GMRF
A Gaussian random field whose precision matrix is sparse under a conditional independence structure. It supports fast sparse linear algebra operations and is a prerequisite for efficient inference in INLA.
Timeline (last 90 days)
Lindgren et al. proved that a Gaussian process with a Matérn covariance function can be represented as the solution of some stochastic partial differential equation (SPDE), and after discretizing it on a triangular mesh with the finite element method, its coefficients form a Gaussian Markov random field (GMRF).
ISBO adopts the INLA-SPDE method, using a Matérn field on a mesh to approximate a dense Gaussian process as a sparse Gaussian Markov random field (GMRF), thereby supporting fast posterior inference.
All Facts (2)
Lindgren et al. proved that a Gaussian process with a Matérn covariance function can be represented as the solution of some stochastic partial differential equation (SPDE), and after discretizing it on a triangular mesh with the finite element method, its coefficients form a Gaussian Markov random field (GMRF).
50%UnverifiedISBO adopts the INLA-SPDE method, using a Matérn field on a mesh to approximate a dense Gaussian process as a sparse Gaussian Markov random field (GMRF), thereby supporting fast posterior inference.
50%