Multiscale Structure of More-than-Binary Variables
arXiv:1705.03927
Abstract
In earlier work, my colleagues and I developed a formalism for using information theory to understand scales of organization and structure in multi-component systems. One prominent theme of that work was that the structure of a system cannot always be decomposed into pairwise relationships. In this brief communication, I refine that formalism to address recent examples which bring out that theme in a novel and subtle way. After summarizing key points of earlier papers, I introduce the crucial new concept of an ancilla component, and I apply this refinement of our formalism to illustrative examples. The goals of this brief communication are, first, to show how a simple scheme for constructing ancillae can be useful in bringing out subtleties of structure, and second, to compare this scheme with another recent proposal in the same genre.
8 pages, 0 figures, 1 term from quantum information theory reappropriated for classical ends; v2: expansions and edits in response to feedback
References in corpus (8)
- Unknown Quantum States: The Quantum de Finetti Representation
- Network information and connected correlations
- Quantifying unique information
- Nonnegative Decomposition of Multivariate Information
- Multivariate Dependence Beyond Shannon Information
- SIC-POVMs and Compatibility among Quantum States
- Negativity Bounds for Weyl-Heisenberg Quasiprobability Representations
- The Partial Entropy Decomposition: Decomposing multivariate entropy and mutual information via pointwise common surprisal