Documentation

MIPRE.Background.LIDT.Co.SelfImprovement.Theorems.Results.HelperSSC.PostDeleteA

Helper strong self-consistency bounds: post-delete transports #

Post-delete-an-A transport lemmas culminating in the moved-quantity bound used before the residual-chain assembly: the counterpart of the vendored SelfImprovement/Theorems/Results/HelperSSC/PostDeleteA.lean (under MIPRE/Background/LIDT/MIPStarRE/LDT/) in the port of planning/c6b-plan.md (milestone M10, section "Port conventions").

Contents #

A strategy is a SymStrat params 𝔓 K; the polynomial submeasurement T, the sandwiched submeasurement H^u and the point projectors live in 𝔓, and every scalar is strategy.state.ev of a joint operator strategy.state.opTensor X Y; leftTensor and rightTensor are strategy.state.L and strategy.state.R, and Xᴴ is star X. The vendored helper_pair_sandwich_operator_sum_le names no state, its opTensor being the Kronecker product on the carrier; here it takes the model (S : SymModel 𝔓 K) as an explicit first argument, as Co AddInUStep34AndTransfer/Selected's addInU_selected_sandwich_tensor_if_sum_le does, and its hypothesis (X h)ᴴ = X h is star (X h) = X h.

Global variance. The variance factor is bounded by M6's summed deviation ∑_g globalVarianceDeviationAtPolynomial params strategy strategy.state T g (Co GlobalVariance/Defs/Families), the independent-points variance of the family u ↦ S.L (A^u_{g(u)}) * S.R √T_g on strategy.state itself rather than the vendored weighted state; it is identified through Co weightedPointConditionedOperator_sq, exactly where the vendored proof rewrites with its vendored counterpart, so the statement keeps its vendored text.

Placement bounds. The two operator bounds ∑_{h,h'} H_{h'} ⊗ T_h = H.total ⊗ T.total ≤ 1 of helper_pair_tensor_mass_le_one and helper_moveOverV_C_contraction, which the vendored file proves twice through opTensor_le_leftTensor and leftTensor_le_one (ι₂ := ι), are the private pair_opTensor_sum_le_one, closed by the keystone's S.opTensor_le_one.

Proofs that differ from the vendored ones.

Dropped hypotheses. The vendored helper_pair_tensor_mass_le_one closes with ev_one_of_isNormalized strategy.state strategy.isNormalized (vendored line 113), and the vendored helperDeleteAClonedQuantity_abs_sub_moveOverVQuantity_le_sqrt_two_delta passes strategy.isNormalized to closenessOfInnerProduct_right (vendored line 607); here strategy.state.ev_one_of_isNormalized and the ported Cauchy--Schwarz bound take no hypothesis. No statement of the vendored file carries a swap, density or normalization hypothesis, so no statement changed beyond the translation. The file sets no option, the vendored file-wide respectTransparency false not being needed.

Not ported #

Every declaration of the vendored file has a counterpart here.

References #

In LionSR/MIPStarRE at commit 507e8122, not in this repository:

Post-delete-an-A transports #

theorem MIPRE.LIDT.Co.SelfImprovement.helper_pair_sandwich_operator_sum_le {𝔓 : Type u_1} [CStarAlgebra 𝔓] [PartialOrder 𝔓] [StarOrderedRing 𝔓] {K : Type u_2} [NormedAddCommGroup K] [InnerProductSpace ℂ K] [CompleteSpace K] (S : SymModel 𝔓 K) (params : MIPStarRE.LDT.Parameters) [MIPStarRE.LDT.FieldModel params.q] (H T : SubMeas (MIPStarRE.LDT.Polynomial params) 𝔓) (X : MIPStarRE.LDT.Polynomial params → 𝔓) (hX_herm : ∀ (h : MIPStarRE.LDT.Polynomial params), star (X h) = X h) :
∑ hh : MIPStarRE.LDT.Polynomial params × MIPStarRE.LDT.Polynomial params, S.opTensor (X hh.1 * H.outcome hh.2 * X hh.1) (T.outcome hh.1) ≤ ∑ h : MIPStarRE.LDT.Polynomial params, S.opTensor (X h * X h) (T.outcome h)

The pair sandwich ∑_{(h, h')} (X_h H_{h'} X_h) ⊗ T_h is at most ∑_h X_h² ⊗ T_h for self-adjoint X_h: summing over h' gives H.total ≤ 1.

theorem MIPRE.LIDT.Co.SelfImprovement.helper_pair_tensor_mass_le_one {𝔓 : Type u_1} [CStarAlgebra 𝔓] [PartialOrder 𝔓] [StarOrderedRing 𝔓] {K : Type u_2} [NormedAddCommGroup K] [InnerProductSpace ℂ K] [CompleteSpace K] (params : MIPStarRE.LDT.Parameters) [MIPStarRE.LDT.FieldModel params.q] (strategy : SymStrat params 𝔓 K) (H T : SubMeas (MIPStarRE.LDT.Polynomial params) 𝔓) :
∑ hh : MIPStarRE.LDT.Polynomial params × MIPStarRE.LDT.Polynomial params, strategy.state.ev (strategy.state.opTensor (H.outcome hh.2) (T.outcome hh.1)) ≤ 1

The pair mass ∑_{(h, h')} ev(H_{h'} ⊗ T_h) of two submeasurements is at most 1.

The variance factor of eq:swap-u-for-v-attack-of-the-clones is at most the summed global-variance deviation: the sandwich (A^u - A^v) H^u_{h'} (A^u - A^v) is dominated by (A^u - A^v)² after summing over h', and averaging over independent points gives the deviation.

The mass factor of eq:swap-u-for-v-attack-of-the-clones is at most 1.

Paper eq:swap-u-for-v-attack-of-the-clones: after delete-an-A, the remaining point projector may be evaluated at an independent point at cost √ζ_variance.

theorem MIPRE.LIDT.Co.SelfImprovement.helper_moveOverV_C_contraction {𝔓 : Type u_1} [CStarAlgebra 𝔓] [PartialOrder 𝔓] [StarOrderedRing 𝔓] {K : Type u_2} [NormedAddCommGroup K] [InnerProductSpace ℂ K] [CompleteSpace K] (params : MIPStarRE.LDT.Parameters) [MIPStarRE.LDT.FieldModel params.q] (strategy : SymStrat params 𝔓 K) (T : SubMeas (MIPStarRE.LDT.Polynomial params) 𝔓) (uv : MIPStarRE.LDT.Point params × MIPStarRE.LDT.Point params) :
(∑ a : MIPStarRE.LDT.Fq params, star (∑ hh : MIPStarRE.LDT.Polynomial params × MIPStarRE.LDT.Polynomial params, if hh.1.toFun uv.2 = a then strategy.state.opTensor ((sandwichedPolynomialSubMeasAt params strategy T uv.1).outcome hh.2) (T.outcome hh.1) else 0) * ∑ hh : MIPStarRE.LDT.Polynomial params × MIPStarRE.LDT.Polynomial params, if hh.1.toFun uv.2 = a then strategy.state.opTensor ((sandwichedPolynomialSubMeasAt params strategy T uv.1).outcome hh.2) (T.outcome hh.1) else 0) ≤ 1

The Cauchy--Schwarz side condition of eq:move-over-v: with C_a = ∑_{(h, h') : h(v) = a} H^u_{h'} ⊗ T_h, ∑ₐ C_a^* C_a ≤ 1.

Paper eq:move-over-v: the cloned delete-an-A quantity can be moved to the right tensor factor at cost √(2δ) from point self-consistency.