SUPPLEMENTARY DATA 1: VERIFIED CORE REPRODUCIBILITY
Continuation-specific reliability assessment of receding-horizon wind-battery-reserve dispatch
Release amendment: 6 October 2026

START HERE
This package preserves the frozen scientific data, model, algorithms and saved controller results of the original final core. It adds a documented portability repair and verification records. Read AMENDMENT_README_20261006.txt for the exact scope and limitations; SCIENTIFIC_VERIFICATION_LEDGER_20261006.md maps claims to evidence.

REQUIREMENTS
The checked environment is Python 3.12.14, NumPy 2.3.5, SciPy 1.17.0 and pandas 2.2.3. HiGHS is supplied with SciPy; no commercial solver is required. In your own virtual environment, install the pinned requirements:
  python -m pip install -r requirements.txt
Installing dependencies may need a network connection. The computations below are offline.

1. VERIFY PACKAGE INTEGRITY
From this extracted package directory:
  python verify_manifest.py
The current MANIFEST_SHA256.csv covers every distributed payload file except itself. Historical manifests under verification_20261006 are provenance records, not the current release manifest.

2. RECOMPUTE ALL PRIMARY STATISTICS FROM SAVED OUTPUTS
  python core_reproducibility/reproduce_round3.py --analysis-only --output ../reproduced_analysis
This recomputes all three analysis groups, including paired events, 2,000-replicate block-bootstrap ranges, resource summaries and target-matched scores. It does not rerun optimization. Use a new empty output directory. Expected outcome: PASS for all three groups in comparison_to_archived.json.

3. REBUILD INPUTS AND RECOMPUTE CONTROLLERS (OPTIONAL, LONGER)
  python core_reproducibility/reproduce_round3.py --recompute --dry-run --output ../recomputed_check
  python core_reproducibility/reproduce_round3.py --recompute --workers 4 --output ../recomputed_full
This entry point uses the bundled UK processed input, frozen model and Australian normalized CSV. Supplementary Data 2 is NOT required for this normalized-input entry point. Its full end-to-end controller run was not repeated during this packaging amendment; command construction and frozen-hash checks were verified. October 6 component replay and sensitivity verification are described separately.

4. REPARSE ORIGINAL AEMO ARCHIVES (SEPARATE OPTIONAL CHECK)
Use the unchanged Supplementary_Data_2_AEMO_Raw_Sources.zip. Copy core_reproducibility/confirmation to a NEW working directory, and copy Supplementary Data 2's raw directory into that working copy. From that copied confirmation directory run:
  python acquire_scada_structure.py --reference-mw 453 --reference-description "Fixed project nameplate normalization reference; not current export capacity"
With all 30 archives present, this reparse is offline. Do not run it over the distributed frozen provenance files.

CONTENTS AND SCOPE
- core_reproducibility: frozen source-derived inputs, model, controller results, statistics, executable reproduction entry point and scientific review records
- robustness: archived sensitivity tables and relevant drivers; README limitations apply to what was freshly rerun
- timing_provenance: source and timestamp interpretation evidence
- auxiliary_scripts: four historical provenance-only construction scripts; NOT supported turnkey manuscript or figure reconstruction
- verification_20261006: verification records, original manifests and documented path-only changes

The historical directory name confirmation is retained for hash continuity. It denotes the 11-28 September 2026 Australian external stress-test cohort. UK data are previously exposed development data. Full claims remain conditional on the model, fixed cohorts, synthetic obligation and timing assumptions. No field-safety, universal calibration, economic superiority or broad cross-site claim is established.
