Formalizing Numerical Analysis: An Agent Pipeline and Quality Audit Beyond Kernel Acceptance
About
Recent work has demonstrated that coding agents can formalize entire advanced mathematics textbooks in Lean 4, yet existing efforts concentrate on branches of mathematics already well-represented in mathlib and measure success solely through kernel acceptance. We address both limitations by applying a coding agent to formalize Numerical Methods for Ordinary Differential Equations, a textbook in numerical analysis that is largely absent from mathlib, stressing the agent's capacity to develop new theory from scratch. We further introduce a systematic, reproducible three-dimensional framework for evaluating the quality of agent-produced formalizations beyond compilation: semantic correctness, Mathlib reuse, and cross-file reuse via LLM-as-judge methods. Applying this framework to our own formalization and to the released outputs of RepoProver and M2F, we uncover recurring unfaithful formalization patterns, including incomplete multi-part statements, added weakening hypotheses, and parameter restrictions, that kernel acceptance entirely obscures. Our results suggest that compilation-based metrics substantially overstate formalization quality, and we provide a reproducible audit methodology to support more rigorous evaluation of future autoformalization systems.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Cross-file dependency reuse analysis | RepoProver | Formalization-reflected Edges1.16e+3 | 4 | |
| Cross-file dependency reuse analysis | OpenMath | Formalization-reflected Edges16 | 3 | |
| Formal Proof Completion | Butcher textbook Numerical Methods for Ordinary Differential Equations | Sorries0.00e+0 | 3 | |
| Library Overlap Analysis | Mathlib | Overlap4 | 3 | |
| Semantic Correctness Evaluation | OpenMath | Faithfulness54 | 2 | |
| Semantic Correctness Evaluation | M2F | Faithfulness73 | 2 | |
| Semantic Correctness Evaluation | RepoProver | Faithfulness82 | 2 |