Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

LIDT definitions: LDT / Preliminaries / SelfConsistency / DataProcessing

Definition
LIDT_MIPStarRE_LDT_Preliminaries_SelfConsistency_DataProcessing

by Dengnifer · Oct 27, 2026 · Mathlib 0df444a (Lean v4.33.1)

lidt

This bundle defines the two ingredients of the self-consistency data-processing step. It records the wrong-register estimate and then derives the data-processing inequality showing that self-consistency controls the processed measurement family.

Principal definitions:

  • wrongSideEstimate: bounds the contribution evaluated on the wrong register.
  • selfConsistencyImpliesDataProcessing: derives the data-processing bound from self-consistency.

Formalization credit: Sirui Lu, Ruixuan Deng, Yanqiao Zhu, and Zhengfeng Ji (arXiv:2609.19814). Mathematical source: Zhengfeng Ji, Anand Natarajan, Thomas Vidick, John Wright, and Henry Yuen (arXiv:2009.12982).

Definition code
-- Generated from MIPStarRE/LDT/Preliminaries/SelfConsistency/DataProcessing.lean by skeleton subtraction.
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_AxisParallelLine
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_DiagonalLine
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_Distribution
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_DistributionAvg
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_DistributionUniformSums
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_LinePolynomials
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_LowDegreePolynomial
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_MeasurementLift
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_OpFamily
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_OperatorExpectations
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_PMFAverages
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_PMFUniformAverages
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_ParametersBase
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_ParametersFiniteAnswers
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_QuantumState
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_SubMeasurementCore
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_SubMeasurementFamilies
import Definitions.Def_LIDT_MIPStarRE_LDT_Basic_TensorPlacement
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_BipartiteSelfConsistency_Completion
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_BipartiteSelfConsistency_Core
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_BipartiteSelfConsistency_Local
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_CauchySchwarz
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_ComparisonCore
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_Defs
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_DistanceBounds
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_SelfConsistency_Extensions
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_SwitchSandwichMain_Completeness
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_SwitchSandwichPrep_ApproxDelta
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_SwitchSandwichPrep_Core
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_SwitchSandwichPrep_InnerProduct
import Definitions.Def_LIDT_MIPStarRE_LDT_Preliminaries_Triangles_Core
import Definitions.Def_LIDT_MIPStarRE_LDT_Tactic_QuantumNonneg
import Definitions.Def_LIDT_MIPStarRE_LDT_Test_Defs
import Definitions.Def_LIDT_MIPStarRE_LDT_Test_StrategyCore
import Definitions.Def_LIDT_MIPStarRE_LDT_Test_StrategyFailures
import Definitions.Def_LIDT_MIPStarRE_LDT_Test_StrategyRole_Algebra
import Definitions.Def_LIDT_MIPStarRE_LDT_Test_StrategyRole_Core
import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_Basic
import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_BlockDiagonal
import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_NormalizedTrace
import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_Order
import Definitions.Def_LIDT_MIPStarRE_Quantum_FiniteMatrix_TracePairing
import Mathlib
import Mathlib.Analysis.MeanInequalitiesPow
import Mathlib.Probability.ProbabilityMassFunction.Integrals
import Mathlib.Probability.ProbabilityMassFunction.Monad

open scoped _root_.Matrix

namespace LIDT

namespace MIPStarRE.LDT.Preliminaries
end MIPStarRE.LDT.Preliminaries

/-!
# Self-consistency: data processing

`prop:self-consistency-implies-data-processing`: self-consistency implies that
postprocessing a measurement cannot substantially decrease consistency.

## References

- `references/ldt-paper/preliminaries.tex`
- `blueprint/src/chapter/ch03_preliminaries.tex`
-/

@[expose] public section

open scoped BigOperators MatrixOrder _root_.Matrix ComplexOrder

namespace MIPStarRE.LDT.Preliminaries

open MIPStarRE.LDT

/-- `prop:self-consistency-implies-data-processing`.

