LIDT definitions: LDT / SelfImprovement / MatrixRealization / Canonical / StrongDuality / Separation
DefinitionLIDT_MIPStarRE_LDT_SelfImprovement_MatrixRealization_Canonical_StrongDuality_Separationlidt
This bundle develops the separation and attainment steps for the canonical matrix SDP. It constructs the primal image cone, characterizes its identity fiber, converts functional dual feasibility into a trace-pairing matrix witness, and proves attainment of the canonical primal objective.
Principal definitions:
matrixSdpCanonicalPrimalConeMap: maps a primal matrix to its constraint and objective values.matrixSdpCanonicalPrimalImageCone: the closed image cone of positive canonical primal matrices.matrixSdpCanonicalPrimalImageCone_mem_of_nonnegative: places a positive primal matrix in the image cone.matrixSdpCanonicalPrimalImageCone_mem_of_feasible: places a feasible matrix on the identity constraint fiber.matrixSdpCanonicalPrimalImageCone_identity_mem_iff_exists_feasible_objective: characterizes objective values on that fiber.matrixSdpCanonicalTracePairingDualMatrix_dualFeasible_of_conicFunctionalDualFeasible: converts a functional dual bound to a feasible matrix dual.matrixSdpCanonicalFunctionalDualObjective_min_of_matrixDualMin: transfers matrix dual minimality to functional dual minimality.matrixSdpCanonicalPrimalObjective_exists_isMaxOn: proves primal objective attainment.
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/SelfImprovement/MatrixRealization/Canonical/StrongDuality/Separation.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_ExpansionHypercubeGraph_Defs_Core
import Definitions.Def_LIDT_MIPStarRE_LDT_ExpansionHypercubeGraph_Defs_Fourier
import Definitions.Def_LIDT_MIPStarRE_LDT_ExpansionHypercubeGraph_MatrixRealization_Core
import Definitions.Def_LIDT_MIPStarRE_LDT_ExpansionHypercubeGraph_MatrixRealization_TraceForms
import Definitions.Def_LIDT_MIPStarRE_LDT_GlobalVariance_Defs_Core
import Definitions.Def_LIDT_MIPStarRE_LDT_GlobalVariance_Defs_Families
import Definitions.Def_LIDT_MIPStarRE_LDT_GlobalVariance_Defs_Operators
import Definitions.Def_LIDT_MIPStarRE_LDT_MainInductionStep_Defs
import Definitions.Def_LIDT_MIPStarRE_LDT_MakingMeasurementsProjective_Defs
import Definitions.Def_LIDT_MIPStarRE_LDT_MakingMeasurementsProjective_NaimarkCore
import Definitions.Def_LIDT_MIPStarRE_LDT_MakingMeasurementsProjective_Projectivization
import Definitions.Def_LIDT_MIPStarRE_LDT_MakingMeasurementsProjective_Statements
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_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_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_SelfImprovement_Defs
import Definitions.Def_LIDT_MIPStarRE_LDT_SelfImprovement_MatrixRealization_Base
import Definitions.Def_LIDT_MIPStarRE_LDT_SelfImprovement_MatrixRealization_Canonical
import Definitions.Def_LIDT_MIPStarRE_LDT_SelfImprovement_MatrixRealization_CanonicalPrimal
import Definitions.Def_LIDT_MIPStarRE_LDT_SelfImprovement_MatrixRealization_Canonical_StrongDuality_Basic
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_Quantum_FiniteConicDuality
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 Definitions.Def_LIDT_MIPStarRE_Quantum_Measurement
import Mathlib
import Mathlib.Analysis.Complex.Basic
import Mathlib.Analysis.Convex.Cone.Dual
import Mathlib.Analysis.Fourier.ZMod
import Mathlib.Analysis.LocallyConvex.WithSeminorms
import Mathlib.Analysis.Matrix.Normed
import Mathlib.Analysis.MeanInequalitiesPow
import Mathlib.Probability.ProbabilityMassFunction.Integrals
import Mathlib.Probability.ProbabilityMassFunction.Monad
import Mathlib.Topology.Algebra.Module.ContinuousLinearMap.PiProd
import Mathlib.Topology.Bases
import Mathlib.Topology.Instances.Matrix
namespace LIDT
namespace MIPStarRE.LDT.Preliminaries
end MIPStarRE.LDT.Preliminaries
/-!
