Lemma I.9: PiV has weight≥1−d/alpha² and ≥1−2N e^(−2mu²alpha²); PiV rho^n PiV lies between PiV 2^(−nH±Kd alpha sqrt(n)); rank≤2^(nH+Kd alpha
ProvedQAlgorithms.Papers.Independent.Renew46.P43.varianceTypicalLet a nonzero finite system have dimension , and choose a fine orthogonal spectral resolution, including zero eigenspaces and a basis inside degeneracies. All states are normalized, , and all typicality tolerances are positive. Write for entropy in bits, , , and . A shadow is an effect with state mass at least . In the variance bounds put , , where counts the positive eigenvalues of and is their minimum, or zero on empty support. In the constant-typical bound , including . The weak-law factor states may be noncommuting, with target and operator-norm tolerance .
This is the source result with its stated operational or information-theoretic domain.
import Definitions.Def_QAlgorithms_Papers_Independent_Renew46_P43
noncomputable section
namespace QAlgorithms.Papers.Independent.Renew46.P43
open scoped BigOperators Matrix ComplexConjugate ComplexOrder ENNReal Matrix.Norms.L2Operator
open Filter MeasureTheory
open QAlgorithms.Papers.Independent.Information100.P08 (IsPure)
open QAlgorithms.Papers.Independent.Information100.P06 (Below)
open QAlgorithms.Papers.Independent.Information100.P60 (copies relativeBits)
open QAlgorithms.Papers.Independent.Dependency50.P19 (count both)
open QAlgorithms.Papers.Independent.Information100.P22 (coherentInformation)
attribute [local instance] Classical.propDecidable Classical.decEq
/-- Lemma I.9, PDF pp.12-13: both concentration estimates, both sandwich sides,
rank bound, and the eta-shadow bound. -/
theorem varianceTypical {d : ℕ} (S : FineResolution d) (n : ℕ) (hn : 0 < n)
(alpha : ℝ) (halpha : 0 < alpha) :
let Pi := varianceProjector S n alpha
let correction := varianceConstant * d * alpha * Real.sqrt n
effectProb (copies n S.state) Pi ≥ 1 - (d : ℝ) / alpha^2 ∧
effectProb (copies n S.state) Pi ≥ 1 - varianceTail S alpha ∧
Below ((exp2 (-n * vnEntropy S.state - correction) : ℂ) • Pi)
(Pi * copies n S.state * Pi) ∧
Below (Pi * copies n S.state * Pi)
((exp2 (-n * vnEntropy S.state + correction) : ℂ) • Pi) ∧
Pi.trace.re ≤ exp2 (n * vnEntropy S.state + correction) ∧
∀ (B : Matrix (Fin n → Fin d) (Fin n → Fin d) ℂ) (eta : ℝ),
IsShadow (copies n S.state) B eta →
B.trace.re ≥ (eta - varianceTail S alpha) * exp2 (n * vnEntropy S.state - correction) := by sorry
end QAlgorithms.Papers.Independent.Renew46.P43
end
Read-back
What the Lean code literally says, in plain math · gpt-6.1-sol
For every natural number , let consist of real weights summing to one and a complex unitary matrix of size ; these data cannot exist when . Put , and let have entries . For every positive integer and real , let consist of all words satisfying
where counts occurrences of , and define
Here zero weights force zero occurrences. Write , where is applied by continuous functional calculus, and put , . Define , let count the indices with , let be the minimum of those positive , or zero if there are none, and put . The declaration asserts
where means is positive semidefinite. Moreover, for every complex matrix indexed by these length- words and every real , if and are positive semidefinite and , then
There is no restriction on ; when no is positive, the defined tail is zero.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.