Schoenberg (1938) showed how Bochner's basic representation theorem for positive definite functions (e.g. correlation function of a stationary stochastic process) `simplifies' for spatial processes (d-dimensional random fields) which are isotropic: the standard Fourier kernel function is replaced by the characteristic function of a random direction in d-space and the spectral measure, instead of being on d-space, is on the positive half-line.
The talk describes how Wendland's `dimension walks', which were defined earlier by Matheron as Descente and Montee in studying relations between d-D and either (d+2)-D or (d-2)-D correlation functions, are equivalent to simple modifications of their d-Schoenberg measures.
Another family of dimension walks arises from projections from unit d-spheres to lower dimensional spheres, first via the kernel functions in the Schoenberg representation and then more generally, for d-Schoenberg measures.
For a spatial point process model fitted to spatial point pattern data, we develop diagnostics for model validation, analogous to the classical measures of leverage and influence and residual plots in a generalized linear model. The diagnostics can be characterised as derivatives of basic functionals of the model.
They can also be derived heuristically (and computed in practice) as the limits of classical diagnostics under increasingly fine discretizations of the spatial domain.
We apply the diagnostics to example datasets where there are concerns about model validity.