Rapid Convergence of the Unadjusted Langevin Algorithm: Isoperimetry Suffices

Santosh Vempala, Andre Wibisono

Neural Information Processing Systems 

Notably, we do not assume convexity or bounds on higher derivatives. We also prove convergence guarantees in Rényi divergence of order q>1 assuming the limit of ULA satisfies either log-Sobolev or Poincaré inequality.