Information Physical Artificial Intelligence in Complex System Dynamics

Breaking Frontiers in Nonlinear Analytics, Model Design and Socio-Environmental Decision Support in a Coevolutionary World

Publication briefs

Publication type: Scientific Report
Author: Rui A. P. Perdigão
Date: 2020, September 30
Title: Information Physical Artificial Intelligence in Complex System Dynamics: Breaking Frontiers in Nonlinear Analytics, Model Design and Socio-Environmental Decision Support in a Coevolutionary World
DOI: https://doi.org/10.46337/200930
Indexed: Yes (Crossref)

Cite as: Perdigão, R.A.P. (2020): Information Physical Artificial Intelligence in Complex System Dynamics: Breaking Frontiers in Nonlinear Analytics, Model Design and Socio-Environmental Decision Support in a Coevolutionary World. https://doi.org/10.46337/200930.

Methodological Keywords: Information Physics, Information Theory, Complex Systems, Dynamical Systems, Mathematical Physics, Non-Ergodic, Chaos, Entropy, Emergence, Synergy, Coevolution, Causation.
Applied Keywords: Evolutionary Cognition, Big Data Analytics, Artificial Intelligence, Machine Learning, Earth System Dynamics, Hydrologic Dynamics, Hydrologic Analytics, Socio-Environmental Dynamics, Decision Support.

For further information and queries, please contact: mdsc[at]meteoceanics.eu

Other formats

A summarized version of this work was orally presented at the cE3c Annual Meeting 2020 on October 1st, 2020.

Abstract

Classical statistical, information-theoretic, machine learning and artificial intelligence techniques capture only aggregate recurrence-based information, whilst ignoring non-ergodic dynamic entanglement at microstate, event-scale level, along with non-recurrent structural-functional coevolutionary interplay.

In order to overcome these issues, the non-ergodic theories of information physics and synergistic dynamic complexity formulated in Perdigão (2017, 2018, 2020) generalized dynamical systems theories, machine learning information retrieval and nonlinear information theory to far-from-equilibrium nonlinear statistical mechanics with event codependence and multiscale multidomain system dynamic innovation.

In the present study this theory is further developed to elicit microphysical coevolution underlying non-ergodic entanglement at the roots of evolutionary complexity from quantum microphysics to classical macrophysics. This brings out a new dynamic framework for coevolutionary information retrieval and synergistic dynamic model design, unveiling hidden information beyond nonlinear spatiotemporal memory loss and extending the predictability horizons in complex system dynamics. The findings are illustrated to elicit hidden structure and predictability in adaptive quantum and classical networks, nonlinear fluid wave resonators, followed by biogeophysical applications within the coevolutionary Earth system dynamics.

Our theoretical and methodological innovations are deployed on our novel hybrid information physical artificial intelligence engine, thereby providing a leading edge on advance prediction of extreme events including floods, droughts, heatwaves, wildfires, along with socio-environmental impacts, crucial to provide decision support to relevant authorities protecting our society and the environment, from local micrometeorology to large scale climate and overall earth system dynamics.

All in all, our contribution ranges from fundamental mathematical physics and complex system science to socio-environmental model design to improve predictability and preparedness for such critical phenomena as otherwise elusive “black swan” extreme events in a changing climate.