Convergence of Kinetic Langevin Monte Carlo on Lie groups
This provides a foundational advancement for sampling on curved spaces, enabling applications in fields like robotics and physics where Lie groups are common, though it is incremental in extending kinetic Langevin methods to non-Euclidean settings.
The paper tackles the problem of sampling from distributions on Lie groups by constructing a kinetic Langevin Monte Carlo sampler that preserves the group structure and proves exponential convergence with explicit rates under the Wasserstein-2 distance, requiring only compactness and smoothness conditions without convexity assumptions.
Explicit, momentum-based dynamics for optimizing functions defined on Lie groups was recently constructed, based on techniques such as variational optimization and left trivialization. We appropriately add tractable noise to the optimization dynamics to turn it into a sampling dynamics, leveraging the advantageous feature that the trivialized momentum variable is Euclidean despite that the potential function lives on a manifold. We then propose a Lie-group MCMC sampler, by delicately discretizing the resulting kinetic-Langevin-type sampling dynamics. The Lie group structure is exactly preserved by this discretization. Exponential convergence with explicit convergence rate for both the continuous dynamics and the discrete sampler are then proved under $W_2$ distance. Only compactness of the Lie group and geodesically $L$-smoothness of the potential function are needed. To the best of our knowledge, this is the first convergence result for kinetic Langevin on curved spaces, and also the first quantitative result that requires no convexity or, at least not explicitly, any common relaxation such as isoperimetry.