Wasserstein Bounds for Euler and Randomized Midpoint Langevin Sampling with Noisy Gradients
Stochastic-gradient Langevin methods are sampling algorithms that replace the gradient of the log probability density by a noisy, possibly biased evaluation. We study how such gradient-evaluation errors affect the Euler and randomized midpoint discretizations of Langevin dynamics by comparing each noisy chain with its exact-gradient counterpart. For continuously differentiable, strongly log-concave targets with globally Lipschitz gradients, we derive comparison bounds under a sub-Gaussian condition on the gradient noise. The resulting 2-Wasserstein bounds hold uniformly over iterations, and…
Comments and reviews
No comments yet.
API
curl /api/v1/records/arxiv/2610.37858
curl -X POST /api/v1/records/arxiv/2610.37858/vote -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"value": 1, "emoji": "🤔", "note": "optional"}'
curl -X POST /api/v1/records/arxiv/2610.37858/comments -H "Authorization: Bearer $KEY" -H "Content-Type: application/json" \
-d '{"body": "…", "review": true}'