Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Haar-spiked states, Born outcome laws and extended KL expectation

Definition
IncoherentPurity_Adaptive_Supplement_Spike

by mikedeng1 · Nov 15, 2026 · Mathlib 0df444a (Lean v4.33.1)

p2o-batch-stoc27sp2o-gran-per-chapterp2o-plan-paperp2o-v1quantum-informationstatistics

For a positive integer dimension ddd, a direction v∈Cdv\in\mathbb C^dv∈Cd is a unit vector, represented by tr⁡(∣v⟩⟨v∣)=1\operatorname{tr}(|v\rangle\langle v|)=1tr(∣v⟩⟨v∣)=1. The published Haar law on unitaries induces a probability measure on such directions through the first column; forgetting the normalization witness gives exactly the published pure-vector Haar law. For 0≤θ≤10\le\theta\le10≤θ≤1, construct the density matrix

ρθ,v=1−θdI+θ∣v⟩⟨v∣.\rho_{\theta,v}=\frac{1-\theta}{d}I+\theta|v\rangle\langle v|.ρθ,v​=d1−θ​I+θ∣v⟩⟨v∣.

Its positive semidefiniteness and unit trace are proved in the constructor. For a fixed POVM MMM on any measurable outcome space, pθ,vp_{\theta,v}pθ,v​ is its actual Born probability measure at this state. Define the Haar information as

Ev D(pt,v∥ps,v).\mathbb E_v\,D(p_{t,v}\Vert p_{s,v}).Ev​D(pt,v​∥ps,v​).

Here DDD is Mathlib's natural-log measure-theoretic KL divergence, valued in [0,+∞][0,+\infty][0,+∞], and the expectation is a nonnegative extended Lebesgue integral. Failure of absolute continuity or integrability gives infinite divergence. This interface supplies the state family and information quantity used in Lemma 5.3 without imposing finite outcomes or full rank.

Formalization Note The constructor also permits the pure endpoint θ=1\theta=1θ=1; the information theorem retains the source's strict s,t<1s,t<1s,t<1. Structural helpers certify measurable maps and probability laws.

Definition code
import Definitions.Def_IncoherentPurity_Adaptive_Estimators

open MeasureTheory ProbabilityTheory Matrix
open scoped ENNReal ComplexOrder
set_option autoImplicit false
noncomputable section

namespace IncoherentPurity.Adaptive.Supplement

/-- Unit vectors: the trace of the rank-one outer product is the squared Euclidean norm. -/
abbrev Direction (d : ℕ) := {v : Fin d → ℂ // (outer v).trace = 1}

/-- The same first-column map used in the published `pureHaar`, with its unit-vector witness. -/
def haarDirection {d : ℕ} (hd : 0 < d) (U : Unitary d) : Direction d := by
  refine ⟨fun i => U.val i ⟨0, hd⟩, ?_⟩
  have h := congrArg (fun A : Op d => A ⟨0, hd⟩ ⟨0, hd⟩)
    (Matrix.UnitaryGroup.star_mul_self U)
  simpa [outer, Matrix.trace, Matrix.diag, Matrix.mul_apply, mul_comm] using h

lemma measurable_haarDirection {d : ℕ} (hd : 0 < d) : Measurable (haarDirection hd) := by
  apply Continuous.measurable
  apply Continuous.subtype_mk
  exact continuous_pi (fun i => (continuous_apply _).comp
    ((continuous_apply i).comp continuous_subtype_val))

/-- Normalized complex Haar law on unit vectors; its underlying vector law is `pureHaar H`. -/
def directionHaar {d : ℕ} (hd : 0 < d) (H : HaarBasis d) : Measure (Direction d) :=
  H.law.map (haarDirection hd)

instance directionHaar_probability {d : ℕ} (hd : 0 < d) (H : HaarBasis d) :
    IsProbabilityMeasure (directionHaar hd H) := by
  let := H.probability
  exact Measure.isProbabilityMeasure_map (measurable_haarDirection hd).aemeasurable

lemma directionHaar_map_val {d : ℕ} (hd : 0 < d) (H : HaarBasis d) :
    (directionHaar hd H).map Subtype.val = pureHaar H := by
  rw [directionHaar, Measure.map_map measurable_subtype_coe (measurable_haarDirection hd)]
  simp [pureHaar, hd, Function.comp_def, haarDirection]

