Another option to fit spde in brms if it is not feasible to fit into the smooth2random re-parametrization is to define a new model formula term in brms like:
brm(resp ~ 1 + spde(mesh = mesh), data = dat)
We would then have more flexibility in how this term is defined and which functions it uses. But we would need to tap deeper into the brms machinery.
Another option to fit spde in brms if it is not feasible to fit into the
smooth2random
re-parametrization is to define a new model formula term in brms like:We would then have more flexibility in how this term is defined and which functions it uses. But we would need to tap deeper into the brms machinery.