Unreliability of mutual information as a measure for variations in total correlations

S. Alipour*, S. Tuohino, A. T. Rezakhani, T. Ala-Nissila

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

6 Citations (Scopus)
111 Downloads (Pure)


Correlations disguised in various forms underlie a host of important phenomena in classical and quantum systems, such as information and energy exchanges. The quantum mutual information and the norm of the correlation matrix are both considered as proper measures of total correlations. We demonstrate that, when applied to the same system, these two measures can actually show significantly different behavior except at least in two limiting cases: when there are no correlations and when there is maximal quantum entanglement. We further quantify the discrepancy by providing analytic formulas for time derivatives of the measures for an interacting bipartite system evolving unitarily. We argue that to properly account for correlations one should consider the full information provided by the correlation matrix (and reduced states of the subsystems). Scalar quantities such as the norm of the correlation matrix or the quantum mutual information can only capture a part of the complex features of correlations. As a concrete example, we show that in describing heat exchange associated with correlations neither of these quantities can fully capture the underlying physics. As a byproduct, we also prove an area law for quantum mutual information in systems with local and short-range interactions, without need to assume Markovianity or final thermal equilibrium.

Original languageEnglish
Article number042311
Pages (from-to)1-6
Number of pages6
JournalPhysical Review A
Issue number4
Publication statusPublished - 7 Apr 2020
MoE publication typeA1 Journal article-refereed




Dive into the research topics of 'Unreliability of mutual information as a measure for variations in total correlations'. Together they form a unique fingerprint.

Cite this