Research map / Probabilistic and scientific modeling / Variational and generative sampling
Poisson-Minibatching for Gibbs Sampling with Convergence Rate Guarantees
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 Poisson minibatching without an additional Metropolis-Hastings correction | Gibbs sampling · Potts model | marginal error | Helped |
| Increase λ from 0.1·L² through 1·L² to 5·L² (ablation) | Poisson-Gibbs · Potts model | marginal error convergence speed: 7, 28 and 132 average factors | Helped |
| Use a sufficiently large second minibatch for the Metropolis-Hastings correction (ablation) | DoubleMIN-Gibbs · Potts model | marginal error convergence: 10^4 · L^2 | Helped |
| Use Poisson minibatching with continuous-state sampling | Gibbs sampling · Continuous spin model | symmetric KL divergence | No clear effect |
| Use double Chebyshev approximation instead of fast inverse transform sampling | Poisson-Gibbs · Continuous spin model | symmetric KL divergence convergence | Helped |
| Use Chebyshev approximation rather than rejection sampling | Gibbs sampling · Continuous spin model | symmetric KL divergence per runtime: greater than 300 | Helped |
| Use Poisson minibatching with continuous-state sampling | Gibbs sampling · Continuous spin model | symmetric KL divergence per runtime | Helped |
| Use PGITS or PGDA instead of rejection sampling | Gibbs sampling · Truncated Gaussian mixture | density estimate accuracy: around 1000 | Helped |
Closest papers by meaning
- Minibatch Gibbs Sampling on Large Graphical Models (ICML 2018 · similarity 0.95)
- Multi-Marginal Couplings for Metropolis--Hastings (NeurIPS 2026 · similarity 0.87)
- Faster Sampling from Log-Concave Densities over Polytopes via Efficient Linear Solvers (ICLR 2024 · similarity 0.86)
- Diffusive Gibbs Sampling (ICML 2024 · similarity 0.86)
- Split Gibbs Discrete Diffusion Posterior Sampling (NeurIPS 2025 · similarity 0.86)
- Scalable Discrete Diffusion Samplers: Combinatorial Optimization and Statistical Physics (ICLR 2025 · 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.