Quantum Discord

In the quest to understand the universe at its most fundamental level, the nature of information and correlation takes on a strange and powerful new meaning. For decades, quantum entanglement was seen as the quintessential feature separating the quantum world from our classical reality. However, this view only tells part of the story. A deeper and more pervasive form of "quantumness" exists, one that challenges our very notion of what it means to know something about a system. This is the realm of quantum discord. This article addresses the knowledge gap left by an entanglement-centric view, introducing discord as a more general measure of quantum correlation. In the following chapters, we will embark on a journey to understand this subtle yet profound concept. We will first explore its "Principles and Mechanisms," defining discord by contrasting it with classical information and revealing its surprising relationship with entanglement. Subsequently, we will witness its broad impact in "Applications and Interdisciplinary Connections," discovering how discord manifests in quantum measurements, fuels quantum technologies, and is even imprinted on the cosmic scale.

A conceptual visualization: A large tetrahedron represents all physical states. Inside it, an octahedron represents the separable (non-entangled) states. The zero-discord states are just three lines (the axes) within the octahedron. The entire volume of the octahedron, excluding these lines, consists of states that have quantum discord but no entanglement.

To truly grasp the essence of quantum discord, we must embark on a journey that begins with a familiar concept—information—and watch as it transforms in the strange and wonderful quantum world. Classical information theory, the bedrock of our digital age, provides a perfect launching pad. It tells us that the correlation between two systems, let’s call them Alice's ($A$) and Bob's ($B$), can be quantified by a single, unambiguous measure: mutual information. But as we'll see, the quantum realm is not so simple. It forces us to ask a deeper question: what kind of information are we talking about?

In the classical world of bits and bytes, the mutual information $I(A:B)$ tells us how much knowing about system $B$ reduces our uncertainty about system $A$. It can be written in two ways that, classically, are perfectly identical.

First, there's the symmetric view: $I(A:B) = H(A) + H(B) - H(A,B)$ Here, $H(X)$ is the Shannon entropy, a measure of the uncertainty or "surprise" associated with a variable $X$. This formula beautifully expresses mutual information as the sum of the individual uncertainties minus their joint uncertainty. It's like saying the shared information is what's left after you account for the total information and subtract what is unique to each part.

The second view is operational and asymmetric: $I(A:B) = H(A) - H(A|B)$ This tells us that the information we share is the total uncertainty in $A$ minus the uncertainty that remains in $A$ after we have learned the state of $B$. It’s the information gained by observation.

Classically, these two expressions are one and the same. But in quantum mechanics, this equivalence shatters, and in that fracture, we find quantum discord.

When we step into the quantum world, our uncertainties are described by the von Neumann entropy, $S(\rho) = -\mathrm{Tr}(\rho \ln \rho)$, the quantum analog of Shannon entropy. The first formula for mutual information translates directly:

This is the ​​quantum mutual information​​, and it serves as our measure of the total correlation—both classical and quantum—between two systems described by the joint density matrix $\rho_{AB}$.

The second formula, however, hits a quantum roadblock. The term $H(A|B)$ implies we can know the state of $B$ without affecting $A$. But in quantum mechanics, the act of "knowing"—of measuring—is an invasive procedure. You cannot simply look at a quantum system; you must interact with it, and that interaction can profoundly disturb the delicate correlations it shares with other systems.

So, how much information can we actually gain about Alice's system by measuring Bob's? The answer depends on how Bob performs his measurement. He could measure spin along the z-axis, the x-axis, or any other direction. Each choice of measurement constitutes a different question he can ask his system. To find the truly "classical" part of the correlation, we must find the measurement that gives Bob the most information about Alice while causing the least possible disturbance. This maximum accessible information is what we call the ​​classical correlation​​, often denoted $J(A:B)$ or $C(A:B)$:

This expression is the quantum incarnation of $H(A) - H(A|B)$. It represents the total information in $A$, $S(\rho_A)$, minus the minimum average uncertainty that remains in $A$ after we have performed the best possible measurement on $B$. This minimization over all possible measurements, $\{\Pi_k^B\}$, is the crucial new ingredient. It's an admission that the information we extract is conditioned on the questions we ask.

