Optical cut weights differ by at most a factor of two
ProvedQAlgorithms.Papers.Independent.Intakenext100.P84.optical_cut_factor_twoFix a cut of a finite quantum network, and let denote its labelled crossing edges in both directions. For each , let be its frequency and let be its intensity transmittance. Then
The inequality compares the two explicit optical cut weights uniformly over the cut, network topology, and frequencies. It remains valid for an empty cut or zero transmittances. Frequencies restricted to one cut need not sum to one.
Formalization Note The finite index type in Lean is precisely the set of crossing-edge labels. The inequality uses explicit real logarithms, rather than a ratio of capacities at zero transmittance.
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P84
noncomputable section
namespace QAlgorithms.Papers.Independent.Intakenext100.P84
open QAlgorithms
open scoped BigOperators
attribute [local instance] Classical.propDecidable Classical.decEq
/-- PDF p.4, Eq.(13): the optical upper cut weight is at most twice the
achievable cut weight. `E` indexes just the crossing edges of the fixed cut;
parallel and oppositely directed links retain separate labels. No positivity
of the cut weight or normalization of its restricted frequencies is required. -/
theorem optical_cut_factor_two {E : Type*} [Fintype E]
(frequency eta : E → ℝ)
(hfrequency : ∀ e, 0 ≤ frequency e)
(heta : ∀ e, 0 ≤ eta e ∧ eta e < 1) :
(∑ e, frequency e * Real.logb 2 ((1 + eta e) / (1 - eta e))) ≤
2 * ∑ e, frequency e * Real.logb 2 (1 / (1 - eta e)) := by sorry
end QAlgorithms.Papers.Independent.Intakenext100.P84
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For every finite type and every pair of real-valued functions , assume and for every . Then
Zero weights and values are allowed; all logarithm arguments are positive under the hypotheses. If is empty, both sums are zero.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.