Speaker
Description
We introduce an AI-driven multi-agent system that automates the processing of Nuclear Resonant Scattering (NRS) measurements. Developed atop the Dr. Sai ecosystem, the system addresses two primary experimental techniques within the NRS domain: coherent scattering methods and inelastic scattering approaches for phonon studies. The agent is optimized for operation at the High Energy Photon Source (HEPS) in China. It wraps existing computational engines, including CONUSS for spectral modeling and PHOENIX for vibrational property extraction, into a unified analytical pipeline. Starting from raw instrument outputs, the assistant manages preprocessing steps such as reformatting detector streams, assembling input decks, and verifying parameter consistency. A standout capability is the orchestration of simultaneous analysis jobs, allowing researchers to test competing theoretical interpretations or model variations across large datasets without manual intervention. The design prioritizes workflow consistency, offering fine-grained control over adjustable parameters. Tasks that were previously error-prone and required substantial manual effort are now performed autonomously by the agent, significantly reducing both the risk of mistakes and the time spent on analysis. Graphical summaries of output results are generated automatically, aiding both real-time decision-making and post-experiment documentation. The system is now in its final deployment phase, with ongoing work focused on integration into routine experimental workflows at HEPS, where it is expected to shorten analysis cycles, enhance output uniformity, and support systematic comparative studies across the NRS user community.