We now have two distinct ways to quantify correlation in a quantum system: the total correlation, $I(A:B)$, and the accessible classical correlation, $J(A:B)$. Unlike in the classical world, these are generally not equal. The total information is almost always more than what we can extract by local measurements. This mysterious surplus, this information that vanishes under the clumsy touch of measurement, is the ​​quantum discord​​:

Quantum discord is the part of the correlation that is inherently quantum. It is a measure of how non-classical a system's correlations are, defined by the disturbance caused by a local measurement. If a state has zero discord, it means there exists at least one local measurement that can reveal everything there is to know about the correlations without causing any quantum disturbance. Such states are considered "classical" in their correlational structure. They have a very specific form, often called ​​classical-quantum states​​:

In these states, there's an orthonormal basis for subsystem $A$ (the $\{|i\rangle_A\}$ basis) such that measuring in this basis reveals the "classical" index $i$ with probability $p_i$, leaving subsystem $B$ in the corresponding state $\rho_B^{(i)}$. Because the basis states are orthogonal, this measurement can be done without introducing any quantum weirdness, and all the mutual information can be accessed. For any state of this form, the quantum discord (measured on A) is exactly zero. Any state that cannot be written this way must have non-zero discord. This happens, for example, when the possible states of one subsystem are not orthogonal, such as $|0\rangle$ and $|+\rangle$, making it impossible to distinguish them perfectly without disturbance.

At this point, you might be thinking: "Isn't this just entanglement?" The answer, surprisingly, is no. While entanglement is a powerful form of quantum correlation, discord is a more general and fundamental concept. All entangled states have discord, but not all states with discord are entangled.

For a ​​pure entangled state​​, like the Bell state $|\Phi^+\rangle = \frac{1}{\sqrt{2}}(|00\rangle+|11\rangle)$, the situation is simple. The total correlation is entirely quantum. The mutual information turns out to be exactly twice the entanglement entropy ($I(A:B) = 2S_A$), and both the classical correlation and the discord are equal to the entanglement entropy ($J(A:B) = D(A:B) = S_A$). In this pristine case, all these measures are just different ways of looking at the same thing: entanglement.

The real surprise comes with ​​mixed states​​. It is entirely possible to construct a quantum state that is ​​separable​​—meaning it is not entangled and can be created using only local operations and classical communication—but which still possesses non-zero quantum discord.

A famous example is the ​​Werner state​​, a mixture of a maximally entangled Bell state and a completely random, uncorrelated state. If the mixture contains a small enough proportion of the Bell state (specifically, for a mixing parameter $p \le 1/3$), the resulting state is separable. It has no entanglement. Yet, a direct calculation shows that its quantum discord is non-zero.

This reveals a fascinating truth: quantum correlations can exist even in the absence of the "spooky action at a distance" we associate with entanglement. These non-entangled-but-discordant states represent a subtle form of quantumness. The correlation exists, but it's encoded in a basis that is non-orthogonal, making it impossible to access without disturbance.

A Gallery of Correlations

We can visualize the landscape of quantum states to better understand this relationship. For a particular family of two-qubit states known as Bell-diagonal states, the state is defined by three parameters $(c_1, c_2, c_3)$. The set of all possible physical states forms a tetrahedron in this parameter space.

• The ​​separable states​​, those without entanglement, form a perfect octahedron nestled inside this tetrahedron.
• The states with ​​zero discord​​ are even more restricted; they lie only along the three axes ($c_1$, $c_2$, $c_3$) passing through the center of the octahedron.

The volume of the axes is zero. This means that almost all separable states are not on these axes. The conclusion is stunning: the set of states with genuinely classical correlations is vanishingly small. If you were to pick a separable quantum state at random, you would almost certainly pick one with quantum discord.

