⚛️ quantum physics

Sklar's Theorem and Quantum State Reconstruction from All-Context Dependence

This paper applies Sklar's theorem to demonstrate that the dependence structure of local binary projective measurements on non-product two-qubit states uniquely determines the quantum state up to a discrete ambiguity, while a corresponding scalar optimization combined with quantum total correlation reproduces global quantum discord.

Original authors: Yi-Yu Lin

Published 2026-09-09

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yi-Yu Lin

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the world of probability, there is a long-standing idea that any collection of related events can be understood by separating two distinct things: how each event happens on its own, and how the events influence one another. Imagine a group of friends; you can describe how often each person laughs individually, and then separately describe the pattern of their shared laughter. In classical statistics, a mathematical principle known as Sklar's theorem provides a rigorous way to peel these layers apart, isolating the pure "dependence" between variables from their individual behaviors. Quantum mechanics, the physics of the very small, produces its own kind of probability tables when scientists measure particles. These tables look just like the statistical records of classical events, yet they arise from a reality that defies everyday intuition. The central puzzle for physicists has always been how much of a quantum system's true nature is hidden in these measurements. If we strip away the individual quirks of each particle to see only how they depend on each other, is anything of the original quantum state left behind, or does the picture dissolve into noise?

A researcher at Fudan University has now answered this question for a specific, fundamental case: a pair of quantum particles known as qubits. They discovered that even if you discard all the information about how each particle behaves individually in every possible measurement setting, the remaining web of connections between them is enough to reconstruct the original state of the pair. The scientist did this by treating the quantum measurement results not as a single snapshot, but as a vast family of snapshots taken from every possible angle. In each snapshot, they removed the individual statistics of the particles, leaving only the "dependence nucleus"—a mathematical object that captures the pure relationship between the two. They found that for any pair of particles that are not completely independent, this collection of pure relationships, gathered from all possible measurement directions, uniquely identifies the quantum state, with only one tiny, discrete ambiguity: the state could be flipped in a specific way, but otherwise, it is fully determined.

The researcher arrived at this conclusion by applying the logic of Sklar's theorem to the data generated by the Born rule, which is the standard method for calculating probabilities in quantum mechanics. When a scientist measures a pair of qubits, they choose a direction to look at each one, and the machine produces a table of outcomes. The team realized that while the numbers in the table change depending on the direction chosen, the underlying structure of how the two particles depend on each other remains consistent across all these choices. By mathematically stripping away the individual probabilities of each particle in every single context, they were left with a "dependence nucleus." They proved that if two different quantum states produced the exact same set of these nuclei across all possible measurement directions, then those two states must be essentially the same, or one is a specific "double spin-flip" version of the other. This result is significant because it shows that the information lost by ignoring individual particle behaviors in one measurement is not truly lost; it is recovered when you consider the entire family of measurements together. The only time this reconstruction fails is when the two particles are completely independent of each other, a scenario where no shared dependence structure exists to begin with.

Beyond simply reconstructing the state, the paper also explored what happens when this complex web of dependence is compressed into a single number. The researcher asked how much of the total connection between the particles could be seen in just one measurement setting. They found that if you take the maximum possible connection visible in any single setting and compare it to the total quantum connection inherent in the state, the difference between them is exactly a quantity physicists call "global quantum discord." This is a measure of how much of the quantum correlation is purely quantum and cannot be explained by classical statistics. The study shows that this famous quantity is not just an arbitrary definition, but a natural outcome of separating the individual behaviors from the shared dependence and then asking how much of that shared dependence can be captured in a single view. It suggests that the strange, non-classical correlations of quantum mechanics are simply the part of the relationship that survives when you try to view the system through the lens of classical statistics.

The work is currently limited to pairs of particles, but the author suggests that the logic could extend to larger groups of particles or even systems with continuous properties, like the position of a particle moving through space. For larger groups, the challenge would be to see if the same family of dependence structures can still pin down the state, and for continuous systems, the mathematics becomes cleaner because the "dependence" object is uniquely defined without the need for the specific adjustments used for discrete particles. The researcher proposes that for continuous systems, looking at the pure dependence structure across all possible measurement angles might also allow for a full reconstruction of the state. This opens a new path for understanding quantum systems, not by looking at the particles themselves, but by looking exclusively at the invisible threads that bind them together, proving that the whole is indeed recoverable from the sum of its relational parts.

