Entanglement-Breaking Channels

• Entanglement-breaking channels are quantum channels that transform any entangled input state into a separable output, effectively erasing quantum correlations.
• They are structurally characterized by a measure-and-prepare (Holevo) form with rank-one Kraus operators, serving as a critical noise model in quantum information.
• These channels underpin analyses of quantum capacity, classical communication limits, and resource theories, bridging the gap between quantum and classical regimes.

An entanglement-breaking channel (EB channel) is a completely positive, trace-preserving quantum channel such that, when applied to any subsystem of a composite quantum system, it outputs a state that is separable across the partition between the acted-upon subsystem and any auxiliary system. Equivalently, EB channels sever all quantum entanglement between the system they act on and any other system, even for the maximally entangled input state. This class of channels is characterized structurally by admitting a Kraus representation with rank-one operators and, most fundamentally, by the measure-and-prepare (Holevo) structure: every EB channel can be realized as a measurement (POVM) followed by a state preparation conditioned on the measurement outcome. Entanglement-breaking channels play a fundamental role in quantum information theory, both as ultimate noise models and as operational separators between quantum and classical communication.

1. Formal Definitions and Structural Characterizations

Let $$\Phi: \mathcal{B}(\mathcal{H}_A) \to \mathcal{B}(\mathcal{H}_B)$$ be a quantum channel. The following conditions are equivalent and define entanglement breaking:

1. For every auxiliary (reference) system $\mathcal{K}$ and joint state $\rho_{A\mathcal{K}}$, $$(\Phi\otimes\mathrm{id}_{\mathcal{K}})(\rho_{A\mathcal{K}})$$ is separable between $\mathcal{H}_B$ and $\mathcal{K}$ (Wang et al., 2010, Tabia et al., 2024).
2. The Choi-Jamiołkowski operator $$C_\Phi = (\mathrm{id}_A \otimes \Phi)(|\Phi^+\rangle\langle\Phi^+|)$$ is separable in $A|B$ (Moravčíková et al., 2010, Long-Mei et al., 2019, Tabia et al., 2024).
3. $\Phi$ admits a Holevo form (measure-and-prepare): there exists a POVM $\{M_k\}$ and states $\mathcal{K}$0 such that

$\mathcal{K}$1

with $\mathcal{K}$2, $\mathcal{K}$3, and $\mathcal{K}$4 density operators (Wang et al., 2010, Ahiable et al., 2021, Kuramochi, 2018).

Every EB channel thus behaves as a purely classical communication channel followed by quantum state repreparation. In finite dimensions, this is equivalent to $\mathcal{K}$5 having a Kraus decomposition with all Kraus operators of rank one. In infinite dimensions, the measure-and-prepare structural form extends via countable decompositions (Muoi et al., 2024).

2. Mathematical Properties and Extended Frameworks

Generalizations: The EB property is well-defined for channels between arbitrary $\mathcal{K}$6-algebras and von Neumann algebras, and all standard characterizations (Holevo form, Choi-separability, no-broadcasting compatibility) extend. In infinite dimensions, the class of strongly entanglement-breaking (SEB) channels is defined as those whose output is always countably separable; SEB channels admit a measure-and-prepare representation with a countable sum and are determined by the commutativity of the range (Kuramochi, 2018, Muoi et al., 2024). In these settings, one must distinguish between separability and countable separability for states.

Primitivity and Stochastic Representations: Every EB channel induces a stochastic matrix via its Holevo form. The nonzero spectrum of the channel coincides with that of the associated stochastic matrix, tightly linking quantum mixing properties to classical Markovian dynamics. Primitivity of the channel is governed by the primitivity of this stochastic matrix, and precise quantitative bounds connect the primitivity index of the channel and matrix (Ahiable et al., 2021).

Schmidt Number and Partially Entanglement-Breaking Channels: For $\mathcal{K}$7-dimensional systems, the hierarchy of $\mathcal{K}$8-partially entanglement-breaking ($\mathcal{K}$9-PEB) channels refines the EB class: a channel is $\rho_{A\mathcal{K}}$0-PEB if its Choi state has Schmidt number at most $\rho_{A\mathcal{K}}$1; $\rho_{A\mathcal{K}}$2 recovers the standard EB channels (Namiki, 2013). This is operationally equivalent to the channel being constructible via one-way LOCC from a resource state with Schmidt number $\rho_{A\mathcal{K}}$3.

3. Physical Realizations and Thresholds

Gaussian Channels: For bosonic channels with transmissivity $\rho_{A\mathcal{K}}$4 and thermal noise $\rho_{A\mathcal{K}}$5, the EB threshold is given by $\rho_{A\mathcal{K}}$6. At or beyond this point, any initial system-environment entanglement is erased (Zhang et al., 2013). This threshold is directly derived from considering the fate of two-mode squeezed vacuum states under loss and noise and applying the Peres-Horodecki PPT criterion.

Finite-Dimensional Channels: For the depolarizing channel on $\rho_{A\mathcal{K}}$7, $\rho_{A\mathcal{K}}$8, the EB threshold is $\rho_{A\mathcal{K}}$9 (Pirandola, 2013). Random-unitary channels, Pauli channels, and phase/amplitude damping channels similarly admit explicit noise thresholds for entanglement breaking (Cuevas et al., 2017).

Entanglement Sudden Death: The occurrence of entanglement sudden death (ESD) in open quantum systems corresponds, at the operator-sum level, to the composition of the system evolution with an effective EB channel, even when the elementary channels are not EB individually (Knoll et al., 2016).

