L_BFGS

L_BFGS()

Limited-memory BFGS optimizer for smooth objectives.

Use this optimizer for differentiable full-batch objectives where a quasi-Newton method is appropriate. The objective callable must accept a parameter vector and a writable gradient vector, and return the scalar objective value.

Attributes

Name	Description
armijoConstant	Armijo line-search constant.
factr	Relative objective tolerance used by the optimizer.
maxIterations	Maximum number of optimizer iterations.
maxLineSearchTrials	Maximum number of line-search attempts per iteration.
maxStep	Upper bound on the line-search step size.
minGradientNorm	Termination tolerance based on gradient norm.
minStep	Lower bound on the line-search step size.
numBasis	Number of correction vectors retained in memory.
wolfe	Wolfe curvature constant.

Methods

Name	Description
optimize	optimize(self: pyensmallen._pyensmallen.L_BFGS, objective: collections.abc.Callable[[typing.Annotated[numpy.typing.ArrayLike, numpy.float64], typing.Annotated[numpy.typing.ArrayLike, numpy.float64]], float], initial_point: typing.Annotated[numpy.typing.ArrayLike, numpy.float64]) -> numpy.typing.NDArray[numpy.float64]

optimize

L_BFGS.optimize()

optimize(self: pyensmallen._pyensmallen.L_BFGS, objective: collections.abc.Callable[[typing.Annotated[numpy.typing.ArrayLike, numpy.float64], typing.Annotated[numpy.typing.ArrayLike, numpy.float64]], float], initial_point: typing.Annotated[numpy.typing.ArrayLike, numpy.float64]) -> numpy.typing.NDArray[numpy.float64]

Optimize an objective from the provided starting point.

Parameters

Name	Type	Description	Default
objective	callable	Callable with signature objective(params, gradient). The callable should write the gradient in place and return the scalar objective.	required
initial_point	ndarray	Initial parameter vector.	required

Returns

Name	Type	Description
	ndarray	Optimized parameter vector.