Skip to content
DraconiticResearch
Search Sources API
203,234 studies31 sourcesUpdated 46 min ago
PreprintCC BY-NC-ND 4.0arXiv

Outram Park Part II: Agentic Verification, Validation and Documentation as a Human Review Gate for Nuclear Data and Neutronics Software

Abstract

NJOY21 is part of a modernisation effort for NJOY2016, but mostly consists of C++ wrappers. Agentic coding represents a unique opporunity for NJOY modernisation in Rust, but agents require human expert supervision, which may be in short supply. Moreover, expert human review is the bottleneck in agentic software development as human resources are limited in this context. This work describes a methodology for how a human manager of AI agents can reasonably translate and refactor scientific software with limited domain expertise using principles borrowed from nuclear quality assurance documents. Moreover, it presents a methodology for when to allocate limited human resources for reviewing the agentically generated code. This methodology centres on agentic V&V and documentation, and it is demonstrated via construction of Monte Carlo web demos for the Rust fork of NJOY an Outram Monte Carlo (OutramMC), Rust ports of NJOY2016 and OpenMC, in Outram Park. In this work, an integrated pipeline for the Rust port of NJOY and OutramMC was demonstrated in web demos, and code to code verification and validation was performed with ICSBEP benchmarks for Godiva, HST009 and LCT008. ACE files produced with Outram Park's NJOY port were identical for more than 30 ACE files. These web demos also served as part of agentic documentation, which prepared the codebase for human review.

Related

  1. Agentic Porting, Construction and Initial Verification and Validation of Libraries within the Open Source Unified TRAnsient Multi-Phase Advanced Reactor simulation Kit (Outram Park) Part I: A Software Engineering Log for Thermal Hydraulics
    Theodore Kay Chen Ong, Sicong Xiao
    arXivPreprintCC BY-NC-ND 4.0Full textComputational physicsarXiv
  2. Agentic programs: an emerging form of scientific software in computational materials science
    Yunsung Lim, Haekwan Jeon, Jaesun Kim +2
    arXivPreprintarXiv licenceFull textCondensed matterarXiv
  3. QBugLM: An Agentic Benchmarking Framework for LLM-based Quantum Software Debugging
    An B. B. Pham, Hoa T. Nguyen, Muhammad Usman
    arXivPreprintCC BY 4.0Full textQuantum physics & quantum informationarXiv
  4. An Open-Source Hardware and Software Toolkit to Enable Agentic RHEED-Guided Thin-Film Synthesis
    Asraful Haque, Christopher M. Rouleau, Rama K. Vasudevan +1
    arXivPreprintCC BY 4.0Full textCondensed matterarXiv
  5. LatticeMind: A Validation-Gated Agentic Framework for Density Functional Theory Workflows
    Nathaniel M. Vegh, Yu Zhu, Raphaël J. Prentki +3
    arXivPreprintarXiv licenceFull textComputational physicsCondensed matterarXiv
BibTeX
@misc{ong2026outram,
  title = {{Outram Park Part II: Agentic Verification, Validation and Documentation as a Human Review Gate for Nuclear Data and Neutronics Software}},
  author = {Theodore Kay Chen Ong and SiCong Xiao},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2609.15258},
  eprint = {2609.15258},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.15258},
  note = {CC BY-NC-ND 4.0},
}
RIS
TY  - UNPB
TI  - Outram Park Part II: Agentic Verification, Validation and Documentation as a Human Review Gate for Nuclear Data and Neutronics Software
AU  - Theodore Kay Chen Ong
AU  - SiCong Xiao
PY  - 2026
DA  - 2026/10/19
PB  - arXiv
DO  - 10.48550/arxiv.2609.15258
UR  - https://arxiv.org/abs/2609.15258
AB  - NJOY21 is part of a modernisation effort for NJOY2016, but mostly consists of C++ wrappers. Agentic coding represents a unique opporunity for NJOY modernisation in Rust, but agents require human expert supervision, which may be in short supply. Moreover, expert human review is the bottleneck in agentic software development as human resources are limited in this context. This work describes a methodology for how a human manager of AI agents can reasonably translate and refactor scientific software with limited domain expertise using principles borrowed from nuclear quality assurance documents. Moreover, it presents a methodology for when to allocate limited human resources for reviewing the agentically generated code. This methodology centres on agentic V&V and documentation, and it is demonstrated via construction of Monte Carlo web demos for the Rust fork of NJOY an Outram Monte Carlo (OutramMC), Rust ports of NJOY2016 and OpenMC, in Outram Park. In this work, an integrated pipeline for the Rust port of NJOY and OutramMC was demonstrated in web demos, and code to code verification and validation was performed with ICSBEP benchmarks for Godiva, HST009 and LCT008. ACE files produced with Outram Park's NJOY port were identical for more than 30 ACE files. These web demos also served as part of agentic documentation, which prepared the codebase for human review.
ER  -