Deep Compositional Spatial Models
Affine transformation on a 1D domain
Affine transformation on a 2D domain
Axial Warping Unit
Bisquare functions on a 1D domain
Bisquare functions on a 2D domain
Deep bivariate compositional spatial model for Gaussian processes
Deep compositional spatial model for Gaussian processes
Deep compositional spatial model for max-stable processes
Deep compositional spatial model (with nearest neighbors) for Gaussian...
Deep compositional spatio-temporal model (with nearest neighbors) for ...
Deep compositional spatial model for r-Pareto processes
Deep trivariate compositional spatial model for Gaussian processes
Deep compositional spatial models
Initialise learning rates
Initialise weights and parameters
LFT (Möbius transformation)
Deep bivariate compositional spatial model
Deep compositional spatial model
Deep compositional spatial model (with nearest neighbors)
Deep compositional spatio-temporal model (with nearest neighbors)
Deep trivariate compositional spatial model
Deep compositional spatial model
Radial Basis Function Warpings
Set TensorFlow seed
Generate simulation data for testing
Deep compositional spatial model for max-stable processes
Deep compositional spatial model for r-Pareto processes
Deep compositional spatial models are standard spatial covariance models coupled with an injective warping function of the spatial domain. The warping function is constructed through a composition of multiple elemental injective functions in a deep-learning framework. The package implements two cases for the univariate setting; first, when these warping functions are known up to some weights that need to be estimated, and, second, when the weights in each layer are random. In the multivariate setting only the former case is available. Estimation and inference is done using `tensorflow`, which makes use of graphics processing units. For more details see Zammit-Mangion et al. (2022) <doi:10.1080/01621459.2021.1887741>, Vu et al. (2022) <doi:10.5705/ss.202020.0156>, Vu et al. (2023) <doi:10.1016/j.spasta.2023.100742>, and Shao et al. (2025) <doi:10.48550/arXiv.2505.12548>.