supports_hessian: bool

__init__(self, func, jacobian=None, hessian=None, *, check_generator=True, logger=None, **generator_opts): 

LLM Docstring

Initialize a Newton step finder, building a direct-Hessian generator from the supplied Jacobian/Hessian unless one is already provided.

• func: Callable

  the objective function

• jacobian: Callable | None

  the gradient function (or a ready generator)

• hessian: Callable | None

  the Hessian function

• check_generator: bool

  build a generator when needed

• logger: object | None

  optional logger

• generator_opts: Any

  extra options for the generator

__call__(self, guess, mask, return_vals=False, projector=None): 

LLM Docstring

Produce a Newton step for the active members.

• guess: np.ndarray

  current parameters

• mask: np.ndarray | tuple

  active-member indices (or chain-minimizer tuple)

• return_vals: bool

  unsupported

• projector: np.ndarray | None

  optional projector applied to the step

• :returns: tuple

  (step, gradient)