Choi Dataset: Quantum Channel Theory

Updated 7 October 2025

Choi Dataset is a collection of mathematical structures and operator representations that map quantum channels via the Jamiołkowski isomorphism.
It provides essential criteria for testing complete positivity, entanglement detection, and resource-theoretic properties in both finite and infinite dimensions.
The dataset underpins practical applications such as non-Markovianity diagnostics, quantum protocol optimization, and experimental channel validation.

The Choi Dataset refers collectively to the mathematical structures, operational definitions, and computational strategies associated with Choi matrices and Choi–Jamiołkowski states, which represent linear maps (i.e. quantum channels and operations) as operators on bipartite spaces or, more generally, tensor products of Hilbert spaces. These objects underpin complete positivity checks, entanglement criteria, resource theory characterizations, non-Markovianity witnesses, and the operational capacity of quantum channels. The dataset encompasses both finite and infinite-dimensional variants, generalizations to resource-theoretic contexts, and measures derived from Choi-state moments, entropy, and rank.

1. Mathematical Foundations of the Choi–Jamiołkowski Isomorphism

The Choi–Jamiołkowski isomorphism provides a bijection between completely positive maps $\Phi$ from matrix algebra %%%%1%%%% to $M_m$ and positive semidefinite matrices in $M_n \otimes M_m$ . For finite dimensions, the standard Choi matrix is given by

$C_\Phi = \sum_{i, j = 1}^{n} e_{ij} \otimes \Phi(e_{ij}),$

where $\{e_{ij}\}$ are matrix units in $M_n$ . Complete positivity of $\Phi$ is equivalent to positivity of $C_\Phi$ (Kye, 2022).

Generalizations permit constructing Choi matrices using alternative bases $\{b_{ij}\}$ ; however, the correspondence between complete positivity and positivity of $C_B = \sum b_{ij} \otimes \Phi(b_{ij})$ holds if and only if $M_B = [C_B][C_B]^T$ is the adjoint action of a nonsingular matrix $s$ , i.e., $M_B = \mathrm{Ad}_s(x) = s^* x s$ (Kye, 2022). This equivalence is necessary and sufficient, thus constraining the basis flexibility.

More broadly, Choi-matrix variants can be built from non-degenerate bilinear forms on the domain, with the standard choice corresponding to the symmetric inner product on matrix algebras (Han et al., 2023). The duality extends to higher-order properties: $k$ -positive maps correspond to $k$ -block-positive Choi matrices, and $k$ -superpositive maps to matrices with Schmidt number $\leq k$ (Han et al., 2023).

In the infinite-dimensional case, Choi–Jamiołkowski correspondence requires refined definitions on dense subspaces, utilising maximally entangled vectors $|\Omega) = \sum_{i=1}^\infty |i\rangle \otimes |i\rangle$ . The CJ form $\Omega_\Phi$ , defined as

$\Omega_\Phi(\psi_B \otimes \psi_A; \psi'_B \otimes \psi'_A) = \langle \psi_A | \Phi(|\overline{\psi}_B\rangle \langle \overline{\psi}'_B|) | \psi'_A \rangle,$

extends to bounded operators if appropriate domain and noise conditions are met (Holevo, 2010, Han et al., 2023). For von Neumann factors, two objects—bounded operator $C_\phi$ and trace-class operator $D_\phi$ derived from projections—play the analog of the Choi matrix, both providing necessary and sufficient conditions for complete positivity (Han et al., 2023).

2. Operational Criteria and Applications

Choi matrices serve as an essential tool for determining the validity and structure of quantum channels and operations:

Entanglement detection: Positive maps that are not completely positive (indecomposable) can be used, via the Choi–Jamiołkowski isomorphism, to construct entanglement witnesses capable of detecting PPT entangled states (Scala et al., 2023).
Optimality and extremality: Generalized Choi maps, particularly in $M_3$ , can be classified as optimal (no subtraction of nonzero completely positive maps preserves positivity) or extremal (no nontrivial additive decompositions remain positive). Optimality is typically easier to certify and crucial for sharp entanglement witnessing; the spanning property links product-vector kernels of $\Phi_W(\psi \psi^\dagger)$ to optimality (Scala et al., 2023).
Schmidt-number and entanglement-breaking criteria: In both finite and infinite dimensions, the Schmidt number of Choi matrices encodes the entanglement properties of corresponding maps. A map is $k$ -superpositive if its Choi matrix is contained in the cone generated by rank $\leq k$ vectors, and is $k$ -partially entanglement breaking (Han et al., 2023, Han et al., 2023).

3. Non-Markovianity Witnesses and Dynamical Diagnostics

Quantum dynamics can depart from Markovian (memoryless) behavior; this is reflected in Choi-state properties:

Linear entropy scheme: For a quantum map $\Lambda$ , the Choi state $C_\Lambda = (I \otimes \Lambda)(|\psi\rangle \langle \psi|)$ is positive semidefinite under Markovian evolution. The linear entropy

$S_{\ell}(\rho) = \frac{d}{d-1} (1 - \mathrm{Tr}(\rho^2))$

is nonnegative for CP maps and becomes negative when CP divisibility breaks (non-Markovian regime) (Zheng et al., 2019). Rényi entropies generalize this criterion and offer adjustable sensitivity across eigenvalue spectra.

