Verification software for Roman Labels Yield One-Round Power Domination

Abstract

Offline verification software accompanying Seth Christopher's preprint “Roman Labels Yield One-Round Power Domination”. Scientific version 1.0.0; packaging revision figshare.2. All 24 Python files are byte-identical to figshare.1. The README, release metadata and checksum manifest have been updated; historical versions of changed software metadata are retained. Roman_Power_Verification_Software_v1.0.0_figshare2.zip contains the verification implementation, the original-definition standalone checker, a strict-JSON wrapper, a bounded public-source replay runner, and a two-pass LaTeX build driver. This authored software subset is offered under MIT to the extent the author holds the relevant rights. It does not relicense the original private archive or third-party material. Required inputs are in Roman_Power_Manuscript_Source_and_Evidence_v1.0.0_figshare2.zip in the companion Preprint item. Extract both archives into one NEW parent directory, merging Roman_Power_v1.0.0. Their paths do not overlap. The optional full historical audit archive is not required for the complete replay. From the merged directory run:python3 -I -S -B verification/check.py inputs/primary_task.json inputs/historical_primary_admitted.json --censuspython3 -I -S -B reproduce.py --out ../roman_power_replay Full replay uses Linux, Python 3.10+ and the standard library. The output directory must be new and outside the release. Expected status: SOURCE_ONLY_ROMAN_POWER_REPLAY_PASS. Fresh testing from the final rebuilt ZIPs completed 83 bounded serial processes; each child defaults to 60 wall seconds and 4 GiB address space. No model call, network service or embedded runtime is needed. A resource stop is unresolved execution, not a counterexample or theorem. The public runner does not recreate the original private sandbox or claim confidential-read isolation. Asserted computational roles are not authenticated agent identities or independent reviewers. The universal theorem rests on the written proof, not these finite tests. AI assistance: ChatGPT provided substantial assistance with target selection, the constructive proof, source research, verification software, test execution, and manuscript preparation. Aurelia generated augmentations; Selene reconstructed neighborhood, payment, and simultaneous-force obligations; Mica reconstructed byte adjacency, checked admission, and freshly rechecked use. These are computational roles, not coauthors or independent peer reviewers. Seth Christopher initiated and directed the research programme and is the sole author responsible for the submitted content.