Repetition Lengths for Almost Bitflip Channels: logical phase law and syndrome concentration
OpenQAlgorithms.Papers.Independent.Intakenext100.P93.independent_phase_and_syndromeLet and let each physical qubit have independent amplitude and phase errors: , , and . For the cat of length , write for the parity of all physical phase errors and for its retained syndrome. Then and are independent, including impossible syndromes, and
For , the weight has the exact law
For every , its mass outside both normalized centers converges to zero:
The empty syndrome at has a separately defined zero tail mass. The centers may coincide at ; no inference of a logical amplitude error is claimed there. This states the phase law and qualitative two-center concentration without adding a finite tail exponent.
import Definitions.Def_QAlgorithms_Papers_Independent_Intakenext100_P93
noncomputable section
namespace QAlgorithms.Papers.Independent.Intakenext100.P93
open QAlgorithms
open scoped BigOperators ComplexConjugate Matrix
attribute [local instance] Classical.propDecidable Classical.decEq
/-- Smith--Smolin, PDF pp.2--3, Repetition Lengths for Almost Bitflip Channels:
phase parity is independent of the entire syndrome; syndrome weight is a
mixture of two binomials and concentrates near k qx and k(1-qx).
The centers may coincide and the endpoint distributions may be deterministic. -/
theorem independent_phase_and_syndrome (qx qz : ℝ)
(hx0 : 0 ≤ qx) (hx1 : qx ≤ 1) (hz0 : 0 ≤ qz) (hz1 : qz ≤ 1) :
(∀ (k : ℕ) (s : Qubits k) (v : Bool),
(∑ u : Bool, catLogicalSyndromeMass k (qx * (1 - qz)) (qx * qz) (qz * (1 - qx)) u v s) =
(if v then (1 - (1 - 2 * qz) ^ (k + 1)) / 2
else (1 + (1 - 2 * qz) ^ (k + 1)) / 2) *
iidCatSyndromeMass k (qx * (1 - qz)) (qx * qz) (qz * (1 - qx)) s) ∧
(∀ k : ℕ, (∑ s : Qubits k, ∑ u : Bool,
catLogicalSyndromeMass k (qx * (1 - qz)) (qx * qz) (qz * (1 - qx)) u true s) =
(1 - (1 - 2 * qz) ^ (k + 1)) / 2) ∧
(∀ (k r : ℕ), r ≤ k →
syndromeWeightMass k (qx * (1 - qz)) (qx * qz) (qz * (1 - qx)) r =
(Nat.choose k r : ℝ) *
((1 - qx) * qx ^ r * (1 - qx) ^ (k - r) +
qx * (1 - qx) ^ r * qx ^ (k - r))) ∧
(∀ epsilon : ℝ, 0 < epsilon →
Filter.Tendsto (fun k : ℕ => concentrationTail k qx qz epsilon)
Filter.atTop (nhds 0)) := by sorry
end QAlgorithms.Papers.Independent.Intakenext100.P93
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For every , including their endpoints, define for each natural number a weight on pairs by
where bits in exponents are interpreted as or . Let , let be the sum of over , , and , and set . These are precisely the masses formed with , , and . The theorem asserts that for every , every syndrome , and every Boolean ,
and that for every ,
For every pair of natural numbers with , it also asserts
where counts true bits. Finally, for every real , the sequence converges to as , where and, for ,
The finite identities include , and all zero powers, including , equal .
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.