Abstract: Discrete Gaussian samplers are used to sample integers from a discrete Gaussian distribution. Since this functionality is used in operations such as key generation, signing, or key ...

We propose Continuous-exploratory Discrete Langevin Sampler (cDLS), a hybrid gradient-based sampler for discrete distributions. cDLS exploits the gradient information to explore the continuous space ...

Sampling from probability distributions with known density functions (up to normalization) is a fundamental challenge across various scientific domains. From Bayesian uncertainty quantification to ...

「DQM」では「制約条件に相当する数式」として「linearパターン」と「quadraticパターン」がある。 ・linear： 各変数を「足し算する形」で制約条件を表現する式の場合。 ・quadratic： 変数同士を「掛け算する形」で制約条件を表現する式の場合。 DQMの数式 ...

This repository officially houses the PyTorch implementation of the paper titled "Learnable Sampler Distillation for Discrete Diffusion Moldes", which is presented at NeurIPS 2025 as a spotlight.