Research map / Optimization / Nonconvex and smooth optimization
Stochastic Runge-Kutta Accelerates Langevin Monte Carlo and Beyond
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use the 1.5-order SRK-LD discretization instead of Euler-Maruyama | Overdamped Langevin sampling from strongly convex potentials · Strongly convex potentials | iterations to ε accuracy in W2: \tilde{\mathcal{O}}(d\epsilon^{-1}) -> \tilde{\mathcal{O}}(d\epsilon^{-2/3}) | Helped |
| Use the 1.0 SRK-ID discretization instead of Euler-Maruyama | Uniformly dissipative general Itô diffusions · Uniformly dissipative general Itô diffusions | iterations to ε accuracy in W2: \tilde{\mathcal{O}}(d \epsilon^{-2}) -> \tilde{\mathcal{O}}(d^{3/4} m^{2} \epsilon^{-1}) | Helped |
| Use SRK-LD rather than Euler-Maruyama at the same step size | Overdamped Langevin sampling · Gaussian mixture and Bayesian logistic regression | squared W2 asymptotic error | Helped |
| Use SRK-ID rather than Euler-Maruyama | Sampling from the stated non-convex potential · Non-convex potential, β=0.33, γ=0.5, d=1 | MSE | Helped |
| Use SRK-ID rather than Euler-Maruyama as sampling dimensionality increases | Sampling from the stated non-convex potential · Non-convex potential at higher dimensionality | MSE | Mixed |
| Use SRK-ID rather than Euler-Maruyama | Sampling from the stated non-convex potential · Non-convex potential | estimated squared W2 | No clear effect |
Closest papers by meaning
- Accelerating Langevin Monte Carlo via Efficient Stochastic Runge-Kutta Methods beyond Log-Concavity (ICML 2026 · similarity 0.95)
- On the Theory of Variance Reduction for Stochastic Gradient Monte Carlo (ICML 2018 · similarity 0.90)
- Langevin Monte Carlo for strongly log-concave distributions: Randomized midpoint revisited (ICLR 2024 · similarity 0.88)
- Provable Convergence and Limitations of Geometric Tempering for Langevin Dynamics (ICLR 2025 · similarity 0.87)
- Sequential Controlled Langevin Diffusions (ICLR 2025 · similarity 0.87)
- The Poisson Midpoint Method for Langevin Dynamics: Provably Efficient Discretization for Diffusion Models (NeurIPS 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.