Proof:
1. First prove the “wrong-side” estimate
   `P_[f_q] ⊗ I ≈_{2δ + 4√ε} I ⊗ A_[f_q]` by:
   - expanding the `qSDD` square,
   - bounding the mass of `P` via `completenessTransferProjectiveP`,
   - comparing the mixed overlap with the diagonal overlap of `A`,
   - and using SSC plus question-dependent postprocessing monotonicity.
2. Apply `twoNotionsOfSelfConsistencyAfterEvaluation` to obtain
   `A_[f_q] ⊗ I ≈_{2δ} I ⊗ A_[f_q]`.
3. Use the `SDDRel` triangle inequality to conclude
   `P_[f_q] ⊗ I ≈_{8δ + 8√ε} A_[f_q] ⊗ I`. -/
private lemma wrongSideEstimate
    {Question α β : Type*} {ι : Type*}
    [Fintype ι] [DecidableEq ι] [Fintype α] [Fintype β]
    (ψ : QuantumState (ι × ι))
    (hperm : PermInvState ψ)
    (hψ : ψ.IsNormalized)
    (𝒟 : Distribution Question)
    (h𝒟 : ∑ q ∈ 𝒟.support, 𝒟.weight q ≤ 1)
    (A : IdxSubMeas Question α ι)
    (P : IdxProjSubMeas Question α ι)
    (δ ε : Error) (f : Question → α → β) :
    BipartiteSSCRel ψ 𝒟 A δ →
    SDDRel ψ 𝒟
      (IdxSubMeas.liftLeft (IdxProjSubMeas.toIdxSubMeas P))
      (IdxSubMeas.liftLeft A) ε →
      SDDRel ψ 𝒟
        (IdxSubMeas.liftLeft (fun q => postprocess ((P q).toSubMeas) (f q)))
        (IdxSubMeas.liftRight (fun q => postprocess (A q) (f q)))
        (2 * δ + 4 * Real.sqrt ε) := by
  intro hssc ⟨hε⟩
  let PL : IdxProjSubMeas Question α (ι × ι) := fun q =>
    { toSubMeas := ((P q).toSubMeas).liftLeft
      proj := fun a => by simp [SubMeas.liftLeft, leftTensor_mul_leftTensor, (P q).proj a] }
  let LPf : IdxSubMeas Question β (ι × ι) :=
    IdxSubMeas.liftLeft (fun q => postprocess ((P q).toSubMeas) (f q))
  let RAf : IdxSubMeas Question β (ι × ι) :=
    IdxSubMeas.liftRight (fun q => postprocess (A q) (f q))
  let crossPost : Question → Error := fun q =>
    qMatchMass ψ (LPf q) (RAf q)
  let crossOrig : Question → Error := fun q =>
    qMatchMass ψ
      ((IdxSubMeas.liftLeft (IdxProjSubMeas.toIdxSubMeas P)) q)
      ((IdxSubMeas.liftRight A) q)
  let overlapA : Question → Error := fun q =>
    qMatchMass ψ ((IdxSubMeas.liftLeft A) q) ((IdxSubMeas.liftRight A) q)
  have hcomp :
      CompTransferStmt ψ 𝒟 (IdxSubMeas.liftLeft A) PL ε := by
    apply completenessTransferProjectiveP ψ 𝒟 hψ h𝒟 (IdxSubMeas.liftLeft A) PL ε
    exact
      sddRel_symm ψ 𝒟
        (IdxSubMeas.liftLeft (IdxProjSubMeas.toIdxSubMeas P))
        (IdxSubMeas.liftLeft A) ε
        ⟨hε⟩
  have hmassP :
      idxSubMeasMass ψ 𝒟 LPf ≤
        idxSubMeasMass ψ 𝒟 (IdxSubMeas.liftLeft A) + 2 * Real.sqrt ε := by
    have hpost :
        idxSubMeasMass ψ 𝒟 LPf =
          idxSubMeasMass ψ 𝒟 (IdxProjSubMeas.toIdxSubMeas PL) := by
      unfold idxSubMeasMass subMeasMass avgOver LPf PL
      refine Finset.sum_congr rfl ?_
      intro q hq
      change
        𝒟.weight q *
            ev ψ (((postprocess ((P q).toSubMeas) (f q)).liftLeft).total) =
          𝒟.weight q * ev ψ ((((P q).toSubMeas).liftLeft).total)
      simp [SubMeas.liftLeft, postprocess_total]
    have hbase := hcomp.completenessTransfer
    rw [hpost]
    linarith
  have hmassA :
      idxSubMeasMass ψ 𝒟 RAf =
        idxSubMeasMass ψ 𝒟 (IdxSubMeas.liftLeft A) := by
    unfold idxSubMeasMass subMeasMass avgOver RAf
    refine Finset.sum_congr rfl ?_
    intro q hq
    change
      𝒟.weight q *
          ev ψ (rightTensor (ι₁ := ι) ((postprocess (A q) (f q)).total)) =
      𝒟.weight q * ev ψ (leftTensor (ι₂ := ι) ((A q).total))
    rw [postprocess_total, ← hperm.swap_ev ((A q).total)]
  have hcross_post_ge :
      avgOver 𝒟 crossPost ≥ avgOver 𝒟 crossOrig := by
