We can add a $\lambda$ parameter that specifically implements an isotropic Gaussian prior inside the SGMCMC algorithms (and perhaps others too such as VI)
This is useful for splitting integrators such as BAOA which can integrate the components of the dynamics that are linear exactly (i.e. OU dynamics) such as those induced by a Gaussian prior.
We can add a$\lambda$ parameter that specifically implements an isotropic Gaussian prior inside the SGMCMC algorithms (and perhaps others too such as VI)
This is useful for splitting integrators such as BAOA which can integrate the components of the dynamics that are linear exactly (i.e. OU dynamics) such as those induced by a Gaussian prior.