Constrained Code Generation with Discrete Diffusion

TL;DR

This paper introduces Constrained Diffusion for Code (CDC), a novel framework that integrates constraint satisfaction into discrete diffusion models to improve code generation quality and feasibility.

Contribution

CDC is a training-free neurosymbolic inference method that enhances discrete diffusion models with constraint-aware denoising operators for better code generation.

Findings

CDC improves constraint satisfaction in code generation benchmarks.

CDC outperforms baseline models in functional correctness, security, and syntax.

CDC achieves these improvements with less corrective computation and localized edits.

Abstract

Discrete diffusion models are a powerful, emerging paradigm for code generation. They construct programs through iterative refinement of partially corrupted token sequences and enable parallel token refinement. Importantly, this paradigm exposes a global program state at each denoising step, which provides a natural intervention point for enforcing program-level functionality and security constraints, guiding the generation before the final code is committed. Building on this observation, the paper introduces Constrained Diffusion for Code (CDC), a training-free neurosymbolic inference framework that integrates constraint satisfaction directly into the reverse denoising process. CDC augments the base discrete diffusion sampler with constraint-aware denoising operators that combine mathematical optimization with program analysis to identify constraint-relevant regions of the intermediate program state and locally adjust the denoising trajectory, steering generation toward feasible programs while remaining close to the base model. Across code generation benchmarks, CDC consistently improves constraint satisfaction in functional correctness, security, and even syntax, outperforming discrete diffusion and autoregressive baselines with less corrective computation and more localized edits.