PreprintarXiv licencearXiv
Agentic programs: an emerging form of scientific software in computational materials science
Abstract
Computational materials science has traditionally delegated algorithmic tasks to computers while leaving scientific judgments to humans. We argue that recent LLM-based agent harnesses enable an emerging form of scientific software, agentic programs, that combine deterministic algorithms with bounded LLM-based judgment, task-specific verification, episodic maturation, and complete delegation in production. We illustrate this concept with DeMARS, an agentic program for constructing atomistic models from experimentally measured disordered crystal structures.
Related
- Recursive Improvement of a Differentiable Scientific Software Ecosystem
- An extensible software platform for automated operational and characterization testing of scientific cameras
- Reliable LLM-Generated Programs for High-Energy Physics Experiments through Graph-Grounded Software Knowledge
- Mined from Scientific Literature: Process Schemas for Atomic Layer Deposition and Etching in Materials Science
- FlowMLLab: An open-source framework for reproducible computational-fluid-dynamics and scientific-machine-learning experiments
BibTeX
@misc{lim2026agentic,
title = {{Agentic programs: an emerging form of scientific software in computational materials science}},
author = {Yunsung Lim and Haekwan Jeon and Jaesun Kim and Jisu Kim and Seungwu Han},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.00795},
eprint = {2609.00795},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.00795},
}
RIS
TY - UNPB
TI - Agentic programs: an emerging form of scientific software in computational materials science
AU - Yunsung Lim
AU - Haekwan Jeon
AU - Jaesun Kim
AU - Jisu Kim
AU - Seungwu Han
PY - 2026
DA - 2026/09/01
PB - arXiv
DO - 10.48550/arxiv.2609.00795
UR - https://arxiv.org/abs/2609.00795
AB - Computational materials science has traditionally delegated algorithmic tasks to computers while leaving scientific judgments to humans. We argue that recent LLM-based agent harnesses enable an emerging form of scientific software, agentic programs, that combine deterministic algorithms with bounded LLM-based judgment, task-specific verification, episodic maturation, and complete delegation in production. We illustrate this concept with DeMARS, an agentic program for constructing atomistic models from experimentally measured disordered crystal structures.
ER -