`isSafeOutput`

definition

show as: view math explainer →

module: IndisputableMonolith.Ethics.SigmaExternalizationAudit
domain: Ethics
line: 21 · github
papers citing: none yet

open explainer

Generate a durable explainer page for this declaration.

open lean source

IndisputableMonolith.Ethics.SigmaExternalizationAudit on GitHub at line 21.

browse module

All declarations in this module, on Recognition.

explainer page

Tracked in the explainer inventory; generation is lazy so crawlers do not trigger LLM jobs.

open explainer

formal source

  18theorem zero_output_zero_impact (s : ℝ) : sigmaImpact 0 s = 0 := by
  19  unfold sigmaImpact; ring
  20
  21def isSafeOutput (impact epsilon : ℝ) : Prop := |impact| < epsilon
  22
  23theorem safe_when_small (impact eps : ℝ) (h : |impact| < eps) :
  24    |impact| < eps := h
  25
  26def auditCost : ℕ := 1
  27
  28theorem audit_is_O1 : auditCost = 1 := rfl
  29
  30end
  31
  32end IndisputableMonolith.Ethics.SigmaExternalizationAudit

papers checked against this theorem

No papers in the current corpus map onto this theorem yet.