    unfold avgOver crossPost crossOrig
    refine Finset.sum_le_sum ?_
    intro q hq
    exact mul_le_mul_of_nonneg_left
      (qMatchMass_leftRight_postprocess_ge ψ ((P q).toSubMeas) (A q) (f q))
      (𝒟.nonnegative q)
  have hcross_close :
      |avgOver 𝒟 crossOrig - avgOver 𝒟 overlapA| ≤ Real.sqrt ε := by
    simpa [crossOrig, overlapA, qMatchMass, IdxProjSubMeas.toIdxSubMeas,
      IdxSubMeas.liftLeft, IdxSubMeas.liftRight] using
      easyApproxFromApproxDelta ψ hψ 𝒟 h𝒟
        (IdxSubMeas.liftLeft (IdxProjSubMeas.toIdxSubMeas P))
        (IdxSubMeas.liftLeft A)
        (IdxSubMeas.liftRight A)
        ε
        ⟨hε⟩
  have hcross_ge :
      avgOver 𝒟 crossPost ≥ avgOver 𝒟 overlapA - Real.sqrt ε := by
    linarith [hcross_post_ge, (abs_le.mp hcross_close).1]
  have hssc_gap :
      idxSubMeasMass ψ 𝒟 (IdxSubMeas.liftLeft A) - avgOver 𝒟 overlapA ≤ δ := by
    rcases hssc with ⟨hssc⟩
    calc
      idxSubMeasMass ψ 𝒟 (IdxSubMeas.liftLeft A) - avgOver 𝒟 overlapA
        = avgOver 𝒟
            (fun q =>
              subMeasMass ψ ((IdxSubMeas.liftLeft A) q) - overlapA q) := by
              unfold idxSubMeasMass subMeasMass avgOver overlapA
              rw [← Finset.sum_sub_distrib]
              refine Finset.sum_congr rfl ?_
              intro q hq
              ring
      _ ≤ avgOver 𝒟 (fun q => qBipartiteSSCDefect ψ (A q)) := by
            apply avgOver_mono
            intro q
            have hdef :
                qBipartiteSSCDefect ψ (A q) =
                  max 0 (subMeasMass ψ ((IdxSubMeas.liftLeft A) q) - overlapA q) := by
              simp [qBipartiteSSCDefect, overlapA, qMatchMass, subMeasMass,
                IdxSubMeas.liftLeft, IdxSubMeas.liftRight,
                leftTensor_mul_rightTensor_eq_opTensor]
            rw [hdef]
            exact le_max_right 0 _
      _ ≤ δ := by
            simpa [bipartiteSSCError, overlapA, qMatchMass, IdxSubMeas.liftLeft,
              IdxSubMeas.liftRight] using hssc
  have hpointwise :
      ∀ q, qSDD ψ (LPf q) (RAf q) ≤
        subMeasMass ψ (LPf q) + subMeasMass ψ (RAf q) - 2 * crossPost q := by
    intro q
    let diagP : Error := ∑ b : β, ev ψ ((LPf q).outcome b * (LPf q).outcome b)
    let diagA : Error := ∑ b : β, ev ψ ((RAf q).outcome b * (RAf q).outcome b)
    have hdiagP_le :
        diagP ≤ subMeasMass ψ (LPf q) := by
      simpa [diagP, subMeasMass] using subMeas_diagMass_le_mass ψ (LPf q)
    have hdiagA_le :
        diagA ≤ subMeasMass ψ (RAf q) := by
      simpa [diagA, subMeasMass] using subMeas_diagMass_le_mass ψ (RAf q)
    have h_expand : qSDD ψ (LPf q) (RAf q) = diagP + diagA - 2 * crossPost q := by
      have hterm :
          ∀ b : β,
            ev ψ (((LPf q).outcome b - (RAf q).outcome b)ᴴ *
                ((LPf q).outcome b - (RAf q).outcome b)) =
              ev ψ ((LPf q).outcome b * (LPf q).outcome b) +
                ev ψ ((RAf q).outcome b * (RAf q).outcome b) -
                2 * ev ψ ((LPf q).outcome b * (RAf q).outcome b) := by
        intro b
        have hcomm :
            ev ψ ((RAf q).outcome b * (LPf q).outcome b) =
              ev ψ ((LPf q).outcome b * (RAf q).outcome b) := by
          exact
            ev_mul_comm_of_psd ψ _ _
              ((RAf q).outcome_pos b)
              ((LPf q).outcome_pos b)
        calc
          ev ψ (((LPf q).outcome b - (RAf q).outcome b)ᴴ *
              ((LPf q).outcome b - (RAf q).outcome b))
            =
              ev ψ
                ((((LPf q).outcome b * (LPf q).outcome b) -
