Necessary and sufficient conditions for positive semidefinite quantum mutual information matrices

Feng Liu, Fei Gao, Su‐Juan Qin, Qiaoyan WenarXiv 2014 · DOI · arXiv

Abstract

In classical information theory, the mutual information matrix—where each entry $(i, j)$ represents the mutual information between two variables—is guaranteed to be positive semidefinite (PSD). This property is foundational for various applications in data analysis and signal processing. However, in the quantum realm, the Quantum Mutual Information (QMI) matrix often fails to be PSD due to the non-classical nature of entanglement. This paper derives the necessary and sufficient conditions for the QMI matrix to be PSD for 2-partite and 3-partite systems. Additionally, the authors propose a new metric, genuine $n$-partite mutual information ($I_G$), which offers a symmetric, non-negative, and more accurate measure of total correlations in multipartite quantum states.

Introduction: The Quantum Departure from Classical Norms

In classical systems, the correlation between two variables is bounded by the entropy of the individual variables. Quantum mechanics disrupts this intuition. Through entanglement, the mutual information between two parties can exceed the entropy of the individual parties, leading to what is effectively "negative conditional entropy."

This paper addresses the mathematical consequences of this phenomenon on the QMI matrix ($M_n$), defined as an $n \times n$ matrix where the $(i, j)$-entry is the quantum mutual information $I(\rho_{A_i A_j})$. While the QMI matrix is used to measure bipartite correlations within a larger system, its failure to remain PSD distinguishes it sharply from its classical counterpart. The authors seek to define the boundaries of this PSD property and provide a more robust alternative for measuring multipartite correlations.

Methodology and Matrix Analysis

The researchers utilized matrix analysis and quantum information theory to dissect the properties of $M_n$. A key technique employed was the use of contract matrices (congruence transformations) to simplify $M_n$ into a diagonal form. By examining the resulting eigenvalues and principal minors, the authors were able to isolate the conditions required for positive semidefiniteness.

Conditions for $n=2$ and $n=3$

For a two-party system ($n=2$), the authors established a clear threshold: the QMI matrix is PSD if and only if the product of the marginal entropies is greater than or equal to the square of the mutual information: $$S(\rho_{A_1})S(\rho_{A_2}) \geq I^2(\rho_{A_1A_2})$$

For three-party systems ($n=3$), the conditions become significantly more complex, involving the interplay between all bipartite mutual information values within the state. The paper provides explicit necessary and sufficient conditions for these cases, though it notes that the complexity scales rapidly as $n$ increases.

Defining Genuine $n$-partite Mutual Information ($I_G$)

Existing measures of multipartite correlation, such as the standard tripartite information ($I_3$), are often "ill-defined" because they can result in negative values or zero even when significant entanglement is present. To resolve this, the authors introduced Genuine $n$-partite quantum mutual information ($I_G$).

Instead of relying on matrix eigenvalues directly, $I_G$ is calculated by:

  1. Normalizing the entries of the QMI matrix.
  2. Treating these normalized entries as a probability distribution.
  3. Calculating the Shannon entropy of this distribution.

This approach ensures that $I_G$ remains symmetric, non-negative, and bounded, making it a more reliable tool for quantifying total correlation.

Key Results and Validation

The theoretical findings were validated using several benchmark quantum states:

  • The Singlet State Counterexample: The authors proved that for a standard 2-qubit entangled state (like a singlet), the QMI matrix is not PSD. Specifically, the matrix contracts to a form with a negative eigenvalue ($-3$), demonstrating how entanglement directly breaks the PSD property.
  • GHZ State Analysis: For the 3-qubit Greenberger–Horne–Zeilinger (GHZ) state, the traditional $I_3$ measure is 0, which erroneously suggests a lack of tripartite correlation. In contrast, the authors' $I_G$ metric yields $\log_2 3$, correctly identifying the presence of significant multipartite correlation.
  • W State Sensitivity: In a 4-partite W state, $I_G$ was calculated at approximately $3.9897$, remarkably close to the theoretical maximum of $4$ ($\log_2 2^2$). This high sensitivity suggests that $I_G$ is far superior to previous methods for detecting subtle multipartite structures.

Conclusion and Future Directions

This work clarifies the mathematical limits of the QMI matrix and provides a rigorous framework for understanding when and why it fails to be PSD. The introduction of $I_G$ offers a promising new path for measuring correlations in complex quantum systems.

Future research may focus on:

  1. Generalizing Conditions: Finding a simple, closed-form condition for PSD-ness in arbitrary $n$-partite systems.
  2. Physical Protocols: Investigating how the PSD status of a QMI matrix impacts the efficiency of quantum secret sharing or quantum error correction.
  3. Many-Body Physics: Applying the $I_G$ metric to condensed matter systems to determine if it can identify quantum phase transitions more effectively than standard entropy-based measures.

Summary updated

Abstract

For any $n$-partite state $ρ_{A_{1}A_{2}\cdot\cdot\cdot A_{n}}$, we define its quantum mutual information matrix as an $n$ by $n$ matrix whose $(i,j)$-entry is given by quantum mutual information $I(ρ_{A_{i}A_{j}})$. Although each entry of quantum mutual information matrix, like its classical counterpart, is also used to measure bipartite correlations, the similarity ends here: quantum mutual information matrices are not always positive semidefinite even for collections of up to 3-partite states. In this work, we obtain necessary and sufficient conditions for the positive semidefinite quantum mutual information matrix. We further define the \emph{genuine} $n$-partite mutual information which can be easily calculated. This definition is symmetric, nonnegative, bounded and more accurate for measuring multipartite states.
BibTeX
@misc{liu2014necessary,
  author = {Feng Liu and Fei Gao and Su‐Juan Qin and Qiaoyan Wen},
  title = {{Necessary and sufficient conditions for positive semidefinite quantum mutual information matrices}},
  year = {2014},
  eprint = {1410.7245},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1410.7245},
}
Figure