English

Generating from Discrete Distributions Using Diffusions: Insights from Random Constraint Satisfaction Problems

Machine Learning 2026-03-24 v1

Abstract

Generating data from discrete distributions is important for a number of application domains including text, tabular data, and genomic data. Several groups have recently used random kk-satisfiability (kk-SAT) as a synthetic benchmark for new generative techniques. In this paper, we show that fundamental insights from the theory of random constraint satisfaction problems have observable implications (sometime contradicting intuition) on the behavior of generative techniques on such benchmarks. More precisely, we study the problem of generating a uniformly random solution of a given (random) kk-SAT or kk-XORSAT formula. Among other findings, we observe that: (i)(i)~Continuous diffusions outperform masked discrete diffusions; (ii)(ii)~Learned diffusions can match the theoretical `ideal' accuracy; (iii)(iii)~Smart ordering of the variables can significantly improve accuracy, although not following popular heuristics.

Keywords

Cite

@article{arxiv.2603.20589,
  title  = {Generating from Discrete Distributions Using Diffusions: Insights from Random Constraint Satisfaction Problems},
  author = {Alankrita Bhatt and Mukur Gupta and Germain Kolossov and Andrea Montanari},
  journal= {arXiv preprint arXiv:2603.20589},
  year   = {2026}
}

Comments

39 pages; 15 figures

R2 v1 2026-07-01T11:30:55.164Z