Moment-based criteria: Assessing higher moments $r_n = \mathrm{Tr}(C_\Lambda^n)$ of the Choi state, violation of $(r_2)^2 \leq r_3$ signals non-Markovianity (Mallick et al., 2023). Complementary measures for unital dynamics define instantaneous non-Markovianity via derivatives of moment deviations, directly benchmarking the non-Markovian behavior and relating to established measures like RHP.

This approach allows efficient experimental certification, including shadow tomography protocols, and facilitates construction of datasets systematically cataloging non-Markovianity properties.

4. Resource Theoretic Perspectives

In resource theories (coherence, entanglement, or more general quantum resources), the structure and properties of the Choi matrix are central:

Choi-defined resource theories: An operation is "free" if and only if its renormalized Choi matrix is a free state; the set of free operations coincides with completely resource-non-generating maps (Zanoni et al., 2024).
Coherence in quantum operations: The Choi–Jamiołkowski isomorphism supports a full resource theory approach to operational coherence. The phase-out superoperation $\Theta(\Phi) = \Delta_o \circ \Phi \circ \Delta_i$ diagonalizes the Choi state, defining maximally incoherent, nonactivating coherent, and de-phase incoherent superoperations with important closure properties (Wang et al., 2022). The fidelity coherence measure $M_f(\Phi)$ quantifies operational coherence and is analytically computable for single-qubit unitaries: $M_f(U) = \min \{ 1 - |\cos(\gamma)|^2,~ 1 - |\sin(\gamma)|^2 \}.$

These frameworks enable transfer of structural resource-theoretic monotones and conversion bounds directly to the space of operations via their Choi representations.

5. Infinite-Dimensional Extensions

In infinite-dimensional Hilbert spaces, the Choi–Jamiołkowski correspondence is nuanced. The CJ form is defined as a positive semidefinite sesquilinear form on a dense subspace; in favorable cases, it may be closable into a bounded operator on $H_B \otimes H_A$ (Holevo, 2010, Bolanos-Servin et al., 2013). Entanglement-breaking channels admit separable CJ operators: $\Omega_\Phi = \int_X \rho_B(x) \otimes M_A(dx).$ Bosonic Gaussian channels admit explicit integral representations in terms of Weyl operators and covariance matrices; a Williamson-type decomposition further stratifies CJ forms into quantum, classical, pure, and trivial noise sectors. Boundedness of the expanded CJ operator is determined by noise matrix nondegeneracy and domain extensibility.

The p–Choi–Jamiołkowski state formalism anchors definitions on a fixed faithful state $p$ , controlling divergences and enabling operational entropy production rate assessment through the von Neumann relative entropy between forward and time-reversed CJ states (Bolanos-Servin et al., 2013).

6. Choi-Rank and Operational Meaning

The rank of the Choi state ("Choi-rank") serves as a structural indicator: for channel $\mathcal{N}$ , $c = \operatorname{rank}(J_{\mathcal{N}})$ lower bounds the Kraus rank and minimal environment dimension. Its operational relevance is highlighted in entanglement-assisted exclusion tasks — a generalization of super-dense coding where Bob, rather than discriminating the message, aims to rule out $k$ possibilities. The bound

$k \leq \left\lfloor N\cdot \frac{d^2 - r}{d^2} \right\rfloor$

shows that Choi-rank fundamentally limits exclusion capability (Stratton et al., 2024). This provides a direct operational interpretation for the Choi Dataset beyond structural or algorithmic convenience.

7. Practical Computation, Data Structure, and Experimental Implications

The Choi Dataset enables:

Channel validation and benchmarking: Complete positivity, CP-divisibility, entanglement-breaking, $k$ -positivity properties, and resource-theoretic status can be verified by Choi matrix evaluation, experimentation, and analysis.
Non-Markovianity cataloging: Recording moments, entropy measures, and corresponding time series from Choi states affords systematic dynamical diagnostics.
Quantum protocol optimization: Task performance such as exclusion, discrimination, or resource conversion can be mapped directly to Choi-state properties, supporting efficient optimization and experimental design.
Flexible representation: Different Choi-matrix variants correspond to distinct bilinear forms or basis choices; provided necessary conditions (complete order isomorphism), these are interchangeable without loss of operational information.
Infinite-dimensional protocols: For type I von Neumann factors, characterization, extension, and detection of channel and functional structure proceed via the bounded and trace-class analogues of Choi matrices.

The significance of the Choi Dataset lies in its centrality to both mathematical formalism and experimental implementation for quantum channel theory, non-Markovian dynamical analysis, resource theory realization, and communication tasks.

Table: Choi Matrix Correspondence

Map Property	Choi Matrix Characteristic	Reference
Complete positivity	Positive semidefiniteness	(Kye, 2022, Han et al., 2023, Han et al., 2023)
k-positivity	k-block-positivity	(Han et al., 2023)
k-superpositivity	Schmidt number $\le k$	(Han et al., 2023, Han et al., 2023)
Entanglement breaking	Integral separable CJ operator	(Holevo, 2010)
Optimal (map)	Spanning set of product vectors	(Scala et al., 2023)
Choi-rank	Bound for exclusion tasks	(Stratton et al., 2024)

The Choi Dataset thus encodes the complete landscape of theoretical and operational quantum channel analysis. Its mathematical structure, resource-theoretic implications, dynamical diagnostics, and experimental viability reinforce its foundational status in contemporary quantum information science.