physbo.gp.core.learning module

class physbo.gp.core.learning.adam(gp, config)[source]

Bases: online

default

get_one_update(params, X, t)[source]
Parameters:

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

  • 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 vector that represents the corresponding negative energy of search candidates.

reset()[source]
class physbo.gp.core.learning.batch(gp, config)[source]

Bases: object

basis class for batch learning

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 vector that represents the corresponding negative energy of search candidates.

Returns:

The parameters which give the minimum marginal likelihood.

Return type:

numpy.ndarray

one_run(params, X, t, max_iter=None)[source]
Parameters:

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

  • 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 vector that represents the corresponding negative energy of search candidates.

  • max_iter (int) – Maximum number of iterations to perform.

Returns:

The solution of the optimization.

Return type:

numpy.ndarray

run(X, t)[source]

Performing optimization using the L-BFGS-B algorithm

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 vector that represents the corresponding negative energy of search candidates.

Returns:

The solution of the optimization.

Return type:

numpy.ndarray

class physbo.gp.core.learning.online(gp, config)[source]

Bases: object

base class for online learning

disp_marlik(params, eval_X, eval_t, num_epoch=None)[source]

Displaying marginal likelihood

Parameters:

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

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

  • eval_t (numpy.ndarray) – N-dimensional vector that represents the corresponding negative energy of search candidates.

  • num_epoch (int) – Number of epochs

get_one_update(params, X, t)[source]

Initial parameter searchs

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 vector that represents the corresponding negative energy of search candidates.

Returns:

The parameter which gives the minimum likelihood.

Return type:

numpy.ndarray

one_run(params, X, t, max_epoch=None, is_disp=False)[source]
Parameters:

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

  • 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 vector that represents the corresponding negative energy of search candidates.

  • max_epoch (int) – Maximum candidate epochs

Returns:

The solution of the optimization.

Return type:

numpy.ndarray

run(X, t)[source]

Run initial search and hyper parameter running.

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 vector that represents the corresponding negative energy of search candidates.

Returns:

The solution of the optimization.

Return type:

numpy.ndarray