Having journeyed through the principles and mechanisms of quantum discord, we might be left with a feeling of abstract curiosity. We have a new tool, a new way of thinking about correlations, but what is it for? What does it do? It is one thing to define a quantity in the mathematical playground of quantum theory; it is another entirely to see it at work in the real world, shaping the phenomena we observe and enabling the technologies we build.

The story of quantum discord's applications is a beautiful illustration of the interconnectedness of physics. We will see that this subtle measure of "quantumness" is not just a theorist's plaything. It appears in the fundamental interactions that underpin quantum measurement, it governs the flow of information in quantum computers and communication networks, it sets limits on what an eavesdropper can know, and, in a breathtaking finale, we will find it imprinted on the very fabric of the cosmos. Its presence—and sometimes, its conspicuous absence—tells a profound story about the nature of our quantum universe.

At the very heart of quantum mechanics lies the strange and wonderful process of measurement. When we "look" at a quantum system, we inevitably interact with it, and this interaction creates correlations. Discord gives us a new lens through which to view this process.

Imagine a classic Stern-Gerlach apparatus, where a particle's spin is coupled to its path through a magnetic field. An initial particle, with its spin pointing in some arbitrary direction, enters the device. The magnetic field kicks the spin-up component one way and the spin-down component another. The spin and the spatial position of the particle become correlated. But what kind of correlation is it? If the two paths are perfectly distinguishable, the correlation is simple entanglement. But in a more realistic scenario, the spatial wavefunctions corresponding to the two paths might overlap. Our calculations show that in this case, a non-zero quantum discord appears between the particle's spin and its spatial degree of freedom. The discord depends intimately on the initial spin orientation and the degree of overlap between the paths. It quantifies the quantum nature of the correlation forged by the measurement-like interaction.

This idea becomes even clearer when we consider the famous two-slit experiment, a cornerstone of quantum complementarity. Suppose we place a "which-path" detector near the slits to find out which one the particle went through. This act of detection entangles the particle's path with the state of the detector. The visibility of the interference pattern on the screen is famously tied to how much information the detector learns. We can quantify the "distinguishability" of the path information stored in the detector. It turns out that the quantum discord between the particle and the detector is directly related to this distinguishability. When the detector states are perfectly orthogonal, the path is known with certainty, the interference pattern vanishes, and the correlations are purely classical (zero discord). When the detector states are identical, no path information is gained, the interference is perfect, and there are no correlations at all. In the fascinating middle ground, where we have partial which-path information, quantum discord is present. It is the signature of the quantum character of the information held by the detector, a direct measure of the "quantumness" that remains in the system-detector correlation.

If the quantum world is so full of these strange correlations, why does our everyday macroscopic world seem so stubbornly classical? Discord helps us understand this transition. Quantum correlations are fragile, easily disrupted by interactions with the surrounding environment—a process called decoherence.

Let's imagine a toy model to see how this works. Suppose Alice and Bob share a pair of perfectly entangled qubits. Alice then takes her qubit and performs an operation that couples it to a third qubit, an "ancilla" from the environment. If this ancilla then flies away, carrying its information with it, the original entanglement between Alice and Bob is damaged. Our analysis reveals something remarkable: this local interaction with the environment can completely destroy the initial entanglement, converting all of it into purely classical correlation. The final state shared by Alice and Bob has zero quantum discord. The "quantumness" has been washed away by the interaction with the environment, leaving only classical echoes behind. This provides a beautiful glimpse into why we don't see quantum weirdness on a large scale; the constant interaction with the environment systematically erodes discord and entanglement, leaving the world appearing classical.

However, the absence of discord isn't always due to environmental destruction. Sometimes, a system's intrinsic nature prevents it from harboring quantum correlations, even at absolute zero. Consider two magnetic particles (qubits) whose interaction energy depends only on whether their spins are aligned or anti-aligned along a specific axis—a simple Ising model interaction. If this system is allowed to reach thermal equilibrium with a reservoir, we find that for any temperature, it possesses exactly zero entanglement and zero quantum discord. The interaction is "too classical"; it doesn't create the kind of superpositions necessary for quantum correlations. This is a profound lesson: not all interactions are created equal. The very structure of the laws governing a system determines its capacity for "quantumness."

