Section 12 pasting: line one-point transport — Cauchy-Schwarz setup #
Internal helper module; part of the file-split for #1127.
References #
references/ldt-paper/ld-pasting.texblueprint/src/chapter/ch09_pasting.tex
The linear (pre-max) form of the bipartite consistency defect.
Internal helper for the LdSandwichLineOnePoint Cauchy--Schwarz setup;
exposed for a future file-split (#1127).
Equations
Instances For
For option-valued families with no none mass, the linear bipartite
consistency defect is the paper's sum against the complementary right outcome.
This is the bookkeeping step that rewrites
⟨ψ|A_total ⊗ B_total|ψ⟩ - Σ_o ⟨ψ|A_o ⊗ B_o|ψ⟩ as
Σ_a ⟨ψ|A_a ⊗ (I - B_a)|ψ⟩ when Bob's family is a measurement and both
none outcomes vanish.
Internal helper for the LdSandwichLineOnePoint Cauchy--Schwarz setup;
exposed for a future file-split (#1127).
The linear consistency defect is nonnegative when the right-hand family is a
measurement. This lets the paper's averaged linear estimate feed the max 0
qBipartiteConsDefect maximum form without needing a pointwise absolute-value gap.
Internal helper for the LdSandwichLineOnePoint Cauchy--Schwarz setup;
exposed for a future file-split (#1127).
If the averaged linear consistency-defect comparison holds and the right
family is measurement-valued, then the averaged max 0 bipartite consistency
error comparison follows.
This is the paper-faithful formulation of lem:ld-sandwich-line-one-point: the
Cauchy--Schwarz argument controls an averaged linear expression, not an average
of pointwise absolute values. Nonnegativity of the linear defects removes the
outer max 0.
Internal helper for the LdSandwichLineOnePoint Cauchy--Schwarz setup;
exposed for a future file-split (#1127).
The original expanded off-diagonal scalar in ld-pasting.tex:960--963.
This is the source side after deleting extraneous tail coordinates and expanding
the linear consistency defect as Σ_a ⟨ψ|A_a ⊗ (I-B_a)|ψ⟩.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The intermediate scalar after the first Cauchy--Schwarz move
ld-pasting.tex:964--986 (eq:gonna-need-a-bigger-cauchy-schwarz).
For an original-order prefix outcome gs, orderedHalf is
G^{x_<i}_{g_<i} G^{x_i}_{g_i} while rotatedHalf is
G^{x_i}_{g_i} G^{x_<i}_{g_<i}. The first CS move replaces only the left half
of the sandwich, leaving orderedHalf† on the right.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The target expanded off-diagonal scalar after the two CS moves.
This is the moved-prefix side. The separate endpoint/prefix-completeness
collapse to ldGbcon is the already-proved
ldSandwichLineOnePointPrefixMoved_eq_endpoint, corresponding to
ld-pasting.tex:1011--1024.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Paper-faithful split of the remaining off-diagonal CS route.
The fields isolate the two uses of Preliminaries.closenessOfIP /
Preliminaries.closenessOfIPAdjoint in ld-pasting.tex:964--1010. The endpoint
collapse after these fields is already packaged by
ldSandwichLineOnePointPrefixMoved_eq_endpoint (ld-pasting.tex:1011--1024).
- firstCauchySchwarz : ldSandwichLineOnePoint_prefix_sourceOutcomeSum params strategy family hi ≤ ldSandwichLineOnePoint_prefix_afterFirstCSOutcomeSum params strategy family hi + √(commuteGHalfSandwichError params gamma zeta (i + 1))
First CS move, paper lines
964--986and labeleq:gonna-need-a-bigger-cauchy-schwarz: move the selectedGto the left of the prefix in the left half of the sandwich. - secondCauchySchwarz : ldSandwichLineOnePoint_prefix_afterFirstCSOutcomeSum params strategy family hi ≤ ldSandwichLineOnePoint_prefix_movedOutcomeSum params strategy family hi + √(commuteGHalfSandwichError params gamma zeta (i + 1))
Second CS move, paper lines
987--1010and labeleq:even-bigger-CS: move the selectedGthrough the adjoint/right half, reaching the moved-prefix scalar.
Instances For
Absolute-value form of the two off-diagonal Cauchy--Schwarz moves.
This is the direct output shape of Preliminaries.closenessOfIPAdjoint and
Preliminaries.closenessOfIP: each field compares the two adjacent scalar
averages from ld-pasting.tex:964--1010 with error √ν₄. The one-sided route
used downstream is only an arithmetic consequence of these absolute-value
bounds.
- firstAbs : |ldSandwichLineOnePoint_prefix_sourceOutcomeSum params strategy family hi - ldSandwichLineOnePoint_prefix_afterFirstCSOutcomeSum params strategy family hi| ≤ √(commuteGHalfSandwichError params gamma zeta (i + 1))
First CS move in the exact absolute-value form of
prop:closeness-of-ip, paper lines964--986and labeleq:gonna-need-a-bigger-cauchy-schwarz. - secondAbs : |ldSandwichLineOnePoint_prefix_afterFirstCSOutcomeSum params strategy family hi - ldSandwichLineOnePoint_prefix_movedOutcomeSum params strategy family hi| ≤ √(commuteGHalfSandwichError params gamma zeta (i + 1))
Second CS move in the exact absolute-value form of
prop:closeness-of-ip, paper lines987--1010and labeleq:even-bigger-CS.
Instances For
Ordered half-product appearing in the line-one-point CS step.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Rotated half-product appearing after moving the selected slice to the front.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Right-hand complement selected by the completed polynomial outcome.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Raw ordered left tensor family used in the generic CS proposition.
