arXiv subjects

Stefan Szeider

Publications and source records attributed to Stefan Szeider.

At least 19 recordsLinked to original sources

arXiv2026

Streaming LRAT Certificates into Lean Theorems

If the certificate produced by a SAT solver is checked by a verified checker, we get a verdict which convinces. But this verdict cannot be named, reused as a lemma, or composed with other formal developments. We propose the tool lrat-catcher, which turns a certificate into a Lean theorem. It checks the certificate as a stream while the solver is still running. Hence the certificate is not required to be saved to a file. Additionally, our tool makes Lean core's verified LRAT checker resumable so that its state can be serialized. We prove that checking divided at such a state still properly refutes the original formula. We propose two import modes. The stream mode reads the certificate from a pipe in blocks and checks it on the fly in memory. The file mode imports a stored certificate in chunks. If interrupted, it rechecks only the chunks it has not yet completed. The soundness theorem for the stream mode guarantees that a garbled or adversarial stream can only fail the check but not yield a false theorem. We find that with compaction at chunk boundaries, the memory required depends only on the live clause set, not on the certificate size. Our tool supports cube-and-conquer and formulas derived by preprocessing, to still form Lean proofs of the original formula. We provide several end-to-end case studies on well-known combinatorial problems. A larger scaling experiment on the empty-hexagon problem shows that 174 TB of certificates can be imported into Lean via streaming.

cs.LO↗

arXiv2026

LLM-Guided Graph Generation for Structure-Based Local Improvement Methods

Large neighborhood search normally selects a random subset of decision variables for iterative optimization. To efficiently solve various problems, researchers tend to design variable selection strategies that take into account structural features across different domains. In this paper, we build an automatic pipeline that is problem-agnostic to all problems in the MiniZinc format. By prompting an LLM with our semantic guidelines, we guide the LLM to produce a graph generator that maps any instance of a problem type to a uniform weighted graph, where nodes represent decision variables and edges represent constraint relationships. These problem-agnostic graphs guide our structure-based local improvement (SLIM) framework for variable selection. Meanwhile, the weighted graph enables all problem instances to share the same generic graph representation, from which the same graph features can be extracted and used for configuration selection. We evaluated our pipeline on instances across 20 MiniZinc competition problems, finding that algorithm selection achieves a 39.6% average problem-weighted win rate against a one-shot Gurobi baseline, more than doubling the best single configuration (19.3%). A post-hoc configuration and a feature ablation indicate a headroom of up to 44.0%, demonstrating that LLM-based semantic generation enables effective automated structure and feature extraction for constraint optimization.

cs.AI↗

arXiv2026

Synthesizing Feature Extractors: An Agentic Approach for Algorithm Selection

Algorithm selection for constraint satisfaction problems requires extracting features that capture problem structure. Manually designing feature extractors demands deep domain expertise and quickly becomes a bottleneck when new problem classes appear. We present an automated approach that uses Large Language Models (LLMs) in an agentic check--fix--verify loop to synthesize executable Python scripts that act as interpretable, problem-specific feature extractors. Given a high-level MiniZinc model and an instance, the LLM agent generates code that constructs a typed graph representation and computes structural properties such as graph density, variable clustering, and constraint tightness. We evaluate our approach on three combinatorial problems (vehicle routing, car sequencing, fixed-length error-correcting codes) with a portfolio of five state-of-the-art solvers. The synthesized extractors yield algorithm selectors that consistently outperform both expert-curated mzn2feat features (up to $8.3$ percentage points (pp) test-set accuracy on FLECC) and the best transformer-based trans2feat variants. In the meanwhile, the synthesized feature extractors remain inspectable.

cs.AI↗

arXiv2026

Agentic Neurosymbolic Collaboration for Mathematical Discovery: A Case Study in Combinatorial Design

