Research map / Large language models / Agents, code and math
LLM-Guided Loop Bound Generation for Program Termination Verification
Zan Gong, Biting Huang, Fei He
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add validation-feedback iteration to Loopy | Loopy (Gemini-2.5-Flash) · Termination benchmarks (171 programs) | programs solved: 125 -> 139 | Improved |
| add validation-feedback iteration to Loopy | Loopy (Gemini-2.5-Flash) · Termination benchmarks (171 programs) | avg time on solved (s): 140.8 -> 336.2 | Hurt |
| use loop bounds with LIFT workflow instead of ranking functions with feedback | LIFT vs Loopy+Iter · Termination benchmarks (171 programs) | programs solved: 139 -> 152 | Improved |
| use ICE-DT as invariant learner instead of LaM4Inv (ablation) | LIFT · Termination benchmarks (171 programs) | programs solved: 120 -> 152 | Improved |
| use ICE-DT as invariant learner instead of LEMUR (ablation) | LIFT · Termination benchmarks (171 programs) | programs solved: 100 -> 152 | Improved |
| add progressive generation (conjunctive then lexicographic) vs unified generation (ablation) | LIFT · Termination benchmarks (171 programs) | programs solved: 141 -> 152 | Improved |
| add diagnostic feedback (violation witness and timeout diagnosis) (ablation) | LIFT (Gemini thinking) · Termination benchmarks (171 programs) | programs solved: 138 -> 152 | Improved |
| use LLM-generated loop bounds with ICE-DT validation instead of ranking functions | LIFT vs PROTON 2.1 · Termination benchmarks (171 programs) | programs solved: 144 -> 152 | Improved |
| use a stronger LLM (Gemini-2.5-Flash vs DeepSeek-V3.1) | LIFT · Termination benchmarks (171 programs) | programs solved: 139 -> 152 | Improved |
Closest papers by meaning
- Towards General Loop Invariant Generation: A Benchmark of Programs with Memory Manipulation (NeurIPS 2024 · similarity 0.87)
- Lemur: Integrating Large Language Models in Automated Program Verification (ICLR 2024 · similarity 0.85)
- CLEVER: A Curated Benchmark for Formally Verified Code Generation (NeurIPS 2025 · similarity 0.85)
- Verified Code Transpilation with LLMs (NeurIPS 2024 · similarity 0.85)
- Evaluating Program Semantics Reasoning with Type Inference in System $F$ (NeurIPS 2025 · similarity 0.83)
- HierSVA: A Synthesis Pipeline, Dataset, and Benchmark for LLM-Driven Hierarchical Hardware Formal Verification (NeurIPS 2026 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
