RationalQuadratic#
- class gpjax.kernels.RationalQuadratic(active_dims=None, lengthscale=1.0, variance=1.0, alpha=1.0, n_dims=None, compute_engine=<gpjax.kernels.computations.dense.DenseKernelComputation object>)[source]#
Bases:
StationaryKernelThe Rational Quadratic kernel.
Computes the covariance for pairs of inputs \((x, y)\) with lengthscale parameter \(\ell\), variance \(\sigma^2\) and shape parameter \(\alpha\).
\[ k(x,y)=\sigma^2\Bigg(1+\frac{\lVert x-y\rVert^2_2}{2\alpha\ell^2}\Bigg)^{-\alpha} \]As \(\alpha \to \infty\) this recovers the
RBFkernel; it is equivalently a scale mixture of RBF kernels with a Gamma-distributed inverse squared lengthscale.- Parameters:
lengthscale (AbstractUnwrappable)
variance (AbstractUnwrappable)
alpha (Any)
n_dims (int | None)
compute_engine (AbstractKernelComputation)