def test_glm_any_model():
nbin = 10
align = 'hold'
lag = 0.1 # s
ds = datasets['small']
dc = DataCollection(ds.get_files())
dc.add_unit(ds.get_units()[0], lag)
bnd = dc.make_binned(nbin=nbin, align=align)
ntask, nrep, nunit, nbin = bnd.shape
# make perfect test counts
# direction-only model
tp = radians([20, 10]) # degrees
b0 = log(10) # log(Hz)
pd = pol2cart(tp)
drn = kinematics.get_idir(bnd.pos, axis=2)
rate = exp(dot(drn, pd) + b0)
window_size = diff(bnd.bin_edges, axis=2)
count_mean = rate * window_size
count = poisson(count_mean)
count = count[:,:,None]
bnd.set_PSTHs(count)
# now fit data for pd
count, pos, time = bnd.get_for_regress()
bnom, bse_nom = glm_any_model(count, pos, time, model='kd')
pd_exp = unitvec(bnom['d'])
tp_exp = cart2pol(pd_exp)
acceptable_err = 0.05 # about 3 degrees absolute error
err = abs(tp - tp_exp)
assert_array_less(err, acceptable_err)
def test_cart2pol():
v = np.array([0., 0., 1.])
np.testing.assert_equal(cart2pol(v), np.array([0.,0.]))
v = np.array([0., 0., -1.])
np.testing.assert_equal(cart2pol(v), np.array([np.pi, 0.]))
v = np.array([1., 0., 0.])
np.testing.assert_equal(cart2pol(v), np.array([np.pi/2., 0.]))
v = np.array([1., 1., 0.])
v /= norm(v)[...,None]
np.testing.assert_equal(cart2pol(v), np.array([np.pi/2., np.pi/4.]))
v = np.array([0., 1., 0.])
v /= norm(v)[...,None]
np.testing.assert_equal(cart2pol(v), np.array([np.pi/2., np.pi/2.]))
v = np.array([[0., 0., -1.],[1., 0., 0.]]) # shape 2,3
exp = np.array([[np.pi, 0.], [np.pi/2., 0.]])
np.testing.assert_equal(cart2pol(v), exp)
v = np.array([[0., 0., -1.],[1., 0., 0.]]).T # shape 2,3
exp = np.array([[np.pi, 0.], [np.pi/2., 0.]]).T
np.testing.assert_equal(cart2pol(v, axis=0), exp)
def test_pol2cart_cart2pol():
v = np.random.random(size=(2, 3))
v /= norm(v, axis=1)[...,None]
np.testing.assert_almost_equal(v, pol2cart(cart2pol(v)))
def plot_gem(data, order='co', fig=None, ax=None,
clim=None, **kwargs):
'''
Parameters
----------
data : ndarray
shape (26,)
order : string, optional
one of 'co' or 'oc', for center-out or out-center data order
determines mapping of data to polygons
defaults to center-out
fig : matplotlib Figure instance
an existing figure to use, optional
ax : SplitLambertAxes instance
an existing axis to use, optional
clim : tuple or list of float, optional
minimum and maximum values for color map
if not supplied, min and max of data are used
'''
if not data.shape == (26,):
raise ValueError('data has wrong shape; should be (26,),'
' actually is %s' % (str(data.shape)))
if order == 'co':
mapping = get_targ_co_dir_mapping()
elif order == 'oc':
mapping = get_targ_oc_dir_mapping()
else:
raise ValueError('`order` not recognized')
if (fig != None) & ~isinstance(fig, Figure):
raise ValueError('`fig` must be an instance of Figure')
if (ax != None) & \
~isinstance(ax, split_lambert_projection.SplitLambertAxes):
raise ValueError('`ax` must be an instance of SplitLambertAxes')
if (clim != None):
if (type(clim) != tuple) and (type(clim) != list):
raise ValueError('clim must be tuple or list')
if len(clim) != 2:
raise ValueError('clim must have 2 elements')
else:
clim = (np.min(data), np.max(data))
if (fig == None):
if (ax != None):
fig = ax.figure
else:
fig = plt.figure()
if (ax == None):
ax = fig.add_subplot(111, projection='split_lambert')
pts, spts, circs = make_26_targets()
pts = pts.astype(float)
pts /= norm(pts, axis=1)[:,None]
tps = cart2pol(pts)
q = 9
cnrs = np.zeros((32, 5, 2)) # for non-polar polys
ends = np.zeros((2, q, 2)) # for poles
# for now use boundaries suggested by square ends
squares = np.unique(tps[circs[0]][:,0])
mids = np.mean(rolling_window(squares, 2), axis=1)
eps = 1e-8
tvs = np.hstack([mids[:2], np.pi / 2. - eps, np.pi / 2., mids[-2:]])
pvs = np.linspace(0, 2 * np.pi, q, endpoint=True) % (2 * np.pi) \
- np.pi / 8.
k = 0
rings = [0,1,3,4]
for i in rings: # 3 rings of 4-cnr polys
for j in xrange(8): # 8 polys per ring
xs = np.array((tvs[i], tvs[i], tvs[i+1], tvs[i+1], tvs[i]))
ys = np.array((pvs[j], pvs[j+1], pvs[j+1], pvs[j], pvs[j]))
cnrs[k] = np.vstack((xs, ys)).T
k += 1
ends[0,:,0] = np.ones(q) * tvs[0]
ends[1,:,0] = np.ones(q) * tvs[-1]
ends[0,:,1] = pvs
ends[1,:,1] = pvs
kwargs.setdefault('edgecolors', 'none')
kwargs.setdefault('linewidths', (0.25,))
coll_cnrs = mcoll.PolyCollection(cnrs, **kwargs)
coll_ends = mcoll.PolyCollection(ends, **kwargs)
mapped_data = data[mapping]
coll_cnrs.set_array(mapped_data[:-2])
coll_ends.set_array(mapped_data[-2:])
if clim == None:
cmin, cmax = mapped_data.min(), mapped_data.max()
else:
cmin, cmax = clim
coll_cnrs.set_clim(cmin, cmax)
coll_ends.set_clim(cmin, cmax)
ax.add_collection(coll_cnrs)
ax.add_collection(coll_ends)
return ax