# Section 9 -- Canonical SDP separation argument
This module contains the closed-image-cone separation argument and the resulting
zero-duality-gap theorem for the canonical finite-dimensional matrix SDP.
## References
- `references/ldt-paper/self_improvement.tex`
-/
@[expose] public section
namespace MIPStarRE.LDT.SelfImprovement
open MIPStarRE.LDT
open MIPStarRE.LDT.ExpansionHypercubeGraph
open MIPStarRE.LDT.GlobalVariance
open MIPStarRE.LDT.MakingMeasurementsProjective
open Filter
open scoped BigOperators MatrixOrder _root_.Matrix ComplexOrder _root_.Matrix.Norms.Elementwise Topology
/-- The map sending a primal matrix to its constraint image and objective value. -/
noncomputable def matrixSdpCanonicalPrimalConeMap
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params) :
MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model) →L[ℝ]
MatrixOperator model.space × ℝ :=
(matrixSdpCanonicalConstraintOperatorCLM params model).prod
(matrixSdpCanonicalPrimalObjectiveCLM params model)
@[simp]
theorem matrixSdpCanonicalPrimalConeMap_apply
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params)
(X : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model)) :
matrixSdpCanonicalPrimalConeMap params model X =
(matrixSdpCanonicalConstraintOperator params model X,
Complex.re (Matrix.trace (matrixSdpCanonicalObjectiveOperator params model * X))) := by
simp [matrixSdpCanonicalPrimalConeMap, matrixSdpCanonicalPrimalObjectiveCLM]
/-- The closed image cone of positive semidefinite canonical primal matrices under
the constraint-objective map. -/
noncomputable def matrixSdpCanonicalPrimalImageCone
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params) :
ProperCone ℝ (MatrixOperator model.space × ℝ) :=
(MIPStarRE.Quantum.opNonnegativeProperCone
(matrixSdpCanonicalBlockHilbertSpace params model).carrier).map
(matrixSdpCanonicalPrimalConeMap params model)
/-- An actual positive semidefinite canonical primal matrix maps into the closed
canonical primal image cone. -/
theorem matrixSdpCanonicalPrimalImageCone_mem_of_nonnegative
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params)
(X : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model))
(hX : 0 ≤ X) :
(matrixSdpCanonicalConstraintOperator params model X,
Complex.re (Matrix.trace (matrixSdpCanonicalObjectiveOperator params model * X))) ∈
matrixSdpCanonicalPrimalImageCone params model := by
rw [← matrixSdpCanonicalPrimalConeMap_apply params model X]
rw [matrixSdpCanonicalPrimalImageCone, ProperCone.mem_map]
exact subset_closure <| by
have hXcone :
X ∈ (MIPStarRE.Quantum.opNonnegativeProperCone
(matrixSdpCanonicalBlockHilbertSpace params model).carrier).toPointedCone := by
change 0 ≤ X
exact hX
exact (PointedCone.mem_map).2 ⟨X, hXcone, rfl⟩
/-- A feasible canonical primal matrix maps to the image-cone point with
constraint component equal to the identity. -/
theorem matrixSdpCanonicalPrimalImageCone_mem_of_feasible
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params)
(X : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model))
(hX : MatrixSdpCanonicalPrimalFeasible params model X) :
((1 : MatrixOperator model.space),