                    (LPf q).outcome b * (RAf q).outcome b) -
                  ((RAf q).outcome b * (LPf q).outcome b -
                    (RAf q).outcome b * (RAf q).outcome b)) := by
                  congr 1
                  simp [sub_mul, mul_sub, SubMeas.outcome_hermitian]
                  abel
          _ =
              ev ψ ((LPf q).outcome b * (LPf q).outcome b) -
                ev ψ ((LPf q).outcome b * (RAf q).outcome b) -
                (ev ψ ((RAf q).outcome b * (LPf q).outcome b) -
                  ev ψ ((RAf q).outcome b * (RAf q).outcome b)) := by
                    rw [ev_sub, ev_sub, ev_sub]
          _ =
              ev ψ ((LPf q).outcome b * (LPf q).outcome b) +
                ev ψ ((RAf q).outcome b * (RAf q).outcome b) -
                2 * ev ψ ((LPf q).outcome b * (RAf q).outcome b) := by
                  rw [hcomm]
                  ring
      unfold qSDD qSDDCore crossPost
      calc
        ∑ b : β,
            ev ψ (((LPf q).outcome b - (RAf q).outcome b)ᴴ *
              ((LPf q).outcome b - (RAf q).outcome b))
          =
            ∑ b : β,
              (ev ψ ((LPf q).outcome b * (LPf q).outcome b) +
                ev ψ ((RAf q).outcome b * (RAf q).outcome b) -
                2 * ev ψ ((LPf q).outcome b * (RAf q).outcome b)) := by
                  refine Finset.sum_congr rfl ?_
                  intro b hb
                  exact hterm b
        _ = diagP + diagA - 2 * qMatchMass ψ (LPf q) (RAf q) := by
              unfold diagP diagA qMatchMass
              rw [Finset.sum_sub_distrib, Finset.sum_add_distrib, Finset.mul_sum]
    rw [h_expand]
    linarith
  constructor
  calc
    sddError ψ 𝒟 LPf RAf
      ≤ avgOver 𝒟
          (fun q =>
            subMeasMass ψ (LPf q) + subMeasMass ψ (RAf q) - 2 * crossPost q) := by
              unfold sddError
              apply avgOver_mono
              intro q
              exact hpointwise q
    _ = idxSubMeasMass ψ 𝒟 LPf +
          idxSubMeasMass ψ 𝒟 RAf -
            2 * avgOver 𝒟 crossPost := by
              unfold idxSubMeasMass subMeasMass avgOver
              calc
                ∑ q ∈ 𝒟.support,
                    𝒟.weight q *
                      (ev ψ (LPf q).total + ev ψ (RAf q).total -
                        2 * crossPost q)
                  =
                    ∑ q ∈ 𝒟.support,
                      (𝒟.weight q * ev ψ (LPf q).total +
                        (𝒟.weight q * ev ψ (RAf q).total -
                          2 * (𝒟.weight q * crossPost q))) := by
                          refine Finset.sum_congr rfl ?_
                          intro q hq
                          ring
                _ =
                    (∑ q ∈ 𝒟.support, 𝒟.weight q * ev ψ (LPf q).total) +
                      ((∑ q ∈ 𝒟.support, 𝒟.weight q * ev ψ (RAf q).total) -
                        2 * ∑ q ∈ 𝒟.support, 𝒟.weight q * crossPost q) := by
                          rw [Finset.sum_add_distrib, Finset.sum_sub_distrib, ← Finset.mul_sum]
                _ = idxSubMeasMass ψ 𝒟 LPf +
                    idxSubMeasMass ψ 𝒟 RAf -
                      2 * avgOver 𝒟 crossPost := by
                          simp [idxSubMeasMass, avgOver, subMeasMass]
                          ring
    _ ≤
        (idxSubMeasMass ψ 𝒟 (IdxSubMeas.liftLeft A) + 2 * Real.sqrt ε) +
          idxSubMeasMass ψ 𝒟 (IdxSubMeas.liftLeft A) -
            2 * (avgOver 𝒟 overlapA - Real.sqrt ε) := by
              rw [hmassA]
              linarith
    _ ≤ 2 * δ + 4 * Real.sqrt ε := by
          linarith

