This paper introduces entropic conjugation as a new framework to analyse high-order interdependencies in complex systems, clarifying existing information-theoretic measures and revealing new ones. The framework allows to disentangle symmetric and skew-symmetric measures, highlighting the O-information as a unique, computationally efficient, and naturally emerging axis of variation of high-order interdependencies in real-world and simulated systems.
- Fernando E. Rosas
- Aaron J. Gutknecht
- Michael Gastpar
