ASPBE Lean Blueprint

4.3. Evolved optimal-control completion🔗

Theorem4.3.1
uses 0used by 1L∃∀N

The evolved child keeps the equality flag and follows it with three parallel bit flips. The reduced active-register map remains a finite permutation.

Lean code for Theorem4.3.11 theorem
  • complete
    theorem QuantumBlockEncoding.OptimalControl.evolvedEqFlipImage_isPermutation :
      QuantumBlockEncoding.OptimalControl.IsPermutation
        QuantumBlockEncoding.OptimalControl.evolvedEqFlipImage
    theorem QuantumBlockEncoding.OptimalControl.evolvedEqFlipImage_isPermutation :
      QuantumBlockEncoding.OptimalControl.IsPermutation
        QuantumBlockEncoding.OptimalControl.evolvedEqFlipImage
    The evolved reduced active-register map is a permutation. 
Theorem4.3.2
uses 1used by 1L∃∀N

The corresponding 16\times16 matrix closes both rational Gram-matrix identities.

Lean code for Theorem4.3.21 theorem
  • complete
    theorem QuantumBlockEncoding.OptimalControl.evolvedEqFlipUnitary_isRationalOrthogonal :
      QuantumBlockEncoding.OptimalControl.IsRationalOrthogonal
        QuantumBlockEncoding.OptimalControl.evolvedEqFlipUnitary
    theorem QuantumBlockEncoding.OptimalControl.evolvedEqFlipUnitary_isRationalOrthogonal :
      QuantumBlockEncoding.OptimalControl.IsRationalOrthogonal
        QuantumBlockEncoding.OptimalControl.evolvedEqFlipUnitary
    The evolved matrix is a concrete rational unitary matrix in the project-local
    real/permutation sense: both its column and row Gram matrices are identity.
    
Theorem4.3.3
uses 1used by 1L∃∀N

Although the off-block unitary completion differs from the parent construction, its clean block is the same transfer operator.

Lean code for Theorem4.3.31 theorem
  • complete
    theorem QuantumBlockEncoding.OptimalControl.evolvedEqFlipUnitary_cleanBlock
      (row col : Fin 8) :
      QuantumBlockEncoding.OptimalControl.evolvedEqFlipUnitary
          (QuantumBlockEncoding.OptimalControl.cleanIndex row)
          (QuantumBlockEncoding.OptimalControl.cleanIndex col) =
        QuantumBlockEncoding.OptimalControl.exampleOperator row col
    theorem QuantumBlockEncoding.OptimalControl.evolvedEqFlipUnitary_cleanBlock
      (row col : Fin 8) :
      QuantumBlockEncoding.OptimalControl.evolvedEqFlipUnitary
          (QuantumBlockEncoding.OptimalControl.cleanIndex
            row)
          (QuantumBlockEncoding.OptimalControl.cleanIndex
            col) =
        QuantumBlockEncoding.OptimalControl.exampleOperator
          row col
    The evolved concrete matrix has the required clean block. 
Definition4.3.4
uses 1used by 1L∃∀N

The depth-two logical construction is packaged as a verified operator block encoding with an explicit circuit transcript, schedule, resource count, and both semantic proofs.

Lean code for Definition4.3.41 definition
  • def QuantumBlockEncoding.OptimalControl.evolvedEqFlipVerified :
      QuantumBlockEncoding.VerifiedOperatorBlockEncoding  3
    def QuantumBlockEncoding.OptimalControl.evolvedEqFlipVerified :
      QuantumBlockEncoding.VerifiedOperatorBlockEncoding
         3
    Verified concrete depth-2 block encoding for `E_1`. 
Definition4.3.5
uses 1used by 0L∃∀N

The exact evolved solution is reused as the baseline for any later approximate optimization.

Lean code for Definition4.3.51 definition
  • def QuantumBlockEncoding.OptimalControl.evolvedEqFlipZeroErrorApprox :
      QuantumBlockEncoding.VerifiedApproximateOperatorBlockEncoding  3
    def QuantumBlockEncoding.OptimalControl.evolvedEqFlipZeroErrorApprox :
      QuantumBlockEncoding.VerifiedApproximateOperatorBlockEncoding
         3
    The exact evolved candidate is also a zero-error approximate block encoding.
    This is the Lean anchor for Scenario 1 of the adaptive exact-to-approximate
    policy: after exact convergence, approximate search may continue, but the
    current champion already satisfies every nonnegative requested tolerance.