Research map / Probabilistic and scientific modeling / Variational and generative sampling
Deterministic Langevin Monte Carlo with Normalizing Flows for Bayesian Inference
Richard D.P. Grumitt, Biwei Dai, Uros Seljak
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| remove Metropolis-Hastings correction (ablation) | DLMC · hierarchical funnel d=101, sigma=5 | posterior quality | Hurt |
| remove normalizing-flow preconditioning (ablation) | DLMC · sparse logistic regression d=51 | convergence speed | Hurt |
| remove Langevin step, use NF-proposal MH only (ablation) | DLMC · funnel d=101, sigma=0.1 | convergence speed | Hurt |
| replace normalizing flow with kernel density estimation (PP-DLMC) (ablation) | DLMC · benchmark examples | posterior quality | Hurt |
| replace DLMC with SVGD (ablation) | particle methods · benchmark examples | posterior quality | Hurt |
| use parallel instead of serial DLMC | DLMC · hierarchical funnel d=101, sigma=5 | wall-clock time: 30 minutes | Improved |
| use DLMC instead of NUTS | sampler · hierarchical funnel d=101, sigma=5 | likelihood evaluations: 15,000 likelihood calls | Improved |
| parallel DLMC+HMC instead of NUTS | sampler · ill-conditioned Gaussian d=1000 | wall-clock evaluations: 10^5 -> 100 | Improved |
| serial DLMC+HMC versus NUTS | sampler · ill-conditioned Gaussian d=1000 | serial likelihood evaluations: 10^5 | No clear effect |
| DLMC alone without HMC | DLMC · ill-conditioned Gaussian d=1000 | bias | Hurt |
| upsample to 1000 instead of 5000 particles | DLMC · sparse logistic regression d=51 | computational cost | No clear effect |
| S-DLMC versus NUTS | sampler · Rosenbrock 32-d | wall-clock quality | No clear effect |
Closest papers by meaning
- Stochastic Normalizing Flows (NeurIPS 2020 · similarity 0.89)
- Sequential Controlled Langevin Diffusions (ICLR 2025 · similarity 0.87)
- Microcanonical Langevin Ensembles: Advancing the Sampling of Bayesian Neural Networks (ICLR 2025 · similarity 0.87)
- Preconditioned Langevin Dynamics with Score-based Generative Models for Infinite-Dimensional Linear Bayesian Inverse Problems (NeurIPS 2025 · similarity 0.86)
- Langevin Monte Carlo for strongly log-concave distributions: Randomized midpoint revisited (ICLR 2024 · similarity 0.86)
- To discretize continually: Mean shift interacting particle systems for Bayesian inference (NeurIPS 2026 · similarity 0.86)
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.
