Reproducibility Artifacts for Falsification-Guided Structure-Preserving Inverse Identification under Simulation-to-Experiment Shift

This record contains the large computational artifacts accompanying the study “Falsification-Guided Structure-Preserving Inverse Identification under Simulation-to-Experiment Shift: Abstention, Label-Free Adaptation, and Independent Physical Consistency.” The archive complements the public source-code and compact-results repository at https://github.com/MohammadAliJavari/two-state_reproducibility (release v1.0.0). Because of file size, the archive is distributed as five independent standard ZIP files. These files are not byte-split fragments. Users should download all five ZIP files and ZENODO_PARTS_SHA256.txt, verify the ZIP-level SHA-256 hashes, and extract all five archives into the same destination directory. The common reproducibility_archive/ directory will then be reconstructed. Each part includes REASSEMBLE_INSTRUCTIONS.txt, and the reconstructed archive contains ARCHIVE_FILE_SHA256.csv for file-level integrity verification. The archive contains generated source-bank artifacts, retained model outputs/checkpoints, prediction products, and associated computational evidence used in the study. Raw third-party CECEN and NEES experimental datasets are not redistributed; provenance and acquisition information are provided in the associated GitHub repository.