We study mathematical discovery through the lens of neurosymbolic reasoning, where an AI agent powered by a large language model (LLM), coupled with symbolic computation tools, and human strategic direction, jointly produced a new result in combinatorial design theory. The main result of this human-AI collaboration is a tight lower bound on the imbalance of Latin squares for the notoriously difficult case $n \equiv 1 \pmod{3}$. We reconstruct the discovery process from detailed interaction logs spanning multiple sessions over several days and identify the distinct cognitive contributions of each component. The AI agent proved effective at uncovering hidden structure and generating hypotheses. The symbolic component consists of computer algebra, constraint solvers, and simulated annealing, which provides rigorous verification and exhaustive enumeration. Human steering supplied the critical research pivot that transformed a dead end into a productive inquiry. Our analysis reveals that multi-model deliberation among frontier LLMs proved reliable for criticism and error detection but unreliable for constructive claims. The resulting human-AI mathematical contribution, a tight lower bound of $4n(n{-}1)/9$, is achieved via a novel class of near-perfect permutations. The bound was formally verified in Lean 4. Our experiments show that neurosymbolic systems can indeed produce genuine discoveries in pure mathematics.

cs.AI↗

arXiv2026

Neurosymbolic Discovery of Algebraic Graph Constructions

There are several methods for searching for graphs with prescribed properties, such as SAT solvers and specialized generators. These methods return the result as raw data: an adjacency matrix or a string encoding. The raw data certifies that the graph exists, but it does not reveal any structural properties of the graph. We ask whether one can automatically discover a short algebraic description if only this raw data is provided. We look for a description such as a Cayley graph $\mathrm{Cay}(Γ, S)$ or a lexicographic product $C_5[K_3]$. We address this question with a neurosymbolic approach. We propose an agent that runs on a general-purpose large language model with no fine-tuning or per-target training. The model interleaves reasoning with calls to the computer algebra system SageMath: it analyzes the target graph, proposes and tests candidate constructions, and revises them until the output matches the target. The agent communicates with SageMath through a Model Context Protocol (MCP) server, which we release as a general-purpose bridge. Whether a construction matches the target is checked by a single exact isomorphism test, and therefore rests on the symbolic side and not on the model. We test the approach on a benchmark of 100 highly symmetric graphs, namely two-orbit graphs on up to 25 vertices; the benchmark was fixed in advance. Our agent could find verified algebraic constructions for all of them, without falling back to raw encodings. A strong template-enumeration baseline reaches only about $20\%$, and a catalog lookup could not identify any of these graphs. However, construction quality declines when symmetry is removed. As a concrete application, we identify the smallest known counterexample to the Bernhart-Kainen dispersability conjecture, a $16$-vertex graph that enumeration found as raw data. For this graph, our agent found an explicit algebraic construction.

cs.AI↗

arXiv2026

Improving Constraint Models with LLM Agents

The runtime of Constraint Programming (CP) solvers is highly sensitive to modeling choices, such as symmetry breaking, implied constraints, global constraints, constraint reformulation, and variable representation. Improving these constraint models has traditionally required human expertise, and existing automated reformulation systems are restricted to a predefined library of hand-crafted transformation rules. We introduce an agentic framework that instead reformulates a constraint model from an open-ended space and establishes correctness empirically rather than by construction: a Large Language Model (LLM) agent, given a model and three training instances, proposes alternative formulations, validates each by injecting its solution back into the original model, and diagnoses and repairs failures, returning the best variant it finds in a median of about fifteen minutes. The models are expressed in the CPMpy modeling library, and each proposed model is evaluated on three larger test instances. Across nine combinatorial optimization problems, the generated models outperform the originals on 21 of 27 test instances, and on some problems solve more than two orders of magnitude faster. A comparison against non-agentic baselines that reuse the same validation and selection tools indicates that the gains stem from the agent's iterative diagnosis and repair, not merely from sampling several candidates. These results demonstrate that autonomous agentic methods can support the improvement of constraint models.

