__init__(self, data, dimension=None, base=None, **metadata): 

• data: Any

• dimension: Any

• base: None | Type[BaseSurface]

• metadata: Any

@property
data(self): 

minimize(self, initial_guess=None, function_options=None, **opts): 

Provides a uniform interface for minimization, basically just dispatching to the BaseSurface implementation if provided

• initial_guess: np.ndarray | None

  initial starting point for the minimization

• function_options: None | dict

• opts: Any

• :returns: _

detect_base(data, opts): 

Infers what type of base surface works for the data that’s passed in. It’s super roughly done so…yeah generally better to pass the base class you want explicitly. But in the absence of that we can do this ?_?

Basic strategy:

1. look for options that go with specific methods
2. look at data structures to guess i. gradient as the first data arg + all data args are ndarrays -> Taylor Series ii. callables as second arg -> Linear expansion or Linear fit iii. just like…one big array -> Interpolatin
   • data: tuple

   • opts: dict

   • :returns: _

__call__(self, gridpoints, **kwargs):