Equations
- MIPStarRE.LDT.Pasting.ldSandwichLineOnePointCS_Aord params family hi q gs = MIPStarRE.LDT.leftTensor (MIPStarRE.LDT.Pasting.ldSandwichLineOnePointCS_orderedHalf params family hi q gs)
Instances For
Raw rotated left tensor family used in the generic CS proposition.
Equations
- MIPStarRE.LDT.Pasting.ldSandwichLineOnePointCS_Arot params family hi q gs = MIPStarRE.LDT.leftTensor (MIPStarRE.LDT.Pasting.ldSandwichLineOnePointCS_rotatedHalf params family hi q gs)
Instances For
The adjoint of the ordered raw CS family, as an indexed operator family.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The adjoint of the rotated raw CS family, as an indexed operator family.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Raw left family for the adjoint-oriented CS input, indexed by original outcomes.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Raw right family for the adjoint-oriented CS input, indexed by original outcomes.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Raw commutation after last-reverse reindexing, lifted to sandwiched-line questions.
The adjoint raw family agrees with the ordered CS family.
The adjoint raw family agrees with the rotated CS family.
The adjoint-oriented raw-core bound needed by the line-one-point CS step.
The $C$ family for the first, right-action CS move.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The $C$ family for the second, left-action CS move.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Raw scalar on the source side of the first CS application.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Raw scalar on the target side of the first CS application.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Raw scalar on the source side of the second CS application.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Raw scalar on the target side of the second CS application.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Exact low-level facts needed to turn the generic closenessOfIP* lemmas into
ld-pasting.tex:964--1010 for the line-one-point statement.
This record separates the generic CS theorem instantiation (proved below) from the paper-specific facts:
- the adjoint-oriented raw square-distance bound corresponding to the first square root in lines 974--985 and reused in lines 1005--1010;
- the two unit-side measurement-completeness bounds from lines 986 and 1008;
- the algebraic regrouping/reindexing that identifies the raw CS scalars with the existing source, intermediate, and moved outcome sums.
The construction lemma below proves the unit bounds and regrouping equalities;
the remaining live analytic endpoint is the adjoint-oriented raw-core estimate
recorded by LdSandwichLineOnePointAdjointRawCoreBound.
- adjointRawCore : (avgOver (uniformDistribution (SandwichedLineQuestion params k)) fun (q : SandwichedLineQuestion params k) => qSDDCore strategy.state (fun (gs : GHatTupleOutcome params (i + 1)) => Matrix.conjTranspose (ldSandwichLineOnePointCS_Aord params family hi q gs)) fun (gs : GHatTupleOutcome params (i + 1)) => Matrix.conjTranspose (ldSandwichLineOnePointCS_Arot params family hi q gs)) ≤ commuteGHalfSandwichError params gamma zeta (i + 1)
- firstUnitBound (q : SandwichedLineQuestion params k) : ∑ gs : GHatTupleOutcome params (i + 1), (∑ u : Unit, ldSandwichLineOnePointCS_Cfirst params strategy family hi q gs u).conjTranspose * ∑ u : Unit, ldSandwichLineOnePointCS_Cfirst params strategy family hi q gs u ≤ 1
- secondUnitBound (q : SandwichedLineQuestion params k) : ∑ gs : GHatTupleOutcome params (i + 1), (∑ u : Unit, ldSandwichLineOnePointCS_Csecond params strategy family hi q gs u) * (∑ u : Unit, ldSandwichLineOnePointCS_Csecond params strategy family hi q gs u).conjTranspose ≤ 1
- source_eq_firstSourceRaw : ldSandwichLineOnePoint_prefix_sourceOutcomeSum params strategy family hi = ldSandwichLineOnePointCS_firstSourceRaw params strategy family hi
- afterFirst_eq_firstTargetRaw : ldSandwichLineOnePoint_prefix_afterFirstCSOutcomeSum params strategy family hi = ldSandwichLineOnePointCS_firstTargetRaw params strategy family hi
- afterFirst_eq_secondSourceRaw : ldSandwichLineOnePoint_prefix_afterFirstCSOutcomeSum params strategy family hi = ldSandwichLineOnePointCS_secondSourceRaw params strategy family hi
- moved_eq_secondTargetRaw : ldSandwichLineOnePoint_prefix_movedOutcomeSum params strategy family hi = ldSandwichLineOnePointCS_secondTargetRaw params strategy family hi
Instances For
The adjoint-oriented raw commutator square-distance bound used by the two
Cauchy--Schwarz applications in the proof of lem:ld-sandwich-line-one-point.
The surrounding endpoint expansions and option-valued match-mass identities are proved directly where they are used; this structure records only the nontrivial orientation of the half-sandwich commutation estimate.
- bound : (avgOver (uniformDistribution (SandwichedLineQuestion params k)) fun (q : SandwichedLineQuestion params k) => qSDDCore strategy.state (fun (gs : GHatTupleOutcome params (i + 1)) => Matrix.conjTranspose (ldSandwichLineOnePointCS_Aord params family hi q gs)) fun (gs : GHatTupleOutcome params (i + 1)) => Matrix.conjTranspose (ldSandwichLineOnePointCS_Arot params family hi q gs)) ≤ commuteGHalfSandwichError params gamma zeta (i + 1)
Instances For
The adjoint-oriented estimate for the paper's eq:add-in-the-bot term.
The generic closenessOfIP* applications need the adjoint-oriented
$D D^\dagger$ square-distance term that appears in ld-pasting.tex:980--985
and is reused at lines 1005--1010.