Skip to content
Conference

From LLM Suggestions to Lean Proofs: Verified Redundant Constraints for MiniZinc

2026 · International Conference on Principles and Practice of Constraint Programming · pp. 17:1-17:19 · 0 citations · 36 references
Computer Science

TL;DR

This work presents a pipeline that combines LLM-based constraint generation with empirical evaluation and formal verification, and handles MiniZinc’s partial semantics by requiring the base model to be safe and separately proving that the proposed constraint is well-defined for all instances and solutions of the base model.

View source

Similar papers

Preprint Jul 2026

LLM-Guided Evolutionary Search for Constraint Model Reformulation to Improve Solver Efficiency

Combinatorial problems appear in numerous industrial applications. A common approach is to formulate these problems as declarative constraint models that can subsequently be compiled to and solved by a range of back-end solvers. Recent work shows that Large Language Models (LLMs) can produce correct models from natural language, but even a correct model can be expensive to solve because performance remains sensitive to modelling choices. In this work, we investigate whether LLMs can automate performance-oriented model reformulation. Inspired by Automatic Heuristic Design (AHD), we use an evolutionary framework in which an LLM proposes candidate reformulations that are verified and benchmarked against the user-defined baseline model. We compare AHD-adapted search strategies that control which prior attempts, instructions, and measured feedback enter each prompt. Existing retention strategies prioritize recency or performance, but do not explicitly diversify the context. To cover this gap, we introduce Profile-Diverse Retention (PDR), which applies Maximal Marginal Relevance (MMR) to instance-level runtime vectors to retain behaviourally diverse attempts. We systematically evaluate the strategies on eight CSPLib problems using validation-based final model selection. The results show that: (i) iterative reformulation can produce substantial held-out speedups; (ii) strategies that keep the retained context diverse outperform those that retain only recent or the fastest attempts; and (iii) validation-based selection improves the held-out speedup of every strategy.

Kostis Michailidis, Dimos Tsouros, Nguyen Dang et al. · 0 citations
Preprint Aug 2026

Improving Constraint Models with LLM Agents

This work introduces an agentic framework that reformulates a constraint model from an open-ended space and establishes correctness empirically rather than by construction, and demonstrates that autonomous agentic methods can support the improvement of constraint models.

Florentina Voboril, Stefan Szeider · 0 citations
2026

Factoring Learned Clauses

This paper revisits previous work on ER, which factors out repeated parts of learned clauses during conflict analysis, and explores how their original strategy benefits from 15 years of improvements in the state-of-the-art solver CaDiCaL, and proposes a new, less intrusive inprocessing approach based on factoring XOR and ITE gates from learned clauses globally.

Florian Pollitt, Zachary Battleman, Mathias Fleury et al. · 1 citation
Preprint Jul 2026

LeanCSP: A Framework for Certifying Constraint Reformulation and Solving in Lean

This work introduces a framework that addresses both verification levels in the Lean theorem prover, and can be used to prove formulation-level properties, such as equivalence, equisatisfiability, and the correctness of symmetry-breaking constraints, parametrically for entire problem families.

Pablo Manrique, Stefan Szeider · 0 citations
Preprint Jul 2026

Can LLMs Build a MaxSAT Solver from Papers? The CoreForge Experience

The experience suggests that LLMs can support solver implementation from papers, while requiring external validation, benchmarking, and human guidance, although performance remains below the best hand-engineered MaxSAT solvers.

Ruben Martins · 0 citations
2026

Beyond Core-Guided MaxSAT

A new proof system for MaxSAT is described, the Comparator Calculus, which models the inferential strategies used in core-guided MaxSAT solvers and two new MaxSAT algorithms are introduced: a core-guided one (CSimple) and one non-core-guided (CSat), which uses heuristics to construct new soft formulas and calls a SAT solver on a single soft formula.

Ilario Bonacina, J. Levy, Ion Mikel Liberal · 0 citations