Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

An Agentic Workflow for Legacy HPC Modernization: Converting the Two-Electron-Integral Core of GAMESS

Source

arXiv

arxiv.org/abs/2608.12249v1

Read the source

Source

  • Source name: arXiv
  • Source kind: paper
  • Authors / organization: Yuzhong Shen, Masha Sosonkina, Peng Xu, Mark S. Gordon
  • Venue: arXiv: cs.AI
  • Published: 2026-08-12
  • Collected: 2026-08-13
  • URL: http://arxiv.org/abs/2608.12249v1
  • PDF: https://arxiv.org/pdf/2608.12249v1
  • DOI: UNKNOWN
  • Registry group: academic_apis_and_preprints
  • Evidence role: canonical_academic_index
  • Language: en
  • Country / region: UNKNOWN

Plain-Language Brief

Researchers propose an agentic workflow to modernize legacy Fortran code at a large scale. They apply this workflow in a case study by converting the two-electron-integral routines of GAMESS (a quantum-chemistry package) from fixed-form Fortran 77 to free-form Fortran 2008. The workflow uses prompt-specialized agent roles and an exact verification oracle, with humans serving as a small number of gates. The authors claim that all twelve source files pass various tests and there are no chemistry-relevant differences across the conversion.

What the Document Says

  • Researchers propose an agentic workflow for modernizing legacy Fortran code at production scale.
  • The workflow involves three prompt-specialized agent roles operating under a version-controlled specification authored and revised by the agents themselves.
  • Humans hold a small number of gates in this arrangement, which is kept safe by an exact verification oracle inherited from the domain.
  • The boundary of safe delegation lies exactly where that oracle stops seeing.
  • The researchers apply this workflow in a case study to convert the two-electron-integral routines of GAMESS from fixed-form Fortran 77 to free-form Fortran 2008.
  • The scope of the work was twelve source files, 56,448 lines, and 225 subroutines for computing electron repulsion integrals.
  • The agents ran as three Claude Code roles in isolated worktrees.
  • The work spanned four Claude model generations.
  • The authors adopt bit-for-bit reproduction of GAMESS's standard test suite energies as the merge criterion.
  • All twelve source files pass a 51-test validation battery consisting of the 49 standard GAMESS tests and two additional calculations.
  • Across 612 test runs, there are no chemistry-relevant differences and every file also passes Jenkins tests used for continuous integration.

Key Concepts Mentioned

  • legacy Fortran modernization
  • agentic workflow
  • prompt-specialized agent roles
  • exact verification oracle
  • GAMESS
  • two-electron-integral routines
  • Fortran 77 to Fortran 2008 conversion

Claims Made by the Document

  • The proposed workflow can modernize legacy Fortran code at production scale.
  • The agents can convert the two-electron-integral routines of GAMESS successfully.
  • All twelve source files pass various tests.

Evidence Used

  • The workflow is proposed and its implementation described.
  • The conversion process and its results are detailed.
  • Test results for the converted source files are presented.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Modernizing legacy Fortran is a problem of volume: the transformations are individually routine, but the codebases can be enormous, and across much of computational science the work simply goes undone. We propose an agentic workflow that takes this work on at production scale, and we set out to measure how far such delegation can reach. In this work, three prompt-specialized agent roles operate under a version-controlled specification that the agents themselves authored and revised, while humans hold a small number of gates. The arrangement is kept safe by an exact verification oracle inherited from the domain, and the boundary of safe delegation lies exactly where that oracle stops seeing. We apply the proposed workflow in a case study, converting the two-electron-integral routines of GAMESS (General Atomic and Molecular Electronic Structure System), a mature quantum-chemistry package with a 48-year development history, from fixed-form Fortran 77 to free-form Fortran 2008. The scope of this work was twelve source files, 56,448 lines, and 225 subroutines for computing electron repulsion integrals. The agents ran as three Claude Code roles in isolated worktrees, and the work spanned four Claude model generations. Because the GAMESS group ships a standard test suite whose printed energies its user community treats as canonical, we could adopt bit-for-bit reproduction of those energies as the merge criterion, where a deviation in the twelfth decimal place counts as a failure rather than drift. All twelve source files pass a 51-test validation battery comprising the 49 standard GAMESS tests and two additional calculations, and across 612 test runs the number of chemistry-relevant differences is zero, and every file also passes the Jenkins tests that are used for continuous integration.