Complex.re (Matrix.trace (matrixSdpCanonicalObjectiveOperator params model * X))) ∈
matrixSdpCanonicalPrimalImageCone params model := by
simpa [hX.constraintEqOne] using
matrixSdpCanonicalPrimalImageCone_mem_of_nonnegative params model X hX.nonnegative
/-- On the identity constraint fiber, the closed primal image cone contains
exactly the objective values of feasible canonical primal matrices. -/
theorem matrixSdpCanonicalPrimalImageCone_identity_mem_iff_exists_feasible_objective
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params) (t : ℝ) :
((1 : MatrixOperator model.space), t) ∈
matrixSdpCanonicalPrimalImageCone params model ↔
∃ X : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model),
MatrixSdpCanonicalPrimalFeasible params model X ∧
Complex.re (Matrix.trace
(matrixSdpCanonicalObjectiveOperator params model * X)) = t := by
classical
constructor
· intro hmem
rw [matrixSdpCanonicalPrimalImageCone, ProperCone.mem_map] at hmem
obtain ⟨u, hu_mem, hu_tendsto⟩ := mem_closure_iff_seq_limit.mp hmem
have hu_witness : ∀ n : ℕ,
∃ X : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model),
X ∈ (MIPStarRE.Quantum.opNonnegativeProperCone
(matrixSdpCanonicalBlockHilbertSpace params model).carrier).toPointedCone ∧
(matrixSdpCanonicalPrimalConeMap params model :
MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model) →ₗ[ℝ]
MatrixOperator model.space × ℝ) X = u n := by
intro n
exact PointedCone.mem_map.mp (hu_mem n)
choose X hXcone hmap using hu_witness
have hXnonneg : ∀ n : ℕ, 0 ≤ X n := by
intro n
have hx := hXcone n
change 0 ≤ X n at hx
exact hx
have hmapCLM : ∀ n : ℕ,
matrixSdpCanonicalPrimalConeMap params model (X n) = u n := by
intro n
simpa using hmap n
let dimR : ℝ := Fintype.card model.space.carrier
let R : ℝ := dimR + 1
let s : Set (MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model)) :=
{Y | 0 ≤ Y ∧ ‖Y‖ ≤ R}
have hsBounded : Bornology.IsBounded s := by
rw [isBounded_iff_forall_norm_le]
exact ⟨R, fun Y hY => hY.2⟩
have hcoords :=
(Prod.tendsto_iff u ((1 : MatrixOperator model.space), t)).mp hu_tendsto
have htrace_cont : Continuous fun Y : MatrixOperator model.space =>
Complex.re (Matrix.trace Y) :=
Complex.continuous_re.comp continuous_id.matrix_trace
have htrace_tendsto_u :
Tendsto (fun n : ℕ => Complex.re (Matrix.trace ((u n).1))) atTop
(𝓝 (Fintype.card model.space.carrier : ℝ)) := by
change Tendsto (((fun Y : MatrixOperator model.space =>
Complex.re (Matrix.trace Y)) ∘ fun n : ℕ => (u n).1)) atTop
(𝓝 (Fintype.card model.space.carrier : ℝ))
simpa [Matrix.trace_one] using
(htrace_cont.tendsto (1 : MatrixOperator model.space)).comp hcoords.1
have htrace_tendsto :
Tendsto
(fun n : ℕ => Complex.re (Matrix.trace
(matrixSdpCanonicalConstraintOperator params model (X n)))) atTop
(𝓝 dimR) := by
have htrace_eq :
(fun n : ℕ => Complex.re (Matrix.trace
(matrixSdpCanonicalConstraintOperator params model (X n)))) =
fun n : ℕ => Complex.re (Matrix.trace ((u n).1)) := by
funext n
rw [← hmapCLM n]
rw [matrixSdpCanonicalPrimalConeMap_apply]
rw [htrace_eq]
simpa [dimR] using htrace_tendsto_u
have hdim_lt_R : dimR < R := by
simp [R]
have heventTrace : ∀ᶠ n : ℕ in atTop,
Complex.re (Matrix.trace
