Information in one measurement of a Haar-random spike
OpenIncoherentPurity.Adaptive.Supplement.one_measurement_informationp2o-batch-stoc27sp2o-gran-per-chapterp2o-plan-paperp2o-v1quantum-informationstatistics
Let be any positive integer, let be a normalized complex Haar-random unit vector in , and let be a fixed POVM on an arbitrary measurable outcome space, chosen independently of . Write for its Born outcome law on . For every ,
The divergence uses natural logarithms and may take value ; the expectation has the same extended nonnegative convention. Both inequalities retain the exact dimension factor, including dimension one, equal parameters, and zero POVM effects. This supplies the one-measurement information estimate for the non-adaptive lower-bound branch.
Differences from the source:
- encoding: Haar-random unit vector and parameters in [0,1) → a trace-one outer-product subtype with the published first-column Haar law, and closed-interval parameters with strict upper bounds; reason: the subtype has exactly the unit vectors and the parameter binders have exactly the source domain
- encoding: KL divergence and its Haar expectation → Mathlib natural-log klDiv and a nonnegative extended Lebesgue integral, with ENNReal.ofReal on the finite bounds; reason: preserves infinite divergence and excludes no measurable outcome space; all right-hand sides are nonnegative on the stated domain
Preamble
import Definitions.Def_IncoherentPurity_Adaptive_Supplement_Spike open MeasureTheory ProbabilityTheory Matrix open scoped ENNReal set_option autoImplicit false noncomputable section
Formal statement
namespace IncoherentPurity.Adaptive.Supplement
/-- Lemma 5.3, PDF = printed p.16; pinned source SHA-256 7d80a6695c9c0b12bfa088a8f47d2a298b93f9fde9668867acfb4a95b19f03ce. -/
theorem one_measurement_information {d : ℕ} (hd : 0 < d) (H : HaarBasis d)
{Ω : Type*} [MeasurableSpace Ω] (M : POVM d Ω)
(s t : Set.Icc (0 : ℝ) 1) (hs : s.val < 1) (ht : t.val < 1) :
haarInformation hd H M s t ≤
ENNReal.ofReal ((t.val-s.val)^2 / (1-s.val) * (((d : ℝ)-1)/((d : ℝ)+1))) ∧
ENNReal.ofReal ((t.val-s.val)^2 / (1-s.val) * (((d : ℝ)-1)/((d : ℝ)+1))) ≤
ENNReal.ofReal ((t.val-s.val)^2 / (1-s.val)) := by sorry
end IncoherentPurity.Adaptive.Supplement
Source
Lee–Wadhwa, Optimal Purity Estimation with Incoherent Measurements, STOC 2027 submission #66 (2026), preprint, pinned source.pdf; https://github.com/sigact-stoc/stoc2027/releases/download/papers-1/66.pdf; SHA-256 7d80a6695c9c0b12bfa088a8f47d2a298b93f9fde9668867acfb4a95b19f03ce; Lemma 5.3, PDF p.16 = printed p.16