4. Information-Theoretic Consequences and Operational Role

Zero Quantum Capacity: Every EB channel has zero quantum capacity; it cannot transmit or preserve quantum entanglement with any ancilla. This class is strictly contained within the set of antidegradable channels and encompasses all bi-PPT channels—complementary pairs where both maps remain completely positive after partial transposition—since the bi-PPT property implies EB (Müller-Hermes et al., 2022).

Classical Capacity and Additivity: For EB channels, the classical capacity equals the Holevo information and is additive: $$(\Phi\otimes\mathrm{id}_{\mathcal{K}})(\rho_{A\mathcal{K}})$$0, and $$(\Phi\otimes\mathrm{id}_{\mathcal{K}})(\rho_{A\mathcal{K}})$$1 (Pereg, 2023). In communication over EB channels with unreliable entanglement assistance, the optimal capacity region is captured by single-letter bounds due to the measure-and-prepare structure (Pereg, 2023).

Quantum Memory Resource: Contrary to the classical intuition, in temporal scenarios (multi-time quantum protocols), a quantum system traversing an EB channel cannot, in general, be simulated by a classical memory of the same dimension; explicit tasks exist where EB channels exhibit stronger temporal correlations than any equivalent-dimension classical Markov process (Vieira et al., 2024). However, for spatial correlations, EB channels fully destroy entanglement as expected.

Universal Generation of Separable States: A fundamental result is that any separable state can be written as the output of an EB channel acting on half of a maximally entangled state (Wang et al., 2010). This gives a necessary and sufficient criterion for separability: $$(\Phi\otimes\mathrm{id}_{\mathcal{K}})(\rho_{A\mathcal{K}})$$2 is separable if and only if $$(\Phi\otimes\mathrm{id}_{\mathcal{K}})(\rho_{A\mathcal{K}})$$3 with $$(\Phi\otimes\mathrm{id}_{\mathcal{K}})(\rho_{A\mathcal{K}})$$4 an entanglement-breaking channel.

5. Amendability, Superactivation, and Nonclassical Effects

Amendability: While by definition a single EB channel cannot transmit or revive entanglement, the insertion of intermediate unitary filters between successive applications (i.e., non-Markovian interventions) can "amend" certain (non-strong) EB channels, rendering the overall process non-EB. This amendability is strictly forbidden for strong EB (SEB) channels (Cuevas et al., 2017, Long-Mei et al., 2019).

Reactivation by Classical Correlations and Superactivation: Combining two (or more) EB channels with appropriately correlated (even purely classical, zero-discord) environments can preserve quantum entanglement that would be destroyed by either channel individually. Distinct classes of input states (e.g., Werner or isotropic states, or EPR states in the bosonic case) survive the correlated action due to decoherence-free subspaces or collective noise evasion (Pirandola, 2013, Pirandola, 2012). This non-Markovian, memory-induced restoration of entanglement constitutes a form of "superactivation" for quantum correlations.

Broadcasting Compatibility and Memory Superactivation: Two EB channels, individually useless as quantum memories, can become collectively nontrivial when used as jointly compatible marginals of a broadcasting channel. In explicit constructions, all broadcasting realizations entangle certain input states between output subsystems, even though the channels are separately EB. This demonstrates superactivation of quantum memory resources by EB channels (Tabia et al., 2024).

6. Relation to Entanglement-Annihilating Channels and Resource Theory

EB channels are distinct from entanglement-annihilating (EA) channels. An EA channel outputs a separable state with respect to internal partitions of its own input but may leave entanglement with external ancillas untouched. The EB class is strictly contained within the EA class: every EB channel is EA, but not every EA channel is EB (Moravčíková et al., 2010). This distinction elucidates subtle resource-theoretic hierarchies in quantum information, with EB channels corresponding to processes that reduce quantum information to fully classical communication, whereas EA channels only guarantee loss of internal entanglement.

7. Structural, Operator-Algebraic, and Resource Properties

Infinite-Dimensional and Algebraic Structure: In the context of operator algebras, all major properties and characterizations of EB channels—Holevo structure, no-broadcasting compatibility, and Choquet decompositions of separable states—extend to general $$(\Phi\otimes\mathrm{id}_{\mathcal{K}})(\rho_{A\mathcal{K}})$$5-algebras and von Neumann algebras (Kuramochi, 2018). The class of EB channels forms an upper and lower Dedekind-closed subset in the poset of channels under the order of CP-randomization.

Null-Space Construction and Commutative-Range Maps: In infinite dimensions, every closed self-adjoint trace-zero subspace of trace-class operators is the null space of a SEB channel, yielding fine-grained control over the privacy properties and fixed-point algebras in quantum dynamics (Muoi et al., 2024).

Resource-Theoretic Classification: The hierarchy of $$(\Phi\otimes\mathrm{id}_{\mathcal{K}})(\rho_{A\mathcal{K}})$$6-partially entanglement-breaking channels provides a resource-theoretic partition of channels according to the Schmidt number of the entanglement they can preserve or require for simulation, leading to one-way LOCC operational equivalence (Namiki, 2013).

---

The theory of entanglement-breaking channels spans deep mathematical structure, precise operational significance, and surprising nonclassical phenomena under extended scenarios. EB channels not only delineate the ultimate classical boundary in quantum processing, but their collective behaviors and interplay with memory, compatibility, and correlated environments yield a multiplicity of resource-theoretic and practical implications in quantum information science.