(matrixSdpCanonicalConstraintOperator params model (X n))) < R :=
htrace_tendsto.eventually (Iio_mem_nhds hdim_lt_R)
have heventS : ∀ᶠ n : ℕ in atTop, X n ∈ s := by
filter_upwards [heventTrace] with n hn
refine ⟨hXnonneg n, ?_⟩
exact (matrixSdpCanonicalNonnegative_norm_le_constraint_trace_re
params model (X n) (hXnonneg n)).trans (le_of_lt hn)
obtain ⟨X₀, _hX₀closure, k, hkmono, hX₀tendsto⟩ :=
tendsto_subseq_of_frequently_bounded hsBounded heventS.frequently
have hX₀nonneg : 0 ≤ X₀ :=
(MIPStarRE.Quantum.isClosed_op_nonnegative
(ι := (matrixSdpCanonicalBlockHilbertSpace params model).carrier)).mem_of_tendsto
hX₀tendsto (Eventually.of_forall fun n => hXnonneg (k n))
have hmap_tendsto_X₀ :
Tendsto
(fun n : ℕ => matrixSdpCanonicalPrimalConeMap params model (X (k n)))
atTop (𝓝 (matrixSdpCanonicalPrimalConeMap params model X₀)) :=
((matrixSdpCanonicalPrimalConeMap params model).continuous.tendsto X₀).comp
hX₀tendsto
have hu_subseq_tendsto :
Tendsto (u ∘ k) atTop (𝓝 ((1 : MatrixOperator model.space), t)) :=
hu_tendsto.comp hkmono.tendsto_atTop
have hmap_tendsto_identity :
Tendsto
(fun n : ℕ => matrixSdpCanonicalPrimalConeMap params model (X (k n)))
atTop (𝓝 ((1 : MatrixOperator model.space), t)) := by
have hfun :
(fun n : ℕ => matrixSdpCanonicalPrimalConeMap params model (X (k n))) =
u ∘ k := by
funext n
exact hmapCLM (k n)
rw [hfun]
exact hu_subseq_tendsto
have hlimit :
matrixSdpCanonicalPrimalConeMap params model X₀ =
((1 : MatrixOperator model.space), t) :=
tendsto_nhds_unique hmap_tendsto_X₀ hmap_tendsto_identity
have hconstraint :
matrixSdpCanonicalConstraintOperator params model X₀ =
(1 : MatrixOperator model.space) := by
have := congrArg Prod.fst hlimit
simpa [matrixSdpCanonicalPrimalConeMap_apply] using this
have hobjective :
Complex.re (Matrix.trace
(matrixSdpCanonicalObjectiveOperator params model * X₀)) = t := by
have := congrArg Prod.snd hlimit
simpa [matrixSdpCanonicalPrimalConeMap_apply] using this
exact ⟨X₀, ⟨hX₀nonneg, hconstraint⟩, hobjective⟩
· rintro ⟨X, hX, hobj⟩
simpa [hobj] using matrixSdpCanonicalPrimalImageCone_mem_of_feasible params model X hX
/-- A feasible functional dual bound on the canonical primal image cone gives a
paper-form dual feasible matrix through the Hermitian trace-pairing
representation. -/
theorem matrixSdpCanonicalTracePairingDualMatrix_dualFeasible_of_conicFunctionalDualFeasible
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params)
(ψ : StrongDual ℝ (MatrixOperator model.space))
(hψ : ∀ z ∈ matrixSdpCanonicalPrimalImageCone params model, z.2 ≤ ψ z.1) :
∀ g : Polynomial params,
0 ≤ matrixSdpDualSlackOperator params model
(MIPStarRE.Quantum.hermitianTracePairingMatrixOfRealCLM ψ) g := by
let W := MIPStarRE.Quantum.hermitianTracePairingMatrixOfRealCLM ψ
refine matrixSdpCanonicalDualConstraint_nonneg_of_trace_pairing_nonneg
params model W ?_ ?_
· exact matrixSdpCanonicalDualOperator_sub_objectiveOperator_isHermitian
params model W (MIPStarRE.Quantum.hermitianTracePairingMatrixOfRealCLM_isHermitian ψ)
· intro X hX
have hle :
Complex.re (Matrix.trace
(matrixSdpCanonicalObjectiveOperator params model * X)) ≤
Complex.re (Matrix.trace
