ApproximatingObjective
class ApproximatingObjective
Bases: abc.ABC
A base class for an optimization problem definition. An implementing class must provide at least an implementation of the objective method. In such case only gradient free optimizers can be used. Both method, objective and gradient, preferable to have in an implementation.
Methods Defined Here
gradient
abstract ApproximatingObjective.gradient(param_values)
Computes a gradient with respect to parameters given a vector of parameter values.
Parameters
param_values (ndarray) – a vector of parameter values for the optimization problem.
Return type
ndarray
Returns
an array of gradient values.
objective
abstract ApproximatingObjective.objective(param_values)
Computes a value of the objective function given a vector of parameter values.
Parameters
param_values (ndarray) – a vector of parameter values for the optimization problem.
Return type
float
Returns
a float value of the objective function.
Attributes
num_thetas
Returns: the number of parameters in this optimization problem.
Return type
int
target_matrix
Returns: a matrix being approximated
Return type
ndarray