def animate_and_follow(individual=0, num_neighbours=15):
test_trajectories_file = os.path.join(dir_of_data, 'test_trajectories.npy')
t = np.load(test_trajectories_file)
tt.interpolate_nans(t)
tt.center_trajectories_and_normalise(t)
s_ = tt.smooth(t)
v_ = tt.smooth_velocity(t)
e_ = tt.normalise(v_[:, individual, :])
s = tt.center_in_individual(s_, individual)
s = tt.fixed_to_comoving(s, e_)
v = tt.fixed_to_comoving(v_, e_)
center = tt.fixed_to_comoving(-s_[:, [individual], :], e_)
indices = ttsocial.give_indices(s, num_neighbours)
sn = ttsocial.restrict(s, indices, individual)
vn = ttsocial.restrict(v, indices, individual)
sf = s[:, [individual], :]
vf = v[:, [individual], :]
anim = ttanimation.scatter_vectors(s, velocities=v, k=10)
anim += ttanimation.scatter_ellipses(s, velocities=v, color='c')
anim += ttanimation.scatter_ellipses(sn, velocities=vn, color='b')
anim += ttanimation.scatter_ellipses(sf, velocities=vf, color='r')
anim += ttanimation.scatter_circle(center)
anim.prepare()
anim.show()
def get_average_local_polarization(t, number_of_neighbours = 1):
if t.number_of_individuals == 1:
return('nan')
else:
indices = ttsocial.give_indices(t.s, number_of_neighbours)
en = ttsocial.restrict(t.e,indices)[...,1:,:]
local_polarization = tt.norm(tt.collective.polarization(en))
return np.nanmean(local_polarization, axis = -1)
def local_polarization(tr, n=1, x=1):
if n < x:
return (np.nan, np.nan)
if tr.number_of_individuals == 1:
return (np.nan, np.nan)
if x >= tr.number_of_individuals:
return (np.nan, np.nan)
else:
indices = ttsocial.give_indices(
tr.s, n) # indices of the closest n neighbors
a = ttsocial.restrict(
tr.e, indices
) # normalized velocity (direction) vectors of focal individual and closest n neighbors
b = np.multiply(
a[:, :, 0, :],
a[:, :,
x, :]) #polarization between focal and xth closest individual
c = np.sum(b, axis=2) #dot product of vectors.
lp = tt.norm(tt.collective.polarization(a))
#return(c)
return (np.nanmean(lp), np.nanmean(c))
def local_polarization_func(tr, n=1, x=1): #shape returns tr.s.shape -2
if n < x:
return (np.nan, np.nan)
if tr.number_of_individuals == 1:
return (np.nan, np.nan)
if x >= tr.number_of_individuals:
return (np.nan, np.nan)
else:
indices = ttsocial.neighbour_indices(
tr.s[1:-1], n) # indices of the closest n neighbors
a = ttsocial.restrict(
tr.e[1:-1], indices
) # normalized velocity (direction) vectors of focal individual and closest n neighbors
b = np.multiply(
a[:, :, 0, :],
a[:, :,
x, :]) #polarization between focal and xth closest individual
c = np.sum(b, axis=2) #dot product of vectors.
lp = tt.norm(tt.collective.polarization(a))
#return(c)
return (np.nanmean(lp[~filter_speed(tr, 5).mask]),
np.nanmean(c[~filter_speed(tr, 5).mask]))
def local_polarization(tr, frames, n=1, x=1): #uses filtered data
if n < x:
return (np.nan)
if tr.number_of_individuals == 1:
return (np.nan)
if x >= tr.number_of_individuals:
return (np.nan)
else:
indices = ttsocial.neighbour_indices(
filter_low_pass(tr), n) # indices of the closest n neighbors
a = ttsocial.restrict(
filter_direction_low_pass(tr)[1:-1], indices
) # normalized velocity (direction) vectors of focal individual and closest n neighbors
b = np.multiply(
a[:, :, 0, :],
a[:, :,
x, :]) #polarization between focal and xth closest individual
b_mask = np.ma.array(b, mask=filter_direction_low_pass(tr)[1:-1].mask)
c = np.nansum(b_mask, axis=2) #dot product of vectors.
#lp = tt.norm(tt.collective.polarization(a))
#return(c)
return (np.nanmean(c[frames]))
def get_average_aligment_score(t, number_of_neighbours = 3):
indices = ttsocial.give_indices(t.s, number_of_neighbours)
en = ttsocial.restrict(t.e,indices)[...,1:,:]
alignment = np.nanmean(tt.dot(np.expand_dims(t.e,2), en), axis = -1)
return np.nanmedian(alignment, axis = -1)