Eq. (2.23) — for
ProvedFuchsThesis.Chernoff.eq_2_23inequalitiesinformation-theoryp2o-gran-per-chapterp2o-plan-bookp2o-strengthenedp2o-v1probability
For any two nonnegative real numbers and any ,
Applied termwise to this inequality converts the Bayes error into a sum of products, which factorizes over independent samplings and gives the Chernoff bound.
Formalization Note Real powers use Lean's convention .
Differences from the source:
- generalization: a, b positive → a, b ≥ 0 (real powers with 0^0 = 1); reason: the inequality remains true when a or b is 0, and the statement specializes to the source's for positive a, b
Preamble
import Mathlib
Formal statement
namespace FuchsThesis.Chernoff
/-- Fuchs, *Distinguishability and Accessible Information in Quantum Theory*,
arXiv:quant-ph/9601020v1, §2.2.2, eq. (2.23), p.17 (PDF p.26): for any two positive numbers
`a`, `b` and any `0 ≤ α ≤ 1`, `min{a, b} ≤ a^α b^{1−α}`.
**Formalization Note.** Stated for `a, b ≥ 0` (the source: positive), with `Real.rpow`
(`0 ^ 0 = 1`); it specializes to the source's statement for positive `a`, `b`. -/
theorem eq_2_23 (a b α : ℝ) (ha : 0 ≤ a) (hb : 0 ≤ b) (hα0 : 0 ≤ α) (hα1 : α ≤ 1) :
min a b ≤ a ^ α * b ^ (1 - α) := by sorry
end FuchsThesis.Chernoff
Source
Fuchs, Distinguishability and Accessible Information in Quantum Theory, arXiv:quant-ph/9601020v1, §2.2.2, eq. (2.23), p.17 (PDF p.26)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.