GRACE: an Agentic AI for Particle Physics Experiment Design and Simulation
arXiv:2602.15039v1 Announce Type: cross Abstract: We present GRACE, a simulation-native agent for autonomous experimental design in high-energy and nuclear physics. Given multimodal input in the form of a natural-language prompt or a published experimental paper, the agent extracts a structured representation of the experiment, constructs a runnable toy simulation, and autonomously explores design modifications using first-principles Monte Carlo methods. Unlike agentic systems focused on operational control or execution of predefined procedures, GRACE addresses the upstream problem of experimental design: proposing non-obvious modifications to detector geometry, materials, and configurations that improve physics performance under physical and practical constraints. The agent evaluates candidate designs through repeated simulation, physics-motivated utility functions, and budget-aware escalation from fast parametric models to full Geant4 simulations, while maintaining strict reproducib
arXiv:2602.15039v1 Announce Type: cross Abstract: We present GRACE, a simulation-native agent for autonomous experimental design in high-energy and nuclear physics. Given multimodal input in the form of a natural-language prompt or a published experimental paper, the agent extracts a structured representation of the experiment, constructs a runnable toy simulation, and autonomously explores design modifications using first-principles Monte Carlo methods. Unlike agentic systems focused on operational control or execution of predefined procedures, GRACE addresses the upstream problem of experimental design: proposing non-obvious modifications to detector geometry, materials, and configurations that improve physics performance under physical and practical constraints. The agent evaluates candidate designs through repeated simulation, physics-motivated utility functions, and budget-aware escalation from fast parametric models to full Geant4 simulations, while maintaining strict reproducibility and provenance tracking. We demonstrate the framework on historical experimental setups, showing that the agent can identify optimization directions that align with known upgrade priorities, using only baseline simulation inputs. We also conducted a benchmark in which the agent identified the setup and proposed improvements from a suite of natural language prompts, with some supplied with a relevant physics research paper, of varying high energy physics (HEP) problem settings. This work establishes experimental design as a constrained search problem under physical law and introduces a new benchmark for autonomous, simulation-driven scientific reasoning in complex instruments.
Executive Summary
The article introduces GRACE, an agentic AI for autonomous experimental design in high-energy and nuclear physics. GRACE uses natural-language prompts or published papers to extract experiment representations, construct simulations, and explore design modifications. The agent evaluates candidate designs through simulation, physics-motivated utility functions, and budget-aware escalation, maintaining reproducibility and provenance tracking. Demonstrated on historical setups, GRACE identifies optimization directions aligning with known upgrade priorities and proposes improvements from natural language prompts. This work establishes experimental design as a constrained search problem under physical law, introducing a new benchmark for autonomous scientific reasoning.
Key Points
- ▸ GRACE is an AI agent for autonomous experimental design in particle physics
- ▸ The agent uses natural-language prompts or published papers as input
- ▸ GRACE evaluates candidate designs through simulation and physics-motivated utility functions
Merits
Autonomous Design Optimization
GRACE's ability to propose non-obvious modifications to detector geometry, materials, and configurations can lead to improved physics performance and efficiency
Demerits
Dependence on Simulation Accuracy
The reliability of GRACE's design proposals depends on the accuracy of the underlying simulations, which may be limited by computational resources or physical models
Expert Commentary
The development of GRACE represents a significant step forward in the application of AI to scientific research. By automating the experimental design process, GRACE has the potential to accelerate discovery and reduce costs in particle physics research. However, the reliability and interpretability of GRACE's design proposals will be crucial to its adoption and impact. As the use of AI in scientific research continues to grow, it is essential to address the challenges of explainability, transparency, and reproducibility in AI-driven discovery.
Recommendations
- ✓ Further research on the interpretability and transparency of GRACE's design proposals
- ✓ Investigation into the potential applications of GRACE in other fields of scientific research
