physbo.blm.lik package

Module contents

class physbo.blm.lik.Cov(params=None)

Bases: object

Covariance

params

half of log of variance

Type:

float

nparams

number of parameters

Type:

int

sigma2

variance

Type:

float

prec

inv. of variance

Type:

float

get_cov(N, params=None)

compute the covariance of prior

Parameters:

• N (int) – dimension

• params – half of log of variance (default: self.params)

Returns:

NxN covariance matrix

Return type:

numpy.ndarray

get_prec(N, params=None)

compute the precision of prior

Parameters:

• N (int) – dimension

• params – half of log of variance (default: self.params)

Returns:

inverse of covariance matrix

Return type:

numpy.ndarray

set_params(params)

set the parameter

Parameters:

params (float) – half of log of variance

class physbo.blm.lik.Gauss(linear, cov)

Bases: object

Gaussian

linear

linear kernel

Type:

physbo.blm.lik.Linear

cov

covariance

Type:

physbo.blm.lik.Cov

stats

get_basis(X)

calculates value of basis function at input

Parameters:

X (numpy.ndarray) – input

See also

physbo.blm.basis.fourier.get_basis

get_cov(N, params=None)

Returns covariance matrix

Parameters:

• N (int) – dimension

• params (float) – half of log of variance (default: self.cov.params)

Returns:

NxN covariance matrix

Return type:

numpy.ndarray

get_mean(X, Psi=None, params=None, bias=None)

calculates mean value

Parameters:

• X (numpy.ndarray) – raw input

• Psi (numpy.ndarray) – value of feature maps

• params (numpy.ndarray) – weight

• bias (float) – bias

See also

physbo.blm.basis.fourier.get_mean

get_prec(N, params=None)

Returns precision matrix

Parameters:

• N (int) – dimension

• params (float) – half of log of variance (default: self.cov.params)

Returns:

NxN precision matrix

Return type:

numpy.ndarray

sampling(fmean)

draws samples

Parameters:

fmean (numpy.ndarray) – means of samples

Returns:

samples

Return type:

numpy.ndarray

set_bias(bias)

sets bias

set_params(params)

sets parameters

class physbo.blm.lik.Linear(basis, params=None, bias=None)

Bases: object

basis

basis for random feature map

nbasis

number of basis

Type:

int

bias

params

_init_params

initial value of the parameter

get_mean(X, Psi=None, params=None, bias=None)

calculate mean values

Parameters:

• X (numpy.ndarray) – input as an N-by-d matrix

• Psi (numpy.ndarray) – feature maps Psi(X) as an N-by-l matrix (default: self.get_basis(X))

• params (numpy.ndarray) – weight as a vector with size l (default: self.params)

• bias (float) – (default: self.bias)

Returns:

Psi * params + bias

Return type:

numpy.ndarray

set_bias(bias)

set bias

Parameters:

bias (float)

set_params(params)

set parameters

Parameters:

params (np.ndarray)