bibkey: johnston2010generalized authors: Nathaniel Johnston and David W. Kribs year: 2011 title: Generalized Multiplicative Domains and Quantum Error Correction doi: 10.1090/S0002-9939-2010-10556-7 url: https://arxiv.org/abs/1004.5112v1 claim: A unital completely positive map has a multiplicative domain characterized by its two square equalities; its use for Schur channels retains the unital and trace-preserving hypotheses. strata_touched: [] license: citation-only triage: anchor
Multiplicative-domain criterion consumed by calibrated five-label blocks
Primary text: arXiv:1004.5112v1, section 3, equation (3), Theorem 3; the preprint is from 2010. Published article: Proceedings of the American Mathematical Society 139, 627–639 (2011). The consumer is Hidden Relations, C-29.5.
For a unital completely positive map , the cited classical criterion is
On this algebra respects multiplication and adjoints. The paper recalls this criterion as Theorem 3; its generalized-domain theorems have additional hypotheses. In particular, Theorems 5 and 7 require a unital CPTP map. Theorem 11 instead treats correctable subsystems of a specified code space using an actual representation. These statements are not an unconditional convergence theorem for repeated channels.
For the consumer’s correlation-matrix Schur map , complete positivity and unit diagonal imply unitality and trace preservation. The two Schwarz defects are positive, and each has trace
Because a positive matrix of trace zero is zero, the cited criterion says that the multiplicative domain consists exactly of entries with . Common unit records make that relation an equivalence relation. In each such class the channel acts by a diagonal unitary; its inverse is an actual correction only when the phase calibration and gate are supplied. Raw fixed points instead require on each nonzero matrix entry.
The five-label consumer applies this distinction to an actual square with an block and three singleton classes. It derives the repeated Schur formula and its phase-corrected limit from matrix-unit eigenvalues and fresh independent environments. Those conclusions are not attributed to this paper’s general theorem. Positive class mass is required before normalizing a conditional block, and class weights remain part of the state.