Technical Summary: Sklar's Theorem and Quantum State Reconstruction from All-Context Dependence

Problem Statement
The paper addresses a fundamental question in quantum information: to what extent can a multipartite quantum state be reconstructed if the marginal statistics of local measurement outcomes are discarded, leaving only the dependence structure between subsystems? While Sklar's theorem in classical probability theory provides a rigorous framework for separating marginal distributions from dependence structures (via copulas), the quantum setting introduces a unique complexity. A single quantum state generates a family of classical joint probability distributions (Born statistics) depending on the choice of local measurement contexts. The central problem is whether the collection of "margin-free" dependence structures, generated across all possible local measurement contexts, contains sufficient information to uniquely identify the underlying quantum state, up to discrete symmetries.

Methodology
The author applies the framework of Sklar's theorem to Born statistics generated by local quantum measurements. The methodology proceeds in three main stages:

Discrete Dependence Representation: Recognizing that quantum measurement outcomes are discrete, the author notes that the standard Sklar copula is not uniquely determined by the joint distribution in the discrete case. To resolve this, they adopt the "dependence nucleus" concept (following Geenens), where two probability tables are considered equivalent if they differ only by positive local reweightings (scaling of rows and columns). For binary discrete variables, this equivalence class is uniquely characterized by the odds ratio, $\omega$ , or its bounded transformation $Q$ .
All-Context Reconstruction (Two Qubits): The author defines a map from a two-qubit state $\rho$ to the family of dependence nuclei $\{D(p_{\rho, B})\}_B$ generated by all local binary projective measurements $B$ . They analyze the reconstruction problem by expressing the Born probabilities in the Fano–Bloch representation. By establishing the relationship between the dependence metric $Q$ and the connected correlation functions, they derive conditions under which the equality of dependence structures across all contexts implies the equality of the underlying states.
Scalar Compression and Quantum Discord: In a secondary analysis, the author compresses the full dependence structure in each context into a single scalar: the classical total correlation (multi-information). They optimize this scalar over all local measurement contexts and compare the result to the state's total quantum correlation.

Key Results

Reconstruction Theorem for Two Qubits: The paper proves that for any non-product two-qubit state, the family of all-context margin-free dependence structures (dependence nuclei) uniquely determines the state up to a discrete ambiguity. Specifically, if two non-product states $\rho$ and $\sigma$ yield identical dependence nuclei for all local binary projective measurements, then $\sigma = \rho$ or $\sigma = \tilde{\rho}$ , where $\tilde{\rho}$ is the "double spin-flip" of $\rho$ (defined as $(\sigma_y \otimes \sigma_y)\rho^*(\sigma_y \otimes \sigma_y)$ ). Product states are identified as the degenerate boundary case where the dependence structure vanishes in all contexts.
Recovery of Global Quantum Discord: When the dependence in each context is reduced to the classical total correlation and optimized over all local bases, the resulting quantity is shown to be exactly the global quantum discord (as defined by Rulli and Sarandy). The paper demonstrates that global quantum discord can be interpreted as the difference between the total quantum correlation and the maximum classical total correlation extractable from any single local measurement context.
Distinction between Scalar and Structural Information: The author emphasizes that while the optimized scalar (global discord) is a well-known quantity, it represents only a partial projection of the full data. The full family of dependence nuclei contains significantly more information, as evidenced by the reconstruction theorem, which recovers the state (up to ambiguity) whereas the scalar optimization does not.

Significance and Claims
The paper claims to provide a novel perspective on quantum state characterization by isolating the "dependence" component of quantum statistics from local marginal properties. Its primary significance lies in demonstrating that information lost when discarding marginals in a single context is not necessarily lost when considering the compatibility of dependence structures across all contexts.

The author positions this work as a structural inquiry rather than a proposal for a new experimental protocol. They highlight that:

The separation of marginals and dependence, standard in classical statistics, can be rigorously extended to quantum mechanics via the dependence nucleus.
The reconstruction theorem establishes that the "relational" data among subsystems, stripped of local identity, is sufficient to reconstruct the global state for non-product two-qubit systems.
The derivation of global quantum discord emerges naturally from this framework as an optimization of marginal-free dependence, offering a structural interpretation of the quantity without invoking pre-existing definitions of quantum correlation.

The paper concludes by noting that while the results are established for two qubits, the conceptual framework extends naturally to multiqubit systems (via log-linear interaction parameters) and continuous-variable systems (where unique copulas exist), though the specific reconstruction proofs for these cases are left as future work.

More like this