LLMs are improving at a breathtaking speed: modern agents prepare entire manuscripts and start to prove non-trivial results. Are we entering an era of AI-assisted physics?
In this hackathon we will explore this question in a hands-on fashion, focusing on quantum physics. Talks will first introduce fundamentals of LLMs and then explain modern prompting techniques, agent orchestration and tooling like MCPs. Specialized scaffolding software for LLMs in a physics context will be introduced. Through our supporters, each participant will be provided with 50€ in API credits from OpenAI and 25$ from Anthropic as well as full access to the AI collaborator platform Lucien, which will enable hands-on experimentation in hacking sessions: we will explore proving mathematical results, carrying out numerical simulations, editing papers and ideating for new projects with LLMs, potentially on our own projects. (Note the focus of the hackathon will gravitate towards more using agents for research tasks like those mentioned instead of measurement data analysis with ML techniques, etc.) Lunch and coffee will be provided and give opportunity to network with other AI for science enthusiasts.
Sign-up: 30 spots are available. Participants should ideally have (quantum) physics research experience. Background knowledge about LLMs is not required, only motivation in using these new tools. If you're interested and don't exactly fit the profile, we encourage you to just apply!
Agenda
9:15 Doors open, Registration9:30 Opening & LogisticsTalk 'Introduction to LLMs & Agents' (Sirui Lu)Short Talk 'Responsible & Ethical Use of AI in Science' (Maximilian Lutz)Tutorial 'TeXRA' (Sirui Lu)11:00 Practical Session: experiment with applying agentic tools to your own research project13:00 Lunch (h-bar)14:00 Talk 'Lucien & MCP: Connecting AI to Your Research Tools' (Path Integral Institute, remote)14:30 Practical Sessions: continue to use agentic tools for your own research project | hack on creating skills/MCPs/agents/etc. (with team formation)17:15 (optional) Mini-tutorial 'Preparing Presentations with AI' (Sirui Lu)17:45 Dinner (Pizza)18:30 Experience sharing session | Flash presentations18:50 Closing Session19:00 EndA reunion event will be organized at a later date to gather further progress on the projects coming out of the hackathon. Participants are encouraged to open source their creations if plausible.
Supporters & Hosts
The Munich Center for Quantum Science and Technology PhD board supports graduate students with educational workshops and community events as part of Munich's quantum science and technology excellence cluster and makes possible venue and catering of the hackathon.
Path Integral Institute builds open platforms for AI-powered research and an AI for science community with initiatives like mcp.science or bench.science, and brings their organizational experience for AI for science hackathons (e.g. MCP x Quantum Science at QFARM, Stanford) with a talk about building Model Context Protocol and Skills for Agents.Anthropic and OpenAI, as providers of the state-of-the-art LLM families powering Claude and ChatGPT respectively, support every participant exploring physics with LLMs with sponsorship of 50€ in API credits each.
Lucien will provide their platform offering an agentic AI collaborator for free during the events to all participants.Sirui Lu is a PhD student at the Max Planck Institute of Quantum Optics, and is a LLM for physics afficionado: he develops TeXRA, supporting agent use with TeX, and researches LLMs for physics in his work. Maximilian Lutz is a fellow PhD student of Sirui, he is curious about pushing LLMs to do actual theoretical physics research and has experience organizing hackathons, workshops and retreats.
Sam Chen
sam@example.org
Quality Score
Quality Score
72/100
High confidence
Organiser16/20
Event Maturity14/20
Sponsors18/25
Participants12/20
Operations12/15
Why this score
Strong organiser track record
Returning event
Well-sponsored
Missing data
Prize details
Code of conduct
AI for Quantum Science Hackathon | Hackathon Radar