Improving Gradient-guided Nested Sampling for Posterior Inference

Pablo Lemos, Nikolay Malkin, Will Handley, Y. Bengio, Y. Hezaveh +1 more
12/1/2023

No official code found. Best implementation: Pablo-Lemos/GGNS (21 stars, last commit 2024-02-02).

Abstract

We present a performant, general-purpose gradient-guided nested sampling algorithm, ${\tt GGNS}$, combining the state of the art in differentiable programming, Hamiltonian slice sampling, clustering, mode separation, dynamic nested sampling, and parallelization. This unique combination allows ${\tt GGNS}$ to scale well with dimensionality and perform competitively on a variety of synthetic and real-world problems. We also show the potential of combining nested sampling with generative flow networks to obtain large amounts of high-quality samples from the posterior distribution. This combination leads to faster mode discovery and more accurate estimates of the partition function.

View on arXivView PDF

Code Implementations(1)

Pablo-Lemos/GGNSCommunity100%

Gradient Based Nested Sampling

210Python, Jupyter NotebookApr 21, 20232 years ago

Similar Papers

PolyChord: next-generation nested sampling

83% match2 repos1012015

Diffusion-based supervised learning of generative models for efficient sampling of multimodal distributions

78% match2025

Particle Guidance: non-I.I.D. Diverse Sampling with Diffusion Models

76% match1 repo772023

Towards Accurate Guided Diffusion Sampling through Symplectic Adjoint Method

76% match1 repo132023

Value Gradient Sampler: Sampling as Sequential Decision Making

76% match1 repo82025

Density-Level Backward Error for Low-NFE Generative Flows

76% match2026

Cite this paper

@article{lemos2023improving,
  title  = {Improving Gradient-guided Nested Sampling for Posterior Inference},
  author = {Pablo Lemos and Nikolay Malkin and Will Handley and Y. Bengio and Y. Hezaveh and Laurence Perreault-Levasseur},
  year   = {2023},
  eprint = {2312.03911},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/2312.03911},
  journal = {ICML 2023 2023}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2312.03911.svg)](https://codeofpaper.com/paper/2312.03911)

Discussion