Four limits on identifying regime-dependent friction and wear laws from a machine joint's own sensors — reproducibility archive

Reproducibility materials supporting “Four limits on identifying regime-dependent friction and wear laws from a machine joint’s own sensors” by Shoniwa Mosekari, Sir Geoffrey Leigh Academy, Dartford, Kent, United Kingdom. The archive contains frozen protocols, research code, declared inputs, per-seed and aggregate results, models, numerical figures, execution and verification records, and scientific document-build inputs. It supports a synthetic friction–wear benchmark separating excitation information, regime observability, response-class restrictions and depth-sensor uncertainty. This extension is distinct from the preserved baseline release at DOI 10.5281/zenodo.23451487. Duplicated submission sets and superseded editorial-only copies are omitted; their exact exclusion inventory and the source-archive checksum are included. Scientific files and build inputs are unchanged. Reproduction: use Python 3.12 with requirements-pinned.txt and revision_mmt/document_build/requirements-documents-v29.txt. From the extracted archive root, run bash reproduce_all.sh prepare /absolute/path/to/new_clean_run, enter that new directory, then run MMT_WORKERS=4 bash reproduce_all.sh fresh all. Document construction additionally requires Pandoc, XeLaTeX, Poppler and LibreOffice. See README.md for the full procedure, stage controls, environment and comparison commands. Verification scope: the recorded composite clean campaign completed 4,166 required files and checked 4,215 generated artifact hashes. Strict comparison retains 16 documented differences; these are not relabelled as exact equality. The original 102-path wear-check producer was not recovered; a documented reconstruction checks the declared 2 ms left-held-forcing convention. Repeating the frozen 50-seed cohort reproduces that cohort. All numerical evidence is synthetic; experimental validation and a validated intervention-selection policy are not claimed. Conditional depth intervals and bound-active fits remain limitations. Full verification records are retained. Rights are mixed and file-specific. Existing MIT and CC BY 4.0 grants apply only to their allocated baseline materials. Manuscript sources and unlicensed extension materials retain author copyright; third-party rights are preserved. See LICENSE and PUBLIC_RIGHTS.txt. No archive-wide open licence is granted. Download and reconstruction: this record provides the unchanged archive in 77 numbered transport segments. Download all segments plus reconstruct_segments.py and UPLOAD_SEGMENTS_MANIFEST.json into one directory, then run python3 reconstruct_segments.py (or py -3 reconstruct_segments.py on Windows). The script checks every segment and reconstructs MMT_Extension_Reproducibility.zip, verifying its exact size (481,075,620 bytes) and SHA-256 (97a456aef1e2a7017fb4e976be5bba0525d5f823fb4496de03b999b5b41444f6). Read READ_ME_SEGMENTS_FIRST.txt before extracting. The segments are not independently extractable ZIP files.