el0ps.solver.BnbSolver

class el0ps.solver.BnbSolver(relative_gap=1e-08, absolute_gap=0.0, time_limit=inf, node_limit=None, queue_limit=None, queue_strategy='bound', verbose=False, **kwargs)

Branch-and-Bound solver for L0-regularized problems.

The problem is expressed as

\[\textstyle\min_{\mathbf{x} \in \mathbb{R}^{n}} f(\mathbf{Ax}) + \lambda\|\mathbf{x}\|_0 + h(\mathbf{x})\]

where \(f\) is a el0ps.datafit.BaseDatafit function, \(\mathbf{A} \in \mathbb{R}^{m \times n}\) is a matrix, \(h\) is a el0ps.penalty.BasePenalty function, and \(\lambda\) is a positive scalar.

Parameters:

relative_gap: float, default=1e-8

Relative gap targeted on the objective value by the solver.

absolute_gap: float, default=0.0

Absolute gap targeted on the objective value by the solver.

time_limit: float, default=None

Limit in second on the solving time.

node_limit: int, default=None

Limit on the number of nodes explored during the BnB algorithm.

queue_limit: int, default=None

Limit on the number of nodes in the queue during the BnB algorithm.

queue_strategy: str, default=”bound”

Queue strategy during the BnB algorithm. Can be one of the following: - “bound”: select nodes with the worst lower bound first - “objective”: select nodes with the best upper bound first - “breadth”: select shallower nodes in the BnB tree first - “depth”: select deepest nodes in the BnB tree first - “fifo”: select nodes lastly added to the queue first - “lifo”: select nodes firstly added to the queue first - “random”: select nodes randomly in the queue - “local_gap”: select nodes with the closest gap between their upper and lower bounds first

verbose: bool, default=True

Toggle solver verbosity.

**kwargs: keyword arguments

Additional keyword arguments passed to a BoundSolver instance used for the bounding step of the Branch-and-Bound algorithm.

__init__(relative_gap=1e-08, absolute_gap=0.0, time_limit=inf, node_limit=None, queue_limit=None, queue_strategy='bound', verbose=False, **kwargs)

Methods

__init__([relative_gap, absolute_gap, ...])
package_results(problem, results)
solve(datafit, penalty, A, lmbd[, x_init])	Solve an L0-regularized problem.

Attributes

accept_jitclass