When we characterize the non-classical nature of quantum systems, we often default to quantum entanglement. However, entanglement is only one facet of quantum correlation. A more pervasive type of non-classicality is quantum discord, which captures correlations that exist even in separable (non-entangled) states.
To fully utilize quantum mechanics for computing and communication, researchers are mapping out the precise mathematical boundaries between entanglement and discord. By studying how these correlations behave under geometric distances, state extensions, and multi-copy scaling, we can understand why certain quantum resources remain robust while others vanish.
The Conceptual Divide: Discord vs. Entanglement
To understand why quantum discord and entanglement diverge, consider what happens when a quantum system interacts with its environment. Entanglement is fragile; as a system decoheres, its entanglement can drop abruptly to zero. Yet, the system may still retain a non-classical character.
As Lan Zhou demonstrates in Quantum correlations of two qubits interacting with a macroscopic medium, while entanglement can never exceed its initial value during environmental interaction, quantum discord can actually be amplified or generated. This occurs because discord measures the fundamental disruption caused by making local measurements on a system.
Mathematically, as detailed by Omar Jiménez et al. in Quantum Discord, Thermal Discord, and Entropy Generation in the Minimum Error Discrimination Strategy, quantum discord is defined as the difference between total quantum mutual information $I(\rho)$ and the maximum classical correlations $C(\rho)$ extractable via local measurements:
$$\mathcal{D}(\rho) = I(\rho) - C(\rho)$$
Because local measurements on quantum states inevitably destroy superpositions, any discrepancy between total information and classical information signals non-classicality. This makes discord a highly sensitive probe. For instance, Ben‐Qiong Liu et al. in Quantum and classical correlations in the one-dimensional XY model with Dzyaloshinskii-Moriya interaction showed that in spin chains, quantum discord can identify quantum phase transitions even at long distances where local entanglement has completely vanished.
Total Quantum Correlations [Mutual Information I(ρ)]
├── Classical Correlations [C(ρ)] (Extractable via local measurement)
└── Quantum Discord [D(ρ)] (Destroyed by local measurement)
└── Contains: Quantum Entanglement + Non-entangled Quantum Anomalies
For specific highly symmetric states, this complex optimization simplifies dramatically. Elias Riedel Gårding et al. proved in Bell Diagonal and Werner State Generation: Entanglement, Non-Locality, Steering and Discord on the IBM Quantum Computer that for Bell diagonal states, standard quantum discord is exactly equivalent to the asymmetric relative entropy of discord, simplifying the analytical tracking of these correlations on real quantum hardware.
Unifying Discord and Entanglement via State Extensions
If discord and entanglement are distinct, how do they relate mathematically? A powerful way to bridge them is through the concept of state extensions—embedding a bipartite state into a larger, hypothetical multipartite system.
C. Xiong et al. uncovered a deep structural link in Entanglement as the cross-symmetric part of quantum discord. They showed that if you take a bipartite state and minimize its quantum discord over all possible cross-symmetric state extensions, the residual discord value precisely equals the state's Bures distance of entanglement.
Why does this work? Cross-symmetry in the extended space acts as a mathematical filter. It isolates and cancels out the non-entangled quantum correlations inherent in the local subsystems, leaving behind only the non-local, non-separable entanglement. This elegant result proves that entanglement can be viewed fundamentally as a constrained, minimal manifestation of quantum discord.
Resource Theories and State-Dependent Distances
To treat these correlations rigorously, physicists use quantum resource theories. In this framework, we define a set of "free states" (states lacking the resource, such as completely separable states) and require that valid resource measures (monotones) never increase under a set of restricted operations, like Local Operations and Classical Communication (LOCC).
One way to measure a resource is geometrically: how far is our state from the nearest free state? Xiao-Bin Liang et al. developed exact analytical solutions for this in Complete optimal convex approximations of qubit states under B2 distance, providing a precise framework for calculating the minimal distance from a given qubit state to the convex set of classical or separable states.
However, traditional geometric measures use a fixed set of free states. A more tailored approach was introduced by Kun Wang et al. in Quantifying the unextendibility of entanglement*. To quantify "unextendible entanglement"—the degree to which a state cannot be shared with a third party—the authors minimized a generalized divergence against a set of state-dependent free states.
By letting the target free states depend dynamically on the state being analyzed, they constructed highly sensitive entanglement measures that are strictly monotone under two-extendible operations and LOCC, ensuring physical consistency.
The Power of Multi-Copy Scaling and Additivity
A fundamental question in quantum information is scalability: if we bundle multiple identical copies of a quantum state together, do its properties scale linearly (additively), or do new behaviors emerge?
Historically, it was uncertain whether the local unitary equivalence of individual states would hold when multiple copies were processed simultaneously. Xiaoming Sun 0001 et al. resolved this in Local Equivalence of Multipartite Entanglement, proving that local unitary equivalence is strictly additive. Two states are locally equivalent if and only if $N$ copies of those states remain equivalent. This means that scaling up the number of copies does not cause new types of multipartite entanglement to suddenly emerge or mutate.
This predictable scaling behavior allows for the creation of robust bounds. For example, Xin Wang et al. introduced a new entanglement monotone, $E_W$, in Improved semidefinite programming upper bound on distillable entanglement. Because $E_W$ is efficiently computable via Semidefinite Programming (SDP) and is perfectly additive, it provides a strict, non-increasing upper bound on exactly how much pure entanglement can be distilled from noisy, multi-copy quantum states.
Multi-copy scaling can also fix apparent anomalies in quantum mechanics, such as the violation of monogamy. Monogamy dictates that if Alice and Bob are maximally entangled, Alice cannot share any entanglement with Charlie. Some measures, like negativity, occasionally violate this rule, labeling certain states as "nonmonogamous."
In Activation of nonmonogamous multipartite quantum states, Tamoghna Das et al. demonstrated a phenomenon called activation. They proved that if you take a state that violates a monogamy inequality and process enough identical copies of it together, the joint state space imposes strict global constraints that force the system to satisfy the monogamy inequality.
Once a state is scaled into a regime where it satisfies monogamy, this property becomes completely stable; adding further copies will never degrade the constraint.
Single Copy (Anomalous) Multiple Copies (Activated)
Alice ───(Negativity)─── Bob [Alice]^N ───(Strict Monogamy)─── [Bob]^N
│ │
└─────(Violates Monogamy)─── Charlie └─────(No leakage allowed)─── [Charlie]^N
Conclusion
By moving beyond simple definitions of entanglement and exploring the broader landscape of quantum discord, resource theories, and multi-copy additivity, researchers have built a rigorous framework for non-classical correlations.
Whether leveraging linear entropy to optimize quantum machine learning circuits (Jiaqi Cheng et al. in Problem-dependent Quantum Circuit Design Based on Entropy Matching) or calculating distillable limits via SDP, understanding why these correlations persist across state extensions and copy scaling is essential for engineering the quantum technologies of tomorrow.
Key Foundational Work
Quantum correlations of two qubits interacting with a macroscopic medium (2015) — contributed 1 concept elements to this direction.
Quantum and classical correlations in the one-dimensional<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mi mathvariant="normal">XY</mml:mi></mml:mrow></mml:math>model with Dzyaloshinskii-Moriya interaction (2011) — contributed 1 concept elements to this direction.