(matrixSdpCanonicalDualOperator params model W * X)) := by
have hmem :=
matrixSdpCanonicalPrimalImageCone_mem_of_nonnegative params model X hX
have hfunctional := hψ
(matrixSdpCanonicalConstraintOperator params model X,
Complex.re (Matrix.trace
(matrixSdpCanonicalObjectiveOperator params model * X))) hmem
have hconstraintHerm :
(matrixSdpCanonicalConstraintOperator params model X).IsHermitian :=
matrixSdpCanonicalConstraintOperator_isHermitian_of_nonnegative params model hX
rw [MIPStarRE.Quantum.hermitianTracePairingMatrixOfRealCLM_apply_of_isHermitian
ψ hconstraintHerm] at hfunctional
rwa [← matrixSdpCanonicalDualOperator_trace_constraint params model X W]
at hfunctional
rw [Matrix.sub_mul, Matrix.trace_sub, Complex.sub_re]
exact sub_nonneg.mpr hle
/-- A paper-form dual minimizer is also minimal among feasible functional dual
bounds on the canonical primal image cone. -/
theorem matrixSdpCanonicalFunctionalDualObjective_min_of_matrixDualMin
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params)
(Z : MatrixOperator model.space)
(hZmin : ∀ W : MatrixOperator model.space,
(∀ g : Polynomial params,
0 ≤ matrixSdpDualSlackOperator params model W g) →
matrixSdpDualObjective model Z ≤ matrixSdpDualObjective model W) :
∀ ψ : StrongDual ℝ (MatrixOperator model.space),
(∀ z ∈ matrixSdpCanonicalPrimalImageCone params model, z.2 ≤ ψ z.1) →
matrixSdpDualObjective model Z ≤ ψ (1 : MatrixOperator model.space) := by
intro ψ hψ
let W := MIPStarRE.Quantum.hermitianTracePairingMatrixOfRealCLM ψ
have hWdual :
∀ g : Polynomial params, 0 ≤ matrixSdpDualSlackOperator params model W g := by
simpa [W] using
matrixSdpCanonicalTracePairingDualMatrix_dualFeasible_of_conicFunctionalDualFeasible
params model ψ hψ
have hd_le_W := hZmin W hWdual
have hI : (1 : MatrixOperator model.space).IsHermitian := Matrix.isHermitian_one
have hWobj : matrixSdpDualObjective model W = ψ (1 : MatrixOperator model.space) := by
have hψI :=
MIPStarRE.Quantum.hermitianTracePairingMatrixOfRealCLM_apply_of_isHermitian ψ hI
simpa [matrixSdpDualObjective, W] using hψI.symm
simpa [hWobj] using hd_le_W
/-- The canonical primal objective attains a maximum on the feasible set. -/
theorem matrixSdpCanonicalPrimalObjective_exists_isMaxOn
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params) :
∃ X : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model),
MatrixSdpCanonicalPrimalFeasible params model X ∧
∀ Y : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model),
MatrixSdpCanonicalPrimalFeasible params model Y →
Complex.re (Matrix.trace (matrixSdpCanonicalObjectiveOperator params model * Y)) ≤
Complex.re (Matrix.trace (matrixSdpCanonicalObjectiveOperator params model * X)) := by
obtain ⟨X, hX, hmax⟩ :=
(matrixSdpCanonicalPrimalFeasible_isCompact params model).exists_isMaxOn
(Set.nonempty_def.mpr (matrixSdpCanonicalPrimalFeasible_nonempty params model))
(continuous_matrixSdpCanonicalPrimalObjective params model).continuousOn
exact ⟨X, hX, fun Y hY => hmax hY⟩
/-- The canonical primal maximizer and dual minimizer exist, and their attained
values satisfy weak duality. This packages the attained optima but is not the
zero-gap strong-duality theorem. -/
