Multi-Timescale Edge–Cloud Control under Correlated Batch Demand and Endogenous Degradation: Structural Analysis and Approximate Optimization
Description
This dataset provides the reproducibility data, manuscript source, publication figures, and verification code supporting the manuscript “Multi-Timescale Edge–Cloud Control under Correlated Batch Demand and Endogenous Degradation: Structural Analysis and Approximate Optimization.” The archive contains the V1.19 LaTeX source, bibliography, reference PDF, 12 frozen CSV ledgers, 11 vector figure files, editable TikZ source for the state-feedback system diagram, Python source code, pinned environment information, SHA-256 checksums, and automated verification scripts. The computational materials cover the equal-mean Poisson, MMPP, and BMAP comparisons; the structural-regime map; the verified batch-size-triggered edge-admission instance; the correlation-induced policy-persistence campaign; the approximation-architecture and algorithmic ablations; the hysteresis analysis; and the state-space and formulation-size audit. The core reproduction script independently regenerates the equal-mean, structural-regime, and N = 20, 40, and 80 scaling campaigns in a separate output directory without overwriting the archived results. Additional frozen ledgers support manuscript-to-data verification of the BTEA, CIPP, approximation-architecture, algorithm-ablation, hysteresis, and extended-scaling results. Automated checks recompute and validate all 156 manuscript-facing numerical claims, verify the integrity of every archived file, regenerate Figures 2–11, and independently reproduce the Appendix A marginal-balance identity with an absolute residual of approximately 3.567 × 10⁻¹². The numerical experiments use analytically specified stochastic models and synthetic data; no confidential, personal, or proprietary data are included. Version 2 synchronizes the repository with manuscript V1.19, including the revised Figure 1 and final prose refinements, while preserving the previously audited mathematical model and numerical core. Detailed installation, execution, verification, and manuscript-compilation instructions are provided in README_V1_19.md.
Files
Steps to reproduce
1. Download and extract CAOR_Paper_V1_19_Reproducibility_Package.zip, and move to the extracted CAOR_V1_19_REPRO directory. 2. Using Python 3.12, create and activate a virtual environment: python3 -m venv .venv Linux or macOS: source .venv/bin/activate Windows: .venv\Scripts\activate 3. Install the pinned dependencies: python -m pip install -r requirements.txt 4. Verify the archive integrity and all manuscript-facing numerical claims: python verify_package_v1_19.py Expected result: CAOR V1.19 PACKAGE VERIFICATION: PASS This check verifies the SHA-256 manifest, all 12 archived CSV ledgers, all 11 figure PDFs, and all 156 manuscript-facing numerical claims. 5. Independently reproduce the Appendix A marginal-balance identity: python verify_appendix_identity_v1_19.py Expected final results: APPENDIX A FULL JOINT IDENTITY: PASS APPENDIX A FROZEN LEDGER COMPARISON: PASS The absolute numerical residual should be approximately 3.567e-12. 6. Regenerate Figures 2–11: python make_publication_figures_v1_19.py --output-dir reproduced_figures Expected result: CAOR V1.19 PUBLICATION FIGURES: PASS (10 PDFs) Figure 1 is provided as editable TikZ source in figures/fig_system_model.tex. 7. Regenerate the equal-mean, structural-regime, and core scaling campaigns: python run_core_v1_19.py --output-dir reproduced_outputs 8. Compare the regenerated results with the frozen manuscript ledgers: python compare_regenerated_v1_19.py reproduced_outputs Expected result: CAOR V1.19 REGENERATED CORE COMPARISON: PASS The regenerated files are written only to reproduced_outputs. The frozen archived files are not overwritten, and machine-dependent wall-clock columns are excluded from comparison. 9. To compile the manuscript PDF, run: pdflatex CAOR_Paper_V1_19.tex bibtex CAOR_Paper_V1_19 pdflatex CAOR_Paper_V1_19.tex pdflatex CAOR_Paper_V1_19.tex The reference build contains 38 pages, with no undefined references or citations, no overfull boxes, no Type-3 fonts, and no raster figures. Further details are provided in README_V1_19.md.
Institutions
- Sangji UniversityGangwon-do, Wŏnju