arXiv:2610.27394
Choi-Jamiołkowski isomorphism(s) and channel/state duality, naturally!
Paul E. Lammert
Abstract
Provided a maximally entangled state vector in ${\mathcal{H}}\otimes{\mathcal{H}}$, the Choi-Jamiolkowski isomorphism of quantum information theory (or channel/state duality) produces a mapping of ${\mathsf CP}({\mathcal{H}},{\mathcal{K}})$ (completely positive maps of density operators on ${\mathcal{H}}$ tothose on ${\mathcal{K}}$) onto the density operators on ${\mathcal{K}}\otimes{\mathcal{H}}$, which extends to a unitary map. Here, ${\mathcal{H}}$ and ${\mathcal{K}}$ are {finite-dimensional} Hilbert spaces.The nature of this construction, the form of the associated formulas, and the possibility of versions which do not require an entangled vector seed, are illuminated from the perspective of {naturality}, in the sense of Eilenberg and MacLane.There is a unique such correspondence (``Choi map'') natural in both ${\mathcal{H}}$ and ${\mathcal{K}}$ which produces density operators on either ${\mathsf L}({\mathcal{H}},{\mathcal{K}})$ (Hilbert-Schmidt inner product) or ${\mathcal{K}}\otimes{\mathcal{H}}^*$ instead of ${\mathcal{K}}\otimes{\mathcal{H}}$. In particular, no entangled state input is required. There are no such binatural maps carried on ${\mathcal{K}}\otimes{\mathcal{H}}$, but those natural in only ${\mathcal{K}}$ are canonically associated with unitary maps ${\mathcal{H}}\rightarrow{\mathcal{H}}^*$ and derive from the binatural maps. The construction recovers the standard Choi-Jamiolkowski formulae and shows how they express naturality.
AI Summary
The paper asks which features of the Choi–Jamiołkowski correspondence are canonical, and which depend on choosing a maximally entangled vector or identifying an input Hilbert space with its dual. It formulates this question using natural transformations on finite-dimensional Hilbert spaces, allowing arbitrary linear maps between the underlying spaces. A “Choi map” is defined as a Hilbert–Schmidt unitary from the space of superoperators to an operator space that sends completely positive maps to positive operators. Two constructions are developed: a reshuffling map $\mathcal J$ whose carrier is $\mathsf L(\mathcal H,\mathcal K)$, and a map $\mathcal C$ whose carrier is $\mathcal K\otimes\mathcal H^*$. Each is shown to be the unique Choi map natural in both input and output spaces for its specified carrier. No such binatural Choi map exists on $\mathcal K\otimes\mathcal H$. The conventional correspondences are recovered by choosing a unitary $V:\mathcal H\to\mathcal H^*$; these retain naturality in the output space. The proofs combine canonical operator–tensor identifications, tensor-factor reshuffling, and naturality squares. Positivity is established through the equivalence between $N$-positivity of a map and nonnegative quadratic forms of its reshuffling on operators of rank at most $N$. Uniqueness arguments reduce naturality to behavior on one-dimensional spaces. The treatment is finite-dimensional because it uses Hilbert–Schmidt structures. Its unitary normalization sends trace-preserving maps to operators of trace $\dim\mathcal H$, rather than trace one; dimension-dependent rescaling sacrifices the stated input-space naturality. Conversely, arbitrary positive operators need not correspond to trace-preserving maps, even after rescaling.
Error found: The abstract conflates positive operators with density operators: the body's Choi maps produce trace dim H on channels, and normalized channel images do not exhaust all density operators. With a normalized maximally entangled seed, the conventional Choi map scales Hilbert–Schmidt norms by 1/dim H and therefore is not unitary when dim H>1. The claimed one-to-one parametrization by unitaries V:H→H* is redundant because multiplying V by a global phase leaves C_V unchanged. The appendix's positivity-preserving-unitary lemma and Proposition 1 are false because matrix transposition preserves positivity and the Hilbert–Schmidt inner product without being unitary conjugation. The proof excluding binatural maps with carrier K⊗H drops the phase ambiguity between operators inducing the same conjugation map, leaving its stated implication unjustified.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.