Reproducibility Data and Verification Code for Cross-Timescale Stability and Stochastic Performance of Reputation-Aware Pricing with Hysteretic O-Cloud Scaling

Description

This dataset provides the reproducibility data, numerical results, and verification code accompanying the manuscript “Cross-Timescale Stability and Stochastic Performance of Reputation-Aware Pricing with Hysteretic O-Cloud Scaling.” The archive contains the exact finite-CTMC and tagged-delay analyses, independent uncensored discrete-event simulation (DES), seven-environment sensitivity experiments, three trace-informed traffic regimes, fixed-buffer and service-distribution robustness studies, computational-scaling experiments, symbolic verification, and an independent stationary-solver cross-check. The primary stationary analysis permits transient enumerated states when the finite CTMC has exactly one closed communicating class, assigning zero stationary mass to transient states. The archive also includes diagnostic analyses of communicating-class assumptions and expanded price domains used to assess the sensitivity of the policy comparisons to design-domain restrictions. Reproduction scripts, frozen numerical ledgers, figure-generation files, integrity manifests, execution logs, and provenance records are included. The Madrid LTE traces are used only to construct traffic-side contrast and persistence scenarios. The original raw trace files are not redistributed; upstream source identifiers, file hashes, preprocessing rules, and processed records are provided for reproducibility. The accompanying code reproduces the exact baseline analysis, sensitivity and robustness experiments, trace-informed scenarios, independent DES validations, and computational-scaling results reported in the manuscript.

Steps to reproduce

1. Download and extract the complete archive into a local working directory. 2. Verify the integrity of the archived evidence before running any numerical experiments: python verify_package.py This check uses only the Python standard library and verifies the file inventory, SHA-256 hashes, numerical-ledger consistency, key analytical anchors, absence of cache artifacts, and the archived verification gates. 3. Create a clean Python environment and install the supplied numerical dependencies: python -m pip install -r requirements.txt The revised computational campaign was executed with Python 3.13.5, NumPy 2.3.5, and SciPy 1.17.0. The optional symbolic checks use SymPy 1.14.0. 4. Reproduce the primary analyses reported in the manuscript in a new, initially empty output directory: python run_reproduction.py --revised --output-dir ../telecomsys_recomputed This reproduces the primary exact CTMC analysis, the certificate-matched sequential diagnostic, seven-environment sensitivity analysis, three trace-informed operating regimes, independent five-policy discrete-event simulation (DES), fixed-buffer robustness, paired exponential/Erlang-2 service-law experiments, computational-scaling runs, and associated verification outputs. 5. To reproduce the preserved early validation stages in addition to the primary analyses, run: python run_reproduction.py --full --output-dir ../telecomsys_full_recomputed The output directory must be empty before execution. 6. Run the independent symbolic verification of the analytical anchors and stability expressions: python revision_checks/verify_algebra.py 7. Compare the newly generated outputs with the frozen numerical ledgers and figures included in the archive. The supplied verification scripts check stationary probabilities, flow balances, tagged-state identities, policy-selection results, DES validation targets, sensitivity results, trace mappings, robustness experiments, and computational-scale quantities. 8. The independent DES experiments use fixed master seeds recorded in the archive. The five-policy DES uses 60 replications per policy, a 2,000-TU warm-up, and 50,000 measured TU per replication. The paired exponential/Erlang-2 experiment uses the same warm-up and collection horizon with 60 replications per policy and service law. 9. The Madrid LTE raw trace files are not redistributed in this dataset. Processed trace records, upstream source identifiers, file hashes, preprocessing rules, and provenance information are included. Reproducing the preprocessing directly from the original raw traces requires obtaining those source files separately from the upstream repository under its applicable terms. 10. Computational timing and memory measurements are machine-dependent. Numerical identities, policy selections, analytical checks, and stochastic validation targets should reproduce within the tolerances implemented in the supplied verification scripts.

Institutions

• Sangji University

  Gangwon-do, Wŏnju

Categories

Licence