🤖 AI Summary
This work addresses sampling from log-concave distributions under convex constraints and from composite target distributions. By employing an epigraph transformation, both problems are unified as approximate uniform sampling in a higher-dimensional space. The proposed algorithm integrates a proximal sampler, cutting-plane methods, and rejection sampling, relying solely on a separation oracle and a subgradient oracle—without requiring projections, reflections, or barrier functions. This approach yields unbiased and practical constrained sampling and leverages a dual epigraph structure to handle composite objectives in a unified framework. For the first time, mixing-time bounds are established for both problems under Rényi and χ² divergences, enabling efficient sampling even when the geometric structure of the constraint set is unknown.
📝 Abstract
We study two log-concave sampling problems: constrained sampling and composite sampling. First, we consider sampling from a target distribution with density proportional to $\exp(-f(x))$ supported on a convex set $K \subset \mathbb{R}^d$, where $f$ is convex. The main challenge is enforcing feasibility without degrading mixing. Using an epigraph transformation, we reduce this task to sampling from a nearly uniform distribution over a lifted convex set in $\mathbb{R}^{d+1}$. We then solve the lifted problem using a proximal sampler. Assuming only a separation oracle for $K$ and a subgradient oracle for $f$, we develop an implementation of the proximal sampler based on the cutting-plane method and rejection sampling. Unlike existing constrained samplers that rely on projection, reflection, barrier functions, or mirror maps, our approach enforces feasibility using only minimal oracle access, resulting in a practical and unbiased sampler without knowing the geometry of the constraint set. Second, we study composite sampling, where the target is proportional to $\exp(-f(x)-h(x))$ with closed and convex $f$ and $h$. This composite structure is standard in Bayesian inference with $f$ modeling data fidelity and $h$ encoding prior information. We reduce composite sampling via an epigraph lifting of $h$ to constrained sampling in $\mathbb{R}^{d+1}$, which allows direct application of the constrained sampling algorithm developed in the first part. This reduction results in a double epigraph lifting formulation in $\mathbb{R}^{d+2}$, on which we apply a proximal sampler. By keeping $f$ and $h$ separate, we further demonstrate how different combinations of oracle access (such as subgradient and proximal) can be leveraged to construct separation oracles for the lifted problem. For both sampling problems, we establish mixing time bounds measured in R\'enyi and $\chi^2$ divergences.