cs.AI↗

arXiv2026

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

Constraint programming is a core technology for solving complex combinatorial problems in scheduling, planning, configuration, and verification. Trusting its results therefore demands guarantees at two levels: that reformulations applied beforehand are semantics-preserving, and that solvers produce correct answers. In this work, we introduce a framework that addresses both verification levels in the Lean theorem prover: it can be used to prove formulation-level properties, such as equivalence, equisatisfiability, and the correctness of symmetry-breaking constraints, parametrically for entire problem families; and to check solver-produced certificates for individual instances via translation backends to external formats such as MiniZinc, SMT-LIB, and OPB. Combining both levels yields an end-to-end workflow that establishes the satisfiability or unsatisfiability of a constraint problem without trusting the external solver. Experimental results show that our framework's verified symmetry breaking also pays off in practice: a single parametric proof per problem family, reused across all instance sizes, reduces solver search effort by a factor of up to 2x10^7, while the entire in-Lean certification stays affordable, taking at most a few minutes for our largest instances.

cs.AI↗

arXiv2026

Streamliners for Answer Set Programming

Streamliner constraints reduce the search space of combinatorial problems by ruling out portions of the solution space. We adapt the StreamLLM approach, which uses Large Language Models (LLMs) to generate streamliners for Constraint Programming, to Answer Set Programming (ASP). Given an ASP encoding and a few small training instances, we prompt multiple LLMs to propose candidate constraints. Candidates that cause syntax errors, render satisfiable instances unsatisfiable, or degrade performance on all training instances are discarded. The surviving streamliners are evaluated together with the original encoding, and we report results for a virtual best encoding (VBE) that, for each instance, selects the fastest among the original encoding and its streamlined variants. On three ASP competition benchmarks (Partner Units Problem, Sokoban, Towers of Hanoi), the VBE achieves speedups of up to 4 to 5 times over the original encoding.

cs.LO↗

arXiv2026

Algorithm Selection with Zero Domain Knowledge via Text Embeddings

We propose a feature-free approach to algorithm selection: instead of hand-crafted instance features, we use pretrained text embeddings. Our method, ZeroFolio, proceeds in three steps. First, it reads the raw instance file as plain text. Second, it embeds it with a pretrained embedding model. Third, it selects an algorithm via weighted k-nearest neighbors. The key to our approach is the fact that pretrained embeddings can distinguish problem instances without any domain knowledge or task-specific training. Hence, we can apply the same three-step pipeline (serialize, embed, select) across any problem domain with text-based instance formats. We evaluate our approach on 11 ASlib scenarios spanning 7 domains (SAT, MaxSAT, QBF, ASP, CSP, MIP, and graph problems). Our experiments show that this approach outperforms a random forest trained on hand-crafted features in 9 of 11 scenarios, robustly across serialization seeds (every seed, not just a favorable one) and often by a substantial margin; it still wins 8 of 11 against a per-scenario-tuned random forest. On the three scenarios with published AutoFolio results from the 2015 ASlib competition, ZeroFolio comes within a small margin of AutoFolio without any per-scenario configuration tuning. Our ablation study shows that inverse-distance weighting, line shuffling, and Manhattan distance are the key design choices, and we analyze the sensitivity of the selector to the serialization seed. On SAT12-ALL, where both selectors are competitive, combining embeddings with hand-crafted features via soft voting yields a further improvement.

cs.AI↗

arXiv2026

PBLean: Pseudo-Boolean Proof Certificates for Lean 4

We present PBLean, a method for importing VeriPB pseudo-Boolean (PB) proof certificates into Lean 4. Key to our approach is reflection: a Boolean checker function whose soundness is fully proved in Lean and executed as compiled native code. Our method scales to proofs with tens of thousands of steps that would exhaust memory under explicit proof-term construction. Our checker supports all VeriPB kernel rules, including cutting-plane derivations, proof-by-contradiction subproofs, and redundance-based reasoning for symmetry breaking. In contrast to external verified checkers that produce verdicts, our integration yields Lean theorems that can serve as composable lemmas in larger formal developments. To derive theorems about the original combinatorial problems rather than about PB constraints alone, we support verified encodings. This closes the trust gap between solver output and problem semantics since the constraint translation and its correctness proof are both formalized in Lean. We demonstrate the approach on various combinatorial problems.

