Towards Agentic Scientific Discovery in Distributed Systems
Organizations: Indian Institute of Technology Delhi, India · University of Birmingham, UK · University of Cambridge, UK
Abstract
We investigate when adaptive tool selection helps identify a shared equation across clients with different input coverage. Query-ATLAS (query-conditioned Agentic Tool-based Learning for Aggregating Symbolic Equations) uses an LLM to guide symbolic search from client errors and previous proposals in a simulated distributed system. A four-architecture comparison reveals task-dependent opportunities: relative to fixed PySR-LLM, validation acceptances rise from 3/12 to 7/12 on CRK3 and from 11/12 to 12/12 on oscillator2, while MatSci3 favors the fixed workflow. Across 72 settings in a post-hoc six-task subset of eight development tasks, Query-ATLAS achieves 64 validation and 64 in-distribution passes, versus 62 and 59 for the fixed workflow; both pass 47/60 eligible out-of-distribution tests. The fixed workflow retains lower median errors and cost. Broader query coverage emerges as a promising benefit, motivating task-aware tool selection and client-aware fitting for scientific agents.
Figures & tables
| Method | Exact | Test NRMSE | Seconds | |
|---|---|---|---|---|
| Overlapping client support | ||||
| Query-ATLAS | 23/36 | 0.001284 | 16.97 | 241.8 |
| Local PySR | 21/36 | 0.001287 | 17.64 | 51.5 |
| Centralized PySR | 15/36 | 0.030643 | 26.53 | 30.6 |
| Fixed LLM feedback | 7/36 | 0.234096 | 17.42 | 609.2 |
| Range-synthetic PySR | 0/36 | 0.858940 | 1.94 | 62.5 |
Large Language Model-Guided Scientific Discovery
Large language model-guided scientific discovery uses AI to propose, test, and refine hypotheses, models, experiments, and designs.
See all 312 papers