CMD 4704313 revision analysis version 3

The current synthetic results use a shared pipe area, conserved phase mass
flow, a prescribed slip closure, and a discrete pressure balance. Synthetic
records are identified as SYNTHETIC. The laboratory branch preserves the
published measured corrosion rates and biological replicate assignments.

Use Python 3.12 and install requirements.txt. From analysis, run sequentially:

python original/benchmark_generator.py --output original/results
python run_extension.py
python submission_information_controls/run_information_controls.py
python make_fixed_archive.py
python run_reviewer_checks.py --archive fixed_results.zip
python added/analyze_revision.py submission_information_controls added/results
python plot_revision.py submission_information_controls/per_realization_metrics.csv figures/Figure_1
python check_phase_consistency.py

The feature_scaling.py module records the fixed synthetic feature divisors.
All thirty seeds and all prescribed scenarios are included. Generated outputs
and fitted development XGBoost models accompany the code.

For the separate laboratory reanalysis, run from analysis:
python run_empirical_mic.py
python submission_empirical_fairness/run_fairness_controls.py
python submission_empirical_fairness/verify_fairness_controls.py
python submission_empirical_fairness/plot_empirical_figure2.py

Experimental source: Ivanovich et al. 2025, Frontiers in Microbiology 16,
1540664, https://doi.org/10.3389/fmicb.2025.1540664. Source-cell identifiers,
exposure duration, extraction audits and repository retrieval scripts are in
empirical_mic. The authors' raw workbooks remain at the linked repository.
This archive includes the derived extraction tables and reanalysis outputs.


Round 2 precision update 13 November 2026
Reported RMSE estimates, differences and interval endpoints use four significant figures.
All raw measurements, predictions and full-precision numerical metrics are retained.
Figures S2, S4, S5 and S11 have updated RMSE labels.
The same four figures can be regenerated without model fitting from this archive:
python analysis/round2_precision/update_figures.py --archive <this_archive.zip> --output round2_figures