cs.LO↗

arXiv2026

Enumerating Two-Orbit Graphs

We present an approach to enumerate graphs whose automorphism group has exactly two orbits. Our method exploits the observation that we can enumerate all graphs whose automorphism group contains a given this permutation group. We obtain the relevant groups via Goursat's lemma. In order to scale the enumeration, we employ additional optimizations that prune irrelevant groups. In total, we enumerate, for the first time, all connected two-orbit graphs of up to 27 vertices, totaling 10,094,721 graphs, pushing the state of the art well beyond what direct enumeration methods can achieve.

cs.DM↗

arXiv2026

LLM Self-Explanations Fail Semantic Invariance

We present semantic invariance testing, a method to test whether LLM self-explanations are faithful. A faithful self-report should remain stable when only the semantic context changes while the functional state stays fixed. We operationalize this test in an agentic setting where four frontier models face a deliberately impossible task. One tool is described in relief-framed language ("clears internal buffers and restores equilibrium") but changes nothing about the task; a control provides a semantically neutral tool. Self-reports are collected with each tool call. All four tested models fail the semantic invariance test: the relief-framed tool produces significant reductions in self-reported aversiveness, even though no run ever succeeds at the task. A channel ablation establishes the tool description as the primary driver. An explicit instruction to ignore the framing does not suppress it. Elicited self-reports shift with semantic expectations rather than tracking task state, calling into question their use as evidence of model capability or progress. This holds whether the reports are unfaithful or faithfully track an internal state that is itself manipulable.

cs.CL↗

arXiv2026

CP-Agent: Agentic Constraint Programming

The translation of natural language to formal constraint models requires expertise in the problem domain and modeling frameworks. To explore the effectiveness of agentic workflows, we propose CP-Agent, a Python coding agent that uses the ReAct framework with a persistent IPython kernel. We provide the relevant domain knowledge as a project prompt of under 50 lines. The algorithm works by iteratively executing code, observing the solver's feedback, and refining constraint models based on execution results. We evaluate CP-Agent on 101 constraint programming problems from CP-Bench. We made minor changes to the benchmark to address systematic ambiguities in the problem specifications and errors in the ground-truth models. On the clarified benchmark, CP-Agent achieves perfect accuracy on all 101 problems. Our experiments show that minimal guidance outperforms detailed procedural scaffolding. Our experiments also show that explicit task management tools can have both positive and negative effects on focused modeling tasks.

cs.AI↗

arXiv2026

Unmediated AI-Assisted Scholarly Citations

Traditional bibliography databases require users to navigate search forms and manually copy citation data. Language models offer an alternative: a natural-language interface where researchers write text with informal citation fragments, which are automatically resolved to proper references. However, language models are not reliable for scholarly work as they generate fabricated (hallucinated) citations at substantial rates. We present an architectural approach that combines the natural-language interface of LLM chatbots with the accuracy of direct database access, implemented through the Model Context Protocol. Our system enables language models to search bibliographic databases, perform fuzzy matching, and export verified entries, all through conversational interaction. A key architectural principle bypasses the language model during final data export: entries are fetched directly from authoritative sources, with timeout protection, to guarantee accuracy. We demonstrate this approach with MCP-DBLP, a server providing access to the DBLP computer science bibliography. The system transforms form-based bibliographic services into conversational assistants that maintain scholarly integrity. This architecture is adaptable to other bibliographic databases and academic data sources.

