Alphabeta Math
LemmaStatement: Literature-sourcedProof: AI-generatedPipeline-generatedjudge pass (gpt-6-luna)
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

Unitary hybrids on a preserved clean-work subspace

Statement

Let J:H→K be an isometry of nonzero finite-dimensional complex Hilbert spaces, V a unitary on H, and U a unitary on K. If ∥UJ−JV∥op≤η, then ∥U∗J−JV∗∥op≤η. The same inequalities hold after adjoining an arbitrary finite reference space or coherently controlling U and V against identity.

More generally, let U1,…,Us be actual unitaries on K and V1,…,Vs ideal unitaries on H, with ∥UjJ−JVj∥op≤ηj. Then ∥Us⋯U1J−JVs⋯V1∥op≤∑j=1sηj. The empty products give zero error. Each bound concerns the clean input subspace ran⁡J; none assumes that an actual intermediate state belongs to that subspace or claims a full-space approximation on arbitrary dirty-work inputs.

Facts & Assumptions

Given: The finite spaces, isometry, unitaries and error bounds in the Statement.

[F1]

The operator norm is submultiplicative and satisfies ||T^*T|| = ||T||^2 gives the operator-norm product inequality. Unitary multiplication preserves vector norms and hence preserves the operator norm of any compatible matrix.

[F2]

Every linear map between finite-dimensional real or complex inner product spaces admits a singular value decomposition diagonalizes the squared norm of a finite rectangular map in orthonormal bases.

Proof

1.1givenF1algebra

The exact identity U∗(JV−UJ)V∗=U∗J−JV∗, together with unitary invariance, gives ∥U∗J−JV∗∥op=∥JV−UJ∥op≤η. Thus reversing the actual circuit supplies the inverse clean-input bound, without applying a forward error promise to dirty work.

2.1givenF2step 1.1algebra

If E:H→K has singular values sj, [F2] gives orthonormal bases in which ∥Eψ∥2=∑jsj2∣ψj∣2. Tensor those bases with any orthonormal reference basis; for Ψ=∑j,rcj,rej⊗fr, ∥(E⊗I)Ψ∥2=∑j,rsj2∣cj,r∣2≤∥E∥op2∥Ψ∥2. A top singular vector tensored with any unit reference vector attains equality, so ∥E⊗I∥op=∥E∥op. If the reference space is zero, the amplified map is zero and its bound is immediate. For coherent control against identity, the difference between the actual and ideal clean-input maps is zero on the control-zero summand and E on the orthogonal control-one summand; its norm is at most ∥E∥op. These statements apply to step 1.1 as well as to the original forward bound.

3.1givenF1algebracases∎

For s≥1, expansion and cancellation give Us⋯U1J−JVs⋯V1=∑j=1sUs⋯Uj+1(UjJ−JVj)Vj−1⋯V1. Every factor outside the local difference is unitary, so each summand has norm at most ηj. The triangle inequality yields the claimed sum. On the right of each local difference are only ideal system unitaries followed by J, so this expansion evaluates that difference on clean ideal inputs. The actual later unitaries on the left propagate the error without any cleanliness assumption. For s=0, both products are identity maps, so the difference is zero.

Depends on

Used by

Dependency tree · two levels

9 results within two dependency steps of this one, each drawn at its shortest distance from it. An arrow runs from a result to what uses it, so the chart reads left to right and ends at this result, which carries a heavier outline. Every node is a link to that result. Click elsewhere on the chart to enlarge it.

Sources