Research map / Probabilistic and scientific modeling / Variational and generative sampling
Entropy-based adaptive Hamiltonian Monte Carlo
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 GSM rather than ESJD to adapt HMC (ablation) | HMC mass-matrix adaptation · Ill-conditioned Gaussian, d=100 | MinESS/sec: 122.3 (15.5) vs 0.1 (0.01); 753.8 (22.2) vs 0.1 (0.02); 570.0 (37.4) vs 0.6 (395.2) | Helped |
| Use GSM rather than L2HMC to adapt HMC (ablation) | HMC mass-matrix adaptation · Ill-conditioned Gaussian, d=100 | MinESS/sec: 122.3 (15.5) vs 0.1 (0.01); 753.8 (22.2) vs 0.1 (0.02); 570.0 (37.4) vs 0.1 (0.05) | Helped |
| Use GSM rather than ESJD to adapt HMC (ablation) | HMC mass-matrix adaptation · Correlated Gaussian, d=51 | MinESS/sec: 63.8 (3.9) vs 0.8 (1.6); 390.0 (5.0) vs 2.0 (5.4); 282.7 (7.8) vs 0.9 (3.7) | Helped |
| Use GSM rather than L2HMC to adapt HMC (ablation) | HMC mass-matrix adaptation · Correlated Gaussian, d=51 | MinESS/sec: 63.8 (3.9) vs 0.3 (0.1); 390.0 (5.0) vs 2.7 (2.3); 282.7 (7.8) vs 0.4 (0.9) | Helped |
| Use a tridiagonal rather than diagonal mass matrix (ablation) | Entropy-based HMC · Stochastic volatility, d=2519 | ESS/sec and R-hat | Helped |
| Use a nonlinear rather than linear transformation (ablation) | Entropy-based HMC · Banana-shaped target, d=2 | ESS | Helped |
| Use a linear transformation (ablation) | HMC on the Banana-shaped target · Banana-shaped target, d=2 | ESS | No clear effect |
| Use entropy-based adaptation rather than ESJD or L2HMC (ablation) | HMC adaptation · Anisotropic Gaussian, d=1000 | MinESS/sec | Helped |
| Use HMC with a moderate number of leapfrog steps rather than MALA | Entropy-based sampler · Stochastic volatility, d=2519 | ESS/sec and R-hat | Helped |
| Use the entropy-based sampler rather than NUTS | Bayesian logistic regression · Australian Credit and Caravan | MinESS/sec | Hurt |
| Use the entropy-based sampler rather than Riemann-Manifold samplers | Log-Gaussian Cox process · d=256 | MinESS/sec | Hurt |
| Use the entropy-based sampler rather than Riemann-Manifold samplers | Log-Gaussian Cox process · d=64 | MinESS/sec | Helped |
Closest papers by meaning
- On the Convergence of Hamiltonian Monte Carlo with Stochastic Gradients (ICML 2021 · similarity 0.92)
- A Gradient Based Strategy for Hamiltonian Monte Carlo Hyperparameter Optimization (ICML 2021 · similarity 0.91)
- Gradient-based Adaptive Markov Chain Monte Carlo (NeurIPS 2019 · similarity 0.91)
- Mixed Hamiltonian Monte Carlo for Mixed Discrete and Continuous Variables (NeurIPS 2020 · similarity 0.90)
- Metropolis Adjusted Microcanonical Hamiltonian Monte Carlo (NeurIPS 2025 · similarity 0.90)
- Stochastic Gradient Hamiltonian Monte Carlo Methods with Recursive Variance Reduction (NeurIPS 2019 · similarity 0.90)
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.