The Mirror Langevin Algorithm Converges with Vanishing Bias

The technique of modifying the geometry of a problem from Euclidean to Hessian metric has proved to be quite effective in optimization, and has been the subject of study for sampling. The Mirror Langevin Diffusion (MLD) is a sampling analogue of mirror flow in continuous time, and it has nice convergence properties under log-Sobolev or Poincare inequalities relative to the Hessian metric, as shown by Chewi et al. (2020). In discrete time, a simple discretization of MLD is the Mirror Langevin Algorithm (MLA) studied by Zhang et al. (2020), who showed a biased convergence bound with a non-vanishing bias term (does not go to zero as step size goes to zero). This raised the question of whether we need a better analysis or a better discretization to achieve a vanishing bias. Here we study the Mirror Langevin Algorithm and show it indeed has a vanishing bias. We apply mean-square analysis based on Li et al. (2019) and Li et al. (2021) to show the mixing time bound for MLA under the modified self-concordance condition introduced by Zhang et al. (2020).

[1]  E. Lieb,et al.  On extensions of the Brunn-Minkowski and Prékopa-Leindler theorems, including inequalities for log concave functions, and with an application to the diffusion equation , 1976 .

[2]  M. V. Tretyakov,et al.  Stochastic Numerics for Mathematical Physics , 2004, Scientific Computation.

[3]  Hongyuan Zha,et al.  Sqrt(d) Dimension Dependence of Langevin Monte Carlo , 2021, ArXiv.

[4]  Arnak S. Dalalyan,et al.  User-friendly guarantees for the Langevin Monte Carlo with inaccurate gradient , 2017, Stochastic Processes and their Applications.

[5]  Volkan Cevher,et al.  Mirrored Langevin Dynamics , 2018, NeurIPS.

[6]  É. Moulines,et al.  Non-asymptotic convergence analysis for the Unadjusted Langevin Algorithm , 2015, 1507.05021.

[7]  M. Fathi.  Quelques applications du transport optimal en analyse et en probabilités , 2019 .

[8]  Yin Tat Lee,et al.  Strong self-concordance and sampling , 2019, STOC.

[9]  Arnak S. Dalalyan,et al.  Further and stronger analogy between sampling and optimization: Langevin Monte Carlo and gradient descent , 2017, COLT.

[10]  Murat A. Erdogdu,et al.  Stochastic Runge-Kutta Accelerates Langevin Monte Carlo and Beyond , 2019, NeurIPS.

[11]  D. Kinderlehrer,et al.  THE VARIATIONAL FORMULATION OF THE FOKKER-PLANCK EQUATION , 1996 .

[12]  Tyler Maunu,et al.  Exponential ergodicity of mirror-Langevin diffusions , 2020, NeurIPS.

[13]  Kwangjun Ahn,et al.  Efficient constrained sampling via the mirror-Langevin algorithm , 2020, NeurIPS.

[14]  Hariharan Narayanan,et al.  Random walks on polytopes and an affine interior point method for linear programming , 2009, STOC '09.

[15]  Andre Wibisono,et al.  Sampling as optimization in the space of measures: The Langevin dynamics as a composite optimization problem , 2018, COLT.

[16]  Alain Durmus,et al.  High-dimensional Bayesian inference via the unadjusted Langevin algorithm , 2016, Bernoulli.

[17]  Mohamed-Jalal Fadili,et al.  Wasserstein Control of Mirror Langevin Monte Carlo , 2020, COLT.

[18]  Santosh S. Vempala,et al.  Rapid Convergence of the Unadjusted Langevin Algorithm: Isoperimetry Suffices , 2019, NeurIPS.