Solving Physics Olympiad via Reinforcement Learning on Physics Simulators

A recipe to convert simulators into scalable data generators.

Solving Physics Olympiad via Reinforcement Learning on Physics Simulators Solving Physics Olympiad via Reinforcement Learning on Physics Simulators Mihir Prabhudesai *^   Aryan Satpathy *^   Yangmin Li ^   Zheyang Qin ^   Nikash Bhardwaj   Amir Zadeh λ   Chuan Li λ   Katerina Fragkiadaki   Deepak Pathak Accepted at ICML 2026 * Project co-leads ·  ^ Core contributors ·  λ Lambda Video explanation Paper arXiv 𝕏 Twitter Code --> --> --> --> --> --> --> scene : --> name : "Fixed Pulley System..." --> entities : --> name : "

saved by

• Anonymous Bat

related reading

• DeepSeek-R1arxiv.org
• [2510.13786] The Art of Scaling Reinforcement Learning Compute for LLMsarxiv.org
• Learning to reason with LLMs | OpenAIopenai.com
• State of RL for reasoning LLMs | A. Weersaweers.de
• the-illusion-of-thinking.pdfml-site.cdn-apple.com
• How to reason from first principles – Casey Handmer's blogcaseyhandmer.wordpress.com
• physics-IQ-benchmark/README.md at main · google-deepmind/physics-IQ-benchmarkgithub.com
• [2602.19362] LLMs Can Learn to Reason Via Off-Policy RLarxiv.org
• Emergent Dexterity via Diverse Resets and Large-Scale Reinforcement Learningarxiv.org
• RL on Incorrect Synthetic Data Scales the Efficiency of LLM Math Reasoning by Eight-Foldarxiv.org
• Reasoning in General Domains without Verifiersarxiv.org
• The State of Reinforcement Learning for LLM Reasoningmagazine.sebastianraschka.com