mgplvm.syndata.gen_data module

class mgplvm.syndata.gen_data.Euclid(d)

Bases: mgplvm.syndata.gen_data.Manif

distance(x, y)

gen(n, n_samples, ell=None, sig=10, prefs=False)

gen_ginit(n, n_samples)

property name

noisy_conds(gs, variability)

class mgplvm.syndata.gen_data.Gen(manifold, n, m, l=0.5, alpha=1, beta=0.3, sigma=0.1, variability=0.1, n_samples=1)

Bases: object

gen_data(gs_in=None, gprefs_in=None, mode='Gaussian', overwrite=True, sigma=None, ell=None, sig=10, rate=10)

tbin is time of each time step (by default each time step is 1 ms) gs_in is optional input latent signal, otherwise random points on manifold generate Gaussian noise neural activities generate IPP spiking from Gaussian bump rate model rate is the mean peak firing rate across neurons

gen_gconds(ell=None, sig=10)

generate conditions for each neuron

gen_gprefs()

generate prefered directions for each neuron

get_data()

get_params()

return parameters

noisy_conds()

add noise to the conditions

set_param(param, value)

set the value of a parameter

class mgplvm.syndata.gen_data.Manif(d)

Bases: object

abstract distance(x, y)

abstract gen(n, n_samples, ell, sig)

abstract name()

class mgplvm.syndata.gen_data.Product(manifs)

Bases: mgplvm.syndata.gen_data.Manif

Does not support product of products at the moment

add_tangent_vector(gs, d, delta)

distance(xs, ys)

distance_scaled(xs, ys, ls)

gen(n, n_samples, ell=None, sig=10, prefs=False)

gen_ginit(n, n_samples)

generate a series of points at the origin

property name

class mgplvm.syndata.gen_data.So3

Bases: mgplvm.syndata.gen_data.Manif

distance(x, y)

gen(n, n_samples, ell=None, sig=None, prefs=False)

generate random points in spherical space according to the prior

gen_ginit(n, n_samples)

property name

noisy_conds(gs, variability)

norm(gs)

class mgplvm.syndata.gen_data.Sphere(d)

Bases: mgplvm.syndata.gen_data.Manif

distance(x, y)

gen(n, n_samples, ell=None, sig=None, prefs=False)

generate random points in spherical space according to the prior

gen_ginit(n, n_samples)

property name

noisy_conds(gs, variability)

norm(gs)

class mgplvm.syndata.gen_data.Torus(d)

Bases: mgplvm.syndata.gen_data.Manif

distance(x, y)

gen(n, n_samples, ell=None, sig=10, prefs=False)

if l is none, draw random samples - otherwise draw from an RBF GP with ell = l

gen_ginit(n, n_samples)

property name

noisy_conds(gs, variability)

norm(gs)

mgplvm.syndata.gen_data.draw_GP(n, d, n_samples, sig, ell, jitter=1e-06)

draw RBF GP samples with (n*n_samples) x samples, ell = l (in units of samples)