While the environment may conspire to destroy discord, physicists and engineers are working to harness it. In the burgeoning field of quantum technologies, discord is emerging as a subtle but significant resource.

In ​​quantum communication​​, we often want to send quantum states from one place to another. The famous teleportation protocol relies on a shared resource of high-quality entanglement. But what if our channel is noisy and the entanglement is imperfect, described by a so-called Werner state? A fascinating calculation shows that even when the entanglement is degraded, a residual quantum discord between the communicated qubit and a reference system can survive. This suggests that discord-based correlations might be more robust to noise than entanglement and could be useful for certain communication tasks even when entanglement fails.

Discord also plays a crucial role in ​​quantum cryptography​​. In the BB84 protocol for quantum key distribution, Alice sends qubits to Bob, and they later compare notes to see if an eavesdropper, Eve, has tampered with the transmission. Any intervention by Eve creates correlations between her own system and the qubits she intercepted. The amount of disturbance she causes is measured by the Quantum Bit Error Rate (QBER). We can model this situation and calculate the quantum discord of the state shared between Alice and Eve as a function of the QBER. This gives us a precise information-theoretic handle on what Eve can learn, revealing that the nature of her information has a distinctly quantum character that is captured by discord.

In ​​quantum computing​​, the laws of physics forbid us from perfectly copying an unknown quantum state—the celebrated no-cloning theorem. However, we can create imperfect copies. An "optimal" quantum cloner produces the best possible copies allowed by quantum mechanics. What kind of correlations do these imperfect clones share? It turns out they are not entangled, but they possess a specific, non-zero amount of quantum discord. Discord is an intrinsic feature of this fundamental information-processing task.

Yet, the story is, as always in quantum mechanics, more nuanced. One might naively think that a powerful quantum algorithm must be saturated with quantum correlations. This is not always true. Consider Simon's algorithm, which provides an exponential speedup over any classical counterpart for a specific problem. If we look at the state of the computer's register after a key computational step, and calculate the discord between two of its qubits, we can find that it is exactly zero. Similarly, in the Steane seven-qubit code, a sophisticated method for protecting quantum information from errors, the discord between pairs of physical qubits making up a logical state can also be zero. This reveals the intricate and non-uniform structure of correlations in complex quantum systems. Powerful quantum computations and robust quantum codes can be built from components that, when viewed in isolation, share only classical correlations. The "quantum advantage" arises from the global properties of the state, not necessarily from local pairwise quantum correlations.

We conclude our tour by lifting our gaze from the laboratory to the heavens. The largest structures in our universe—galaxies, and clusters of galaxies—are thought to have grown from tiny, primordial density fluctuations in the very early universe. According to the theory of cosmic inflation, these fluctuations were themselves born from quantum vacuum fluctuations, stretched to astronomical sizes by the exponential expansion of space.

The quantum state describing these primordial perturbations for a pair of modes with opposite wavevectors is a two-mode squeezed vacuum state—a canonical example of a highly quantum-correlated system. For such a pure state, the quantum discord is simply a measure of its entanglement. The astonishing link is that the amount of squeezing, and therefore the amount of discord, is directly related to the amplitude of the power spectrum of the Cosmic Microwave Background (CMB) radiation—the afterglow of the Big Bang that we observe today.

By measuring the temperature fluctuations in the CMB, cosmologists determine the power spectrum $\mathcal{P_R}$. From this observable number, we can directly calculate the quantum discord that existed between modes of the quantum field that seeded all cosmic structure. The correlations that we study in quantum information labs are the very same type of correlations that were present at the dawn of time, and whose consequences are now writ large across the entire observable universe. It is a humbling and awe-inspiring thought—a testament to the profound unity of the laws of physics, from the smallest of qubits to the grandest of cosmic canvases.