lumiculelight does the math

Appendix — evidence

How we operate, and what we quoted

  • No fabricated results: every number of ours is backed by a reproducible run.
  • Simulated versus measured is stated; validated versus projected is stated.
  • We publish our own limits, including where a process supports only one DRC engine.
  • Verification reports three outcomes, not two — clear, gaps, failed baseline.
  • Agents propose, engines verify, no agent approves its own work, every tool call is recorded.
  • Third-party figures on this deck are quoted with their source and are never presented as ours.