Selecting Optimal Variable Order in Autoregressive Ising Models
Shiba Biswal, Marc Vuffray, Andrey Y. Lokhov · 2026
Read Selecting Optimal Variable Order in Autoregressive Ising Models by Shiba Biswal, Marc Vuffray, Andrey Y. Lokhov (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-2602-20394.txt.
Description
Shiba Biswal and two co-authors improve autoregressive sampling by choosing a better variable order. Autoregressive models sample tractably from learned distributions, but their quality depends heavily on the ordering used to factorize the distribution, through the complexity of the resulting conditionals. The authors first learn the Markov random field behind the data, then use its graph structure to build optimized orderings.
On two-dimensional image-like models, a structure-aware ordering narrows the conditioning sets, cutting model complexity. Numerical tests on Ising models with discrete data show graph-informed orderings give higher-fidelity samples than naive orderings. The lesson is that ordering is not a minor implementation detail: it shapes the difficulty of every conditional the model must learn. Structure-aware ordering thus offers a principled way to boost sample quality without changing the model class.
Catalog record
| CREATOR | Shiba Biswal, Marc Vuffray, Andrey Y. Lokhov |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-20394