Theorem 3.1 — random purification from dimension reduction
OpenRandomDimReduction.Channel.Supplement.theorem_3_1p2o-batch-stoc27sp2o-gran-per-chapterp2o-plan-paperp2o-v1quantum-informationrepresentation-theory
Let be positive integers with . Use the three explicit maps of Eqs. (4)–(5), defined from the unnormalized reference-first Choi operator of the published . The map is complex-linear, completely positive and trace preserving on every operator.
For every density operator on of actual matrix rank at most , and every fixed unit purification of ,
Here is normalized Haar on , a single sampled is used in all copies, and paired registers are canonically regrouped. This identifies the explicit Choi extension as a rank-dependent random purification channel.
Differences from the source:
- encoding: the explicit function P^CPTP is a quantum channel → a bundled complex-linear CPTP map agrees with the explicit Eq. (5) function on every operator; reason: the equality makes linearity and full-domain CPTP validity conclusions rather than assumptions
- encoding: the n-fold tensor power of a purification on paired registers → canonical reindexing from (A tensor B)^n to A^n tensor B^n; reason: both sides of Eq. (6) must use the same physical registers
Preamble
import Definitions.Def_RandomDimReduction_Channel_Supplement_Purification set_option autoImplicit false noncomputable section open Matrix open scoped Kronecker Matrix ComplexConjugate Matrix.Norms.Elementwise
Formal statement
namespace RandomDimReduction.Channel.Supplement
/-- Theorem 3.1, pp.9–10, Eq.(6). Linearity and CPTP validity of the explicit Eq.(5)
function are conclusions, expressed by its equality with a bundled CPTP linear map. -/
theorem theorem_3_1 (n k d : ℕ) (hn : 0 < n) (hk : 0 < k) (hkd : k ≤ d) :
(∃ Phi : Op (Word n d) →ₗ[ℂ] Op (Word n d × Word n k),
(∀ X, Phi X = purificationCPTP n k d X) ∧ WildeQIT.EntBreaking.IsQuantumChannel Phi) ∧
∀ rho : Op (Fin d), MixedCloning.Optimal.IsState rho → rho.rank ≤ k →
∀ phi : EuclideanSpace ℂ (Fin d × Fin k), IsPurification rho phi →
purificationCPTP n k d (tensor n rho) = purificationCP n k d (tensor n rho) ∧
purificationCP n k d (tensor n rho) = purification n k d (tensor n rho) ∧
purification n k d (tensor n rho) = randomPurificationAverage n phi := by sorry
end RandomDimReduction.Channel.Supplement
Source
Angus Lowe, Xinyu Tan, Random dimension reduction and learning symmetric properties of quantum states, STOC 2027 submission #994 (October 2026), https://github.com/sigact-stoc/stoc2027/releases/download/papers-2/994.pdf, Theorem 3.1, p. 9