Conditional Diffusion Guidance under Hard Constraint: A Stochastic Analysis Approach
Zhengyi Guo, Wenpin Tang, Renyuan Xu · 2026
Read Conditional Diffusion Guidance under Hard Constraint: A Stochastic Analysis Approach by Zhengyi Guo, Wenpin Tang, Renyuan Xu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-05533.txt.
Description
Diffusion models can generate samples that fit a pattern, but forcing every sample to meet a hard constraint is a stricter demand. Zhengyi Guo, Wenpin Tang, and Renyuan Xu study conditional generation where outputs must satisfy a set event with certainty, as in safety-critical work and rare-event simulation. Soft guidance based on rewards cannot promise that, so they build on the probabilistic roots of diffusion models.
Their framework uses Doob's h-transform to add an explicit drift correction to a pretrained model, leaving the trained score network untouched. Two new learning algorithms estimate the needed correction from the model's own trajectories, using losses based on martingales and their covariation. The authors prove error bounds in total variation and Wasserstein distance that account for score and guidance errors. Numerical tests show the methods enforce hard constraints and produce rare-event samples well. Code for the experiments is shared online.
Catalog record
| CREATOR | Zhengyi Guo, Wenpin Tang, Renyuan Xu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 99 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-05533