Research map / Learning theory and algorithms / Combinatorial and search optimization
Learning Loop Invariants for Program Verification
Xujie Si, Hanjun Dai, Mukund Raghothaman, Mayur Naik, Le Song
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 counterexample-based reward (ablation) | Code2Inv (with attention) · 133 loop invariant benchmarks | instances solved: 94 -> 106 | Improved |
| add counterexample-based reward (ablation) | Code2Inv (with attention) · 133 loop invariant benchmarks | max #Z3 queries: 147K -> 276 | Improved |
| add attention over program memory (ablation) | Code2Inv (with CE) · 133 loop invariant benchmarks | instances solved: 95 -> 106 | Improved |
| add attention over program memory (ablation) | Code2Inv (with CE) · 133 loop invariant benchmarks | max #parameter updates: 337K -> 290K | Improved |
| graph neural network program embedding instead of LSTM over tokens (ablation) | Code2Inv · 133 loop invariant benchmarks | instances solved: 93 -> 106 | Improved |
| graph neural network program embedding instead of LSTM over tokens (ablation) | Code2Inv · 133 loop invariant benchmarks | max #parameter updates: 661K -> 290K | Improved |
| graph neural network program embedding instead of LSTM over tokens (ablation) | Code2Inv · 133 loop invariant benchmarks | max #Z3 queries: 32 -> 276 | Hurt |
| pre-train on augmented programs before solving new ones | Code2Inv · augmented programs with 1 or 5 confounding variables | Z3 queries | Improved |
Closest papers by meaning
- Towards General Loop Invariant Generation: A Benchmark of Programs with Memory Manipulation (NeurIPS 2024 · similarity 0.89)
- StepCodeReasoner: Aligning Code Reasoning with Stepwise Execution Traces via Reinforcement Learning (ICML 2026 · similarity 0.85)
- Process-Verified Reinforcement Learning for Theorem Proving via Lean (ICLR 2026 · similarity 0.85)
- Scaling Code-Assisted Chain-of-Thoughts and Instructions for Model Reasoning (NeurIPS 2025 · similarity 0.85)
- Chain of Execution Supervision Promotes General Reasoning in Large Language Models (NeurIPS 2025 · similarity 0.84)
- V1: Unifying Generation and Self-Verification for Parallel Reasoners (ICML 2026 · similarity 0.84)
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.