theorem matrixSdpCanonicalPrimalDualOptima_exist
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params) :
∃ X : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model),
∃ Z : MatrixOperator model.space,
MatrixSdpCanonicalPrimalFeasible params model X ∧
(∀ g : Polynomial params,
0 ≤ matrixSdpDualSlackOperator params model Z g) ∧
(∀ Y : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model),
MatrixSdpCanonicalPrimalFeasible params model Y →
Complex.re (Matrix.trace
(matrixSdpCanonicalObjectiveOperator params model * Y)) ≤
Complex.re (Matrix.trace
(matrixSdpCanonicalObjectiveOperator params model * X))) ∧
(∀ W : MatrixOperator model.space,
(∀ g : Polynomial params,
0 ≤ matrixSdpDualSlackOperator params model W g) →
matrixSdpDualObjective model Z ≤ matrixSdpDualObjective model W) ∧
Complex.re (Matrix.trace
(matrixSdpCanonicalObjectiveOperator params model * X)) ≤
matrixSdpDualObjective model Z := by
obtain ⟨X, hX, hXmax⟩ :=
matrixSdpCanonicalPrimalObjective_exists_isMaxOn params model
obtain ⟨Z, hZ, hZmin⟩ :=
matrixSdpCanonicalDualObjective_exists_isMinOn params model
refine ⟨X, Z, hX, hZ, hXmax, hZmin, ?_⟩
exact matrixSdpCanonicalWeakDuality params model X hX Z
(matrixSdpCanonicalDualConstraint_nonneg_of_dualFeasible params model Z hZ)
/-- The canonical matrix SDP has a primal-dual optimal pair with zero duality
gap.
This is the finite-dimensional strong-duality conclusion used in the paper's
Section 9 SDP argument. The proof combines compact attainment, separation of
the closed primal image cone, and the trace-pairing representation of the
normalized separator; it does not add an auxiliary dominance hypothesis on the
dual matrix. -/
theorem matrixSdpCanonicalStrongDuality
(params : Parameters) [FieldModel params.q]
(model : MatrixSdpRealization params) :
∃ X : MatrixOperator (matrixSdpCanonicalBlockHilbertSpace params model),
∃ Z : MatrixOperator model.space,
MatrixSdpCanonicalPrimalFeasible params model X ∧
(∀ g : Polynomial params,
0 ≤ matrixSdpDualSlackOperator params model Z g) ∧
Complex.re (Matrix.trace
(matrixSdpCanonicalObjectiveOperator params model * X)) =
matrixSdpDualObjective model Z := by
obtain ⟨X, Z, hX, hZ, hXmax, hZmin, hweak⟩ :=
matrixSdpCanonicalPrimalDualOptima_exist params model
let p : ℝ := Complex.re (Matrix.trace
(matrixSdpCanonicalObjectiveOperator params model * X))
let d : ℝ := matrixSdpDualObjective model Z
have hp_eq_d : p = d := by
refine MIPStarRE.Quantum.conic_primalValue_eq_dualValue_of_fiber_max_dual_min
(C := matrixSdpCanonicalPrimalImageCone params model)
(y := (1 : MatrixOperator model.space)) (p := p) (d := d) ?_ ?_ ?_ ?_
· simpa [p] using matrixSdpCanonicalPrimalImageCone_mem_of_feasible params model X hX
· intro t hmem
obtain ⟨Y, hY, hYobj⟩ :=
(matrixSdpCanonicalPrimalImageCone_identity_mem_iff_exists_feasible_objective
params model t).mp hmem
have ht_le : t ≤ p := by
simpa [p, hYobj] using hXmax Y hY
exact ht_le
· intro ψ hψ
simpa [d] using
matrixSdpCanonicalFunctionalDualObjective_min_of_matrixDualMin
params model Z hZmin ψ hψ
· simpa [p, d] using hweak
refine ⟨X, Z, hX, hZ, ?_⟩
simpa [p, d] using hp_eq_d
end MIPStarRE.LDT.SelfImprovement
end
end LIDT
Source