Research map / Probabilistic and scientific modeling / Variational and generative sampling
De-Sequentialized Monte Carlo: a parallel-in-time particle smoother
Adrien Corenflos, Nicolas Chopin, Simo Särkkä
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 |
|---|---|---|---|
| use c-dSMC instead of bootstrap cSMC with backward sampling in particle Gibbs | particle Gibbs sampler · Particle Gibbs, theta-logistic model (nutria, T+1=120) | time for 10^5 samples (seconds): 4,083 -> 480 | Improved |
| use c-dSMC instead of cSMC with backward sampling | particle Gibbs sampler · Particle Gibbs, theta-logistic model (nutria, T+1=120) | trajectory update rate (%): 90 -> 85 | Hurt |
| use c-dSMC instead of cSMC with backward sampling | particle Gibbs sampler · Particle Gibbs, theta-logistic model (nutria, T+1=120) | parameter auto-correlation | No clear effect |
| use dSMC instead of sequential GPU FFBS | smoothing of Poisson state-space model · Fisher score, T=32 to 512 | running time | Improved |
| use dSMC instead of FFBS | smoothing of Poisson state-space model · Fisher score, T=32 to 512 | average relative error | Hurt |
| use rejection-resampling lazy dSMC instead of systematic-resampling dSMC and FFBS at large N (ablation) | dSMC · constrained random walk, sigma 0.3-0.5 | run time | Improved |
| use rejection-resampling lazy dSMC instead of systematic-resampling dSMC (ablation) | dSMC · constrained random walk, sigma 0.3-0.5 | variance of Fisher score estimate | No clear effect |
| add lazy resampling to dSMC at low particle count N (ablation) | dSMC · constrained random walk | run time | Hurt |
| raise number of time steps T for lazy dSMC versus FFBS | rs-dSMC · constrained random walk | relative performance | Improved |
| increase number of particles in dSMC on a GPU | dSMC · Fisher score, T=32 to 512 | parallelizable time steps | Hurt |
Closest papers by meaning
- Parallelizing MCMC Across the Sequence Length (NeurIPS 2025 · similarity 0.87)
- Cache Coherent Resampling for Efficient Test Time Scaling in LLM Reasoning via Adaptive Sequential Monte Carlo (ICML 2026 · similarity 0.86)
- Reverse Diffusion Sequential Monte Carlo Samplers (NeurIPS 2025 · similarity 0.86)
- Sequential Controlled Langevin Diffusions (ICLR 2025 · similarity 0.85)
- Online Variational Sequential Monte Carlo (ICML 2024 · similarity 0.85)
- Tuning Sequential Monte Carlo Samplers via Greedy Incremental Divergence Minimization (ICML 2025 · similarity 0.84)
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.
