Skip to content

fix(ftip): define integrable evaluation and inverse-cost domains [AGENT] - #175

Merged
utensil merged 1 commit into
mainfrom
kepler/ftip-repair-r2-domains-20260908
Sep 8, 2026
Merged

fix(ftip): define integrable evaluation and inverse-cost domains [AGENT]#175
utensil merged 1 commit into
mainfrom
kepler/ftip-repair-r2-domains-20260908

Conversation

@utensil

@utensil utensil commented Sep 8, 2026

Copy link
Copy Markdown
Owner

Expected performance and gain were written as real-valued quantities without sufficient integrability assumptions, while inverse-frontier ratios could divide by zero even when every individual intervention had positive cost.

This change defines the measurable evaluation law and absolute-integrability requirements for each compared protocol and baseline. It declares empty/unbounded supremum conventions, excludes infinite expected costs from finite budgets, and restricts real differences and inverse-cost ratios to their stated domains. It also distinguishes an unattained frontier threshold from an executable witness, and measured cost ratios from inverse-frontier ratios. The changes cover 16 related FTIP pages and preserve their stable addresses.

Validation: analytic checks for signed heavy tails, empty/unbounded/unattained frontiers, positive costs with zero infimum, and bounded scalar/vector examples; all 22 prose regression tests; just chk; full just build; just verify-render (2,266 page pairs); and the 231-page focused FTIP PDF. The 16 affected HTML pages and surrounding routes render without KaTeX errors; affected HTML/PDF layout was inspected. All 773 source pages and 773 HTML pages plus the PDF are archived and hash-verified. A fresh independent adversarial review approved exact head f83dabc1b913c8a34ac9b762df54fa5105b69ed3 with no unresolved findings. It independently reconstructed the analytic cases and verified all archived source and output payloads.

Declare measurable evaluation spaces and the independent protocol/evaluation product law, with absolute integrability for every compared protocol and baseline. Give scalar and vector expectations, cost budgets, empty or unbounded suprema, and inverse thresholds explicit domains. Require finite cost-ratio operands and a strictly positive denominator, distinguish measured costs from inverse-frontier infima, and propagate finite-value guards to the direct difference and saturation statements.
@utensil
utensil merged commit 0014e30 into main Sep 8, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant