SUPPLEMENTARY DATA 1 -- reconstruction instructions and evidence scope

Extract Supplementary_Data_1.zip and Supplementary_Code_1.zip into the SAME parent directory. Both contain SR_Reproducibility/. Keep this folder tree intact. No files with the same path have different contents. See README_CODE.txt for the one-command offline reproduction.

analysis/: the authors' self-collected phone telemetry is the main empirical experiment. combined_aligned_samples.csv retains all 43,278 CSV row positions and their original order; raw_selected_fields.csv retains all 43,174 selected raw electrical records. Column selection removes application/foreground package names, network identifiers, diagnostic dumps, original file paths and device/user identifiers. Acquisition timestamps and session stems are retained to preserve chronological splits, outages and time-support weights. Wi-Fi-enabled and cellular-attached are binary hardware states, not identifiers. No rows are dropped to anonymize the data. The phone is one ZTE 7552N device; repeated samples, rotations and forward blocks are not independent devices. Read DATA_DICTIONARY.csv.

The original private sessions.zip and diagnostic/raw streams are not included. Field-limited aligned telemetry is sufficient for all supplied phone comparisons, coverage plots and reported source-sensitivity checks. Re-running original alignment requires the private source; aggregate alignment/source audits record that provenance without exposing excluded content. Metadata counts of raw files/malformed lines describe the private original, not files included here. No human enrollment or participant identity is inferred.

round2/capacity/data/: 400 real UCL EIL-015 discharge-capacity observations and 636 real NASA observations in four cells (168,168,168,132). CSV decimal values and cycle ordering are unchanged from original extraction. Large NASA ZIP/MAT and UCL full electrochemical CSV are not included. Offline fitting uses the extracted CSVs; optional scripts retrieve official raw sources and recheck extraction. See SOURCES_AND_REUSE.txt and retained extraction hashes. Capacity experiments are supporting public-cell evidence and are separate from the authors' phone experiment.

round2/androwatts/aggregated.csv: a 13-column subset of the public CC-BY-4.0 1,000-profile aggregate, preserving every source value string and row. ID is an index, not verified chronology. The target uses a nominal 4.4 V power proxy, not measured terminal power. field_selection.json explains the derived-file checksum. Source README and original selected-file manifest are included for attribution.

reference_results/: original analysis results for numerical equivalence checks; these are outputs, not additional observations. In particular reference_results/round2/numerical contains explicitly designed numerical verification, NOT measured data. No simulated values are mixed with phone or public experimental records.

DATA_MANIFEST.json lists SHA-256 for every Data-archive payload file. See PACKAGE_CHANGELOG.txt in the Code archive for portability adaptations. These files are supplied for review; no public repository deposit is represented by this package.
