Research map / Probabilistic and scientific modeling / Variational and generative sampling
Mixed Hamiltonian Monte Carlo for Mixed Discrete and Continuous Variables
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 |
|---|---|---|---|
| Evolve discrete and continuous variables in tandem with frequent discrete proposals | M-HMC · 24D GMM | MRESS | Helped |
| Evolve discrete and continuous variables in tandem with frequent discrete proposals | M-HMC · Variable selection in BLR | MRESS | Helped |
| Evolve discrete and continuous variables in tandem with frequent discrete proposals | M-HMC · AP CTM, 17 documents where samplers agree | MRESS: 57.32 times larger than Gibbs; 8.76 times larger than NwG; 8.65 times larger than HwG | Helped |
| Use M-HMC rather than DHMC's one-dimensional discrete-state embedding | mixed-support sampling · 1D GMM with reordered component means | sampling coverage: M-HMC works well with 10^6 samples; DHMC failed even after 10^7 samples | Helped |
| Increase the number of discrete updates L while holding travel time T fixed (ablation) | M-HMC · Variable selection in BLR | MRESS | Helped |
| Increase L beyond a suitable trajectory length (ablation) | M-HMC · Variable selection in BLR | MRESS | Mixed |
| Set n_D=1 rather than batching more discrete sites per update, holding n_D L constant (ablation) | M-HMC · Variable selection in BLR | MRESS | Helped |
| Use naive MH updates inside HMC instead of the M-HMC correction | mixed HMC on 1D GMM · 1D GMM | sampling accuracy | Hurt |
| Use a non-identity mass matrix for the continuous kinetic energy | M-HMC · AP CTM | sampling performance | Helped |
Closest papers by meaning
- Entropy-based adaptive Hamiltonian Monte Carlo (NeurIPS 2021 · similarity 0.90)
- On the Convergence of Hamiltonian Monte Carlo with Stochastic Gradients (ICML 2021 · similarity 0.89)
- Pseudo-Extended Markov chain Monte Carlo (NeurIPS 2019 · similarity 0.88)
- Metropolis Adjusted Microcanonical Hamiltonian Monte Carlo (NeurIPS 2025 · similarity 0.88)
- Automatically marginalized MCMC in probabilistic programming (ICML 2023 · similarity 0.88)
- Hamiltonian Monte Carlo Inference of Marginalized Linear Mixed-Effects Models (NeurIPS 2024 · 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.