Skip to content

reproducibility.md

Latest commit

 

History

History
65 lines (46 loc) · 1.76 KB

reproducibility.md

File metadata and controls

65 lines (46 loc) · 1.76 KB
Raw

Reproducibility and evidence

Evidence classes

VoltCast keeps three kinds of material separate:

  1. Mechanism: equations and invariants implemented in src/voltcast/.
  2. Synthetic inputs: transparent workload profiles in scenarios/.
  3. Demonstration outputs: versioned CSV summaries in artifacts/baseline/.

No bundled result is presented as a hardware measurement.

Reproduce the baseline

From an editable installation:

voltcast compare \
  scenarios/reading-dark.csv \
  scenarios/mixed-day.csv \
  scenarios/streaming.csv \
  scenarios/gaming.csv \
  --output artifacts/baseline/new.csv

voltcast compare \
  scenarios/reading-dark.csv \
  scenarios/mixed-day.csv \
  scenarios/streaming.csv \
  scenarios/gaming.csv \
  --aged \
  --output artifacts/baseline/aged.csv

python scripts/render_baseline_svg.py

Run the behavioral checks:

python -m unittest discover -s tests -v

Claims supported by the baseline

The committed baseline demonstrates that the implementation:

  • produces shorter endurance under heavier constant power;
  • produces shorter endurance from lower initial SOC;
  • exposes an earlier voltage boundary after the demo aging transform;
  • preserves workload transitions in a zero-order-hold profile;
  • records the state needed to inspect each stop decision.

It does not establish prediction error against a physical phone.

Adding measured data

A measured profile must document:

  • device and battery configuration;
  • sampling method and units;
  • operating-system and workload conditions;
  • license and redistribution permission;
  • preprocessing steps;
  • uncertainty or repeatability information.

Do not add proprietary traces, copyrighted papers, personal identifiers, or screenshots containing account information.