physbo.gp.cov.gauss module

class physbo.gp.cov.gauss.gauss(num_dim, width=3, scale=1, ard=False, max_width=1000000.0, min_width=1e-06, max_scale=1000000.0, min_scale=1e-06)

Bases: object

gaussian kernel

cat_params(width, scale)

Taking the logarithm of width and scale parameters and concatinate them into one ndarray

Parameters:

• width (int) –

• scale (int) –

Returns:

params – Parameters

Return type:

numpy.ndarray

decomp_params(params)

decompose the parameters defined on the log region into width and scale parameters

Parameters:

params (numpy.ndarray) – parameters

Returns:

• width (float)

• scale (float)

get_cand_params(X, t)

Getting candidate parameters.

Parameters:

• X (numpy.ndarray) – N x d dimensional matrix. Each row of X denotes the d-dimensional feature vector of search candidate.

• t (numpy.ndarray) – N dimensional array. The negative energy of each search candidate (value of the objective function to be optimized).

Returns:

params

Return type:

numpy.ndarray

get_cov(X, Z=None, params=None, diag=False)

compute the covariant matrix :param X: N x d dimensional matrix. Each row of X denotes the d-dimensional feature vector of search candidate. :type X: numpy.ndarray :param Z: N x d dimensional matrix. Each row of Z denotes the d-dimensional feature vector of search candidate. :type Z: numpy.ndarray :param params: Parameters :type params: numpy.ndarray :param diag: If X is the diagonalization matrix, true. :type diag: bool

Returns:

G – covariant matrix

Return type:

numpy.ndarray

get_grad(X, params=None)

Getting gradiant values of X

Parameters:

• X (numpy.ndarray) – N x d dimensional matrix. Each row of X denotes the d-dimensional feature vector of search candidate.

• params (numpy.ndarray) –

Returns:

grad

Return type:

numpy.ndarray

get_params_bound()

Getting boundary array.

Returns:

bound – A num_params-dimensional array with the tuple (min_params, max_params).

Return type:

list

load(file_name)

Recovering the Gaussian kernel from file :param file_name: file name to load the information of the kernel :type file_name: str

prepare(params=None)

Setting parameters

Parameters:

params (numpy.ndarray) – parameters

Returns:

• params (numpy.ndarray)

• width (int)

• scale (int)

print_params()

show the current kernel parameters

rand_expans(num_basis, params=None)

Kernel Expansion

Parameters:

• num_basis (int) – total number of basis

• params (numpy.ndarray) – Parameters

Return type:

tupple (W, b, amp)

save(file_name)

save the gaussian kernel

Parameters:

file_name (str) – file name to save the information of the kernel

set_params(params)

set kernel parameters

Parameters:

params (numpy.ndarray) – Parameters for optimization.

supp_params(params)

Set maximum (minimum) values for parameters when the parameter is greater(less) than this value.

Parameters:

params (numpy.ndarray) – Parameters for optimization. Array of real elements of size (n,), where ‘n’ is the number of independent variables.

Returns:

params

Return type:

numpy.ndarray