Research map / Learning theory and algorithms / Combinatorial and search optimization
Massively Parallel Proof-Number Search for Impartial Games and Beyond
Tomáš Čížek, Martin Balko, Martin Schmid
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 child-ordering heuristic with parameters tuned for it (ablation) | PNS-PDFPN · Sprouts 47-spot, 1024 cores | speedup: 97.60 -> 332.97 | Improved |
| add child-ordering heuristic without retuning parameters (ablation) | PNS-PDFPN · Sprouts 47-spot, 1024 cores | speedup: 208.67 -> 169.24 | Hurt |
| add child-ordering heuristic without retuning parameters (ablation) | PNS-PDFPN · Sprouts 47-spot, 128 cores | speedup: 57.43 -> 45.76 | Hurt |
| add child-ordering heuristic without retuning parameters (ablation) | PNS-PDFPN · Sprouts 47-spot, 16 cores | speedup: 14.37 -> 10.15 | Hurt |
| add child-ordering heuristic with retuned parameters (ablation) | PNS-PDFPN · Sprouts 47-spot, 128 cores | speedup: 29.44 -> 87.02 | Improved |
| add retention of Grundy numbers in workers (ablation) | PNS-PNS · Sprouts 29-spot | speedup | Improved |
| add retention of proof and disproof numbers (PNS-DFPN instead of PNS-PNS) (ablation) | PNS-DFPN · Sprouts 29-spot | speedup | Improved |
| add sharing of Grundy numbers among workers (GN sync) (ablation) | PNS-DFPN · Sprouts 29-spot | speedup | Improved |
| add worker grouping to reduce synchronization overhead (ablation) | PNS-DFPN with GN sync · Sprouts 47-spot, above 512 cores | speedup | Improved |
| add second-level parallelization (threads per worker) (ablation) | PNS-DFPN with grouping · Sprouts 47-spot, 1024 cores | speedup: 21.35 -> 208.67 | Improved |
| use PNS-PDFPN instead of shared-memory PDFPN (ablation) | PDFPN · Sprouts 47-spot | speedup: 21.35 -> 208.67 | Improved |
| use SPOTS PNS instead of GLOP PNS | GLOP · Sprouts 29-spot | computation time: roughly 100-fold reduction | Improved |
Closest papers by meaning
- STP: Self-play LLM Theorem Provers with Iterative Conjecturing and Proving (ICML 2025 · similarity 0.83)
- HardcoreLogic: Challenging Large Reasoning Models with Long-tail Logic Puzzle Games (ICLR 2026 · similarity 0.83)
- MPS-Prover: Advancing Stepwise Theorem Proving by Multi-Perspective Search and Data Curation (NeurIPS 2025 · similarity 0.83)
- DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search (ICLR 2025 · similarity 0.83)
- FaithSieve: Fine-Grained Evaluation of Math Proofs with Faithful Formal Evidence (NeurIPS 2026 · similarity 0.82)
- 3D-Prover: Diversity Driven Theorem Proving With Determinantal Point Processes (NeurIPS 2025 · similarity 0.82)
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.
