saealib.RBFKernel#

class saealib.RBFKernel[source]#

Bases: ABC

Abstract base for RBF kernel implementations.

A subclass implements the radial profile phi(r); kernel-specific parameters (e.g. length_scale, nu) are held as object state rather than passed through a shared call signature.

Methods

`__init__`
`evaluate`	Evaluate `phi(r)` for a radial-distance matrix.
`pairwise`	Evaluate the kernel between every row of `x1` and `x2`.
`resolve`	Return a kernel whose data-dependent parameters are resolved.
`validate`	Validate static kernel configuration.

Method Details

RBFKernel.__init__()#

abstractmethod RBFKernel.evaluate(r)[source]#

Evaluate phi(r) for a radial-distance matrix.

Parameters:: r (ndarray)
Return type:: ndarray

RBFKernel.pairwise(x1, x2)[source]#

Evaluate the kernel between every row of x1 and x2.

Parameters:

x1 (ndarray)
x2 (ndarray)

Return type:

ndarray

RBFKernel.resolve(train_x)[source]#

Return a kernel whose data-dependent parameters are resolved.

Calls validate() before returning, so a custom subclass that overrides validate() but not resolve() still gets its configuration checked at resolve time.

Parameters:: train_x (ndarray)
Return type:: Self

RBFKernel.validate()[source]#

Validate static kernel configuration.

Return type:: None