GRAFT-ATHENA: Self-Improving Agentic Teams for Autonomous Discovery and Evolutionary Numerical Algorithms
Abstract
Language-model agents plan, execute, and evaluate scientific workflows. However, the problem--method relationships guiding their choices remain implicit, so plausible outputs can violate the assumptions of the scientific problem. We introduce Graph Reduction to Adaptive Factored Trees (GRAFT), which makes this problem-to-method map explicit as an expandable probabilistic structure of admissible problems, methods, and dependencies. Because this structure follows scientific semantics, knowledge from documentation, papers, and domain experts can be encoded, while factorization keeps the structure tractable. Semantic fingerprints of encoded problems measure their similarity, so expertise and experience guide related problems. GRAFT-guided workflows exceeded published expert baselines in physics-informed learning, recovered blood-rheology trends, predicted Apollo Command Module hypersonic loading within 1.8\% of experiment, and established machine-checked error bounds for regularized brain-flow reconstruction and a spectral physics-informed architecture. The representation retains evaluated outcomes and incorporates new methods, allowing scientific knowledge and experience to accumulate.
Related
- Improving the Last-Iterate Guarantees of Anytime Algorithms for Stochastic Monotone Variational Inequalities
- Support in the domain of numerical algorithms in the education of future secondary school mathematics teachers
- Real-Time Model Predictive Control Algorithms for Autonomous Spacecraft Guidance
- Randomized algorithms in numerical linear algebra for scientific computing
- Autonomous Structural Discovery Of Bioprocess Dynamics And Kinetics Via LLM-Driven Agents
BibTeX
@misc{toscano2026graftathena,
title = {{GRAFT-ATHENA: Self-Improving Agentic Teams for Autonomous Discovery and Evolutionary Numerical Algorithms}},
author = {Juan Diego Toscano and Zhaojie Chai and George Em Karniadakis},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2605.11117},
eprint = {2605.11117},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2605.11117},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - GRAFT-ATHENA: Self-Improving Agentic Teams for Autonomous Discovery and Evolutionary Numerical Algorithms
AU - Juan Diego Toscano
AU - Zhaojie Chai
AU - George Em Karniadakis
PY - 2026
DA - 2026/10/16
PB - arXiv
DO - 10.48550/arxiv.2605.11117
UR - https://arxiv.org/abs/2605.11117
AB - Language-model agents plan, execute, and evaluate scientific workflows. However, the problem--method relationships guiding their choices remain implicit, so plausible outputs can violate the assumptions of the scientific problem. We introduce Graph Reduction to Adaptive Factored Trees (GRAFT), which makes this problem-to-method map explicit as an expandable probabilistic structure of admissible problems, methods, and dependencies. Because this structure follows scientific semantics, knowledge from documentation, papers, and domain experts can be encoded, while factorization keeps the structure tractable. Semantic fingerprints of encoded problems measure their similarity, so expertise and experience guide related problems. GRAFT-guided workflows exceeded published expert baselines in physics-informed learning, recovered blood-rheology trends, predicted Apollo Command Module hypersonic loading within 1.8\% of experiment, and established machine-checked error bounds for regularized brain-flow reconstruction and a spectral physics-informed architecture. The representation retains evaluated outcomes and incorporates new methods, allowing scientific knowledge and experience to accumulate.
ER -