/-- `prop:self-consistency-implies-data-processing`.

A projective approximation to a strongly self-consistent family remains close
after postprocessing on the left register. The proof combines the wrong-side
estimate, self-consistency after evaluation, and the `SDDRel` triangle
inequality. -/
theorem selfConsistencyImpliesDataProcessing
    {Question α β : Type*} {ι : Type*}
    [Fintype ι] [DecidableEq ι] [Fintype α] [Fintype β]
    (ψ : QuantumState (ι × ι))
    (hperm : PermInvState ψ)
    (hψ : ψ.IsNormalized)
    (𝒟 : Distribution Question)
    (h𝒟 : ∑ q ∈ 𝒟.support, 𝒟.weight q ≤ 1)
    (A : IdxSubMeas Question α ι)
    (P : IdxProjSubMeas Question α ι)
    (δ ε : Error) (f : Question → α → β) :
    BipartiteSSCRel ψ 𝒟 A δ →
    SDDRel ψ 𝒟
      (IdxSubMeas.liftLeft (IdxProjSubMeas.toIdxSubMeas P))
      (IdxSubMeas.liftLeft A) ε →
      SDDRel ψ 𝒟
        (IdxSubMeas.liftLeft (fun q => postprocess ((P q).toSubMeas) (f q)))
        (IdxSubMeas.liftLeft (fun q => postprocess (A q) (f q)))
        (8 * δ + 8 * Real.sqrt ε) := by
  intro hssc hsdd
  have hwrong :
      SDDRel ψ 𝒟
        (IdxSubMeas.liftLeft (fun q => postprocess ((P q).toSubMeas) (f q)))
        (IdxSubMeas.liftRight (fun q => postprocess (A q) (f q)))
        (2 * δ + 4 * Real.sqrt ε) :=
    wrongSideEstimate ψ hperm hψ 𝒟 h𝒟 A P δ ε f hssc hsdd
  have hself :
      SDDRel ψ 𝒟
        (IdxSubMeas.liftLeft (fun q => postprocess (A q) (f q)))
        (IdxSubMeas.liftRight (fun q => postprocess (A q) (f q)))
        (2 * δ) :=
    twoNotionsOfSelfConsistencyAfterEvaluation ψ hperm 𝒟 A δ f hssc
  have hself_symm :
      SDDRel ψ 𝒟
        (IdxSubMeas.liftRight (fun q => postprocess (A q) (f q)))
        (IdxSubMeas.liftLeft (fun q => postprocess (A q) (f q)))
        (2 * δ) :=
    sddRel_symm ψ 𝒟
      (IdxSubMeas.liftLeft (fun q => postprocess (A q) (f q)))
      (IdxSubMeas.liftRight (fun q => postprocess (A q) (f q)))
      (2 * δ)
      hself
  have htri :
      SDDRel ψ 𝒟
        (IdxSubMeas.liftLeft (fun q => postprocess ((P q).toSubMeas) (f q)))
        (IdxSubMeas.liftLeft (fun q => postprocess (A q) (f q)))
        (2 * ((2 * δ + 4 * Real.sqrt ε) + (2 * δ))) := by
    exact
      stateDependentDistanceRel_triangle ψ 𝒟
        (IdxSubMeas.liftLeft (fun q => postprocess ((P q).toSubMeas) (f q)))
        (IdxSubMeas.liftRight (fun q => postprocess (A q) (f q)))
        (IdxSubMeas.liftLeft (fun q => postprocess (A q) (f q)))
        (2 * δ + 4 * Real.sqrt ε) (2 * δ)
        hwrong hself_symm
  exact
    stateDependentDistanceRel_mono ψ 𝒟
      (IdxSubMeas.liftLeft (fun q => postprocess ((P q).toSubMeas) (f q)))
      (IdxSubMeas.liftLeft (fun q => postprocess (A q) (f q)))
      (2 * ((2 * δ + 4 * Real.sqrt ε) + (2 * δ)))
      (8 * δ + 8 * Real.sqrt ε)
      (by linarith)
      htri

end MIPStarRE.LDT.Preliminaries

end

end LIDT
Source
https://github.com/LionSR/MIPStarRE/blob/f382c801026f53a9efc260fdee707cb35e0adb15/MIPStarRE/LDT/Preliminaries/SelfConsistency/DataProcessing.lean#L28-L343

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