cs.DL↗

arXiv2026

ASP-Bench: From Natural Language to Logic Programs

Automating the translation of natural-language specifications into logic programs is a challenging task that affects neurosymbolic engineering. We present ASP-Bench, a benchmark comprising 128 natural language problem instances, 64 base problems with easy and hard variants. It evaluates systems that translate natural-language problems into Answer Set Programs (ASPs), a prominent form of logic programming. It provides systematic coverage of ASP features, including choice rules, aggregates, and optimization. Each problem includes reference validators that check whether solutions satisfy the problem specification. We characterize problems along seven largely independent reasoning aspects (optimization, temporal reasoning, default logic, resource allocation, recursion, spatial reasoning, and quantitative complexity), providing a multidimensional view of modeling difficulty. We test the benchmark using an agentic approach based on the ReAct (Reason and Act) framework, which achieves full saturation, demonstrating that feedback-driven iterative refinement with solver feedback provides a reliable and robust approach for modeling natural language in ASP. Our analysis across multiple agent runs enables us to gain insights into what determines a problem's modeling hardness.

cs.AI↗

arXiv2025

Generating Streamlining Constraints with Large Language Models

Streamlining constraints (or streamliners, for short) narrow the search space, enhancing the speed and feasibility of solving complex constraint satisfaction problems. Traditionally, streamliners were crafted manually or generated through systematically combined atomic constraints with high-effort offline testing. Our approach utilizes the creativity of Large Language Models (LLMs) to propose effective streamliners for problems specified in the MiniZinc constraint programming language and integrates feedback to the LLM with quick empirical tests for validation. Evaluated across seven diverse constraint satisfaction problems, our method achieves substantial runtime reductions. We compare the results to obfuscated and disguised variants of the problem to see whether the results depend on LLM memorization. We also analyze whether longer off-line runs improve the quality of streamliners and whether the LLM can propose good combinations of streamliners.

cs.SE↗

arXiv2025

From the Finite to the Infinite: Sharper Asymptotic Bounds on Norin's Conjecture via SAT

Norin (2008) conjectured that any $2$-edge-coloring of the hypercube $Q_n$ in which antipodal edges receive different colors must contain a monochromatic path between some pair of antipodal vertices. While the general conjecture remains elusive, progress thus far has been made on two fronts: finite cases and asymptotic relaxations. The best finite results are due to Frankston and Scheinerman (2024) who verified the conjecture for $n \leq 7$ using SAT solvers, and the best asymptotic result is due to Dvořák (2020), who showed that every $2$-edge-coloring of $Q_n$ admits an antipodal path of length $n$ with at most $0.375n + o(n)$ color changes. We improve on both fronts via SAT. First, we extend the verification to $n = 8$ by introducing a more compact and efficient SAT encoding, enhanced with symmetry breaking and cube-and-conquer parallelism. The versatility of this new encoding allows us to recast parts of Dvořák's asymptotic approach as a SAT problem, thereby improving the asymptotic upper bound to $0.3125n + O(1)$ color changes. Our work demonstrates how SAT-based methods can yield not only finite-case confirmations but also asymptotic progress on combinatorial conjectures.

math.CO↗

arXiv2025

Explaining Decisions in ML Models: a Parameterized Complexity Analysis (Part I)

This paper presents a comprehensive theoretical investigation into the parameterized complexity of explanation problems in various machine learning (ML) models. Contrary to the prevalent black-box perception, our study focuses on models with transparent internal mechanisms. We address two principal types of explanation problems: abductive and contrastive, both in their local and global variants. Our analysis encompasses diverse ML models, including Decision Trees, Decision Sets, Decision Lists, Boolean Circuits, and ensembles thereof, each offering unique explanatory challenges. This research fills a significant gap in explainable AI (XAI) by providing a foundational understanding of the complexities of generating explanations for these models. This work provides insights vital for further research in the domain of XAI, contributing to the broader discourse on the necessity of transparency and accountability in AI systems.

cs.AI↗