format(d),
horizontalalignment='left',
verticalalignment='top',
transform=axes[ri, ci].transAxes)
axes[2, 0].set_ylabel('effects of delays')
# plt.xlim()
plt.show()