/-- The actual spiked density matrix (5.4), including positivity and trace-one witnesses.
The closed parameter interval also permits the pure endpoint without any full-rank assumption. -/
def spikeState {d : ℕ} (hd : 0 < d) (θ : Set.Icc (0 : ℝ) 1) (v : Direction d) : State d := by
  refine ⟨((1 - θ.val) / (d : ℝ)) • (1 : Op d) + θ.val • outer v.val, ?_, ?_⟩
  · exact (Matrix.PosSemidef.one.smul (div_nonneg (sub_nonneg.mpr θ.property.2)
      (Nat.cast_nonneg d))).add
      ((Matrix.posSemidef_vecMulVec_self_star v.val).smul θ.property.1)
  · have hd' : (d : ℂ) ≠ 0 := by exact_mod_cast (Nat.ne_of_gt hd)
    simp only [Matrix.trace_add, Matrix.trace_smul, Matrix.trace_one, Fintype.card_fin,
      v.property]
    simp only [Algebra.smul_def]
    push_cast
    field_simp
    ring

lemma continuous_spikeState {d : ℕ} (hd : 0 < d) (θ : Set.Icc (0 : ℝ) 1) :
    Continuous (spikeState hd θ) := by
  apply Continuous.subtype_mk
  change Continuous (fun v : Direction d =>
    ((1 - θ.val) / (d : ℝ)) • (1 : Op d) + θ.val • outer v.val)
  apply continuous_pi
  intro i
  apply continuous_pi
  intro j
  change Continuous (fun v : Direction d =>
    ((1 - θ.val) / (d : ℝ)) • (1 : Op d) i j + θ.val • (v.val i * star (v.val j)))
  exact continuous_const.add
    ((((continuous_apply i).comp continuous_subtype_val).mul
      (((continuous_apply j).comp continuous_subtype_val).star)).const_smul θ.val)

/-- Actual Born outcome law of a fixed POVM on the constructed spike state. -/
def spikeOutcome {d : ℕ} (hd : 0 < d) {Ω : Type*} [MeasurableSpace Ω]
    (M : POVM d Ω) (θ : Set.Icc (0 : ℝ) 1) (v : Direction d) : Measure Ω :=
  M.born (spikeState hd θ v)

lemma measurable_spikeOutcome {d : ℕ} (hd : 0 < d) {Ω : Type*} [MeasurableSpace Ω]
    (M : POVM d Ω) (θ : Set.Icc (0 : ℝ) 1) : Measurable (spikeOutcome hd M θ) :=
  M.born.measurable.comp (continuous_spikeState hd θ).measurable

instance spikeOutcome_probability {d : ℕ} (hd : 0 < d) {Ω : Type*} [MeasurableSpace Ω]
    (M : POVM d Ω) (θ : Set.Icc (0 : ℝ) 1) (v : Direction d) :
    IsProbabilityMeasure (spikeOutcome hd M θ v) := by
  let := M.markov
  exact inferInstanceAs (IsProbabilityMeasure (M.born (spikeState hd θ v)))

/-- Natural-log KL, with value infinity on failure of absolute continuity or integrability,
and its extended nonnegative Haar expectation. No real totalized expectation is used. -/
def haarInformation {d : ℕ} (hd : 0 < d) (H : HaarBasis d)
    {Ω : Type*} [MeasurableSpace Ω] (M : POVM d Ω)
    (s t : Set.Icc (0 : ℝ) 1) : ℝ≥0∞ :=
  ∫⁻ v, InformationTheory.klDiv (spikeOutcome hd M t v) (spikeOutcome hd M s v)
    ∂directionHaar hd H

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; Section 5, PDF p.16 = printed p.16, Eqs. (5.4)–(5.10); Definition 2.1, PDF p.8 = printed p.8

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me