Toward a unified view of complex multiscale stochastic systems: A generalized theory of interactions and its computational infrastructure for their universal and efficient investigation

The Internet, societies and the brain exemplify systems whose compositional structure contains numerous interacting parts and coupled scales of irreversible action, driven to a large extent by intrinsic random perturbations that cannot be removed without losing critical information for their scientific understanding. Systems that match this signature are ubiquitous across a large number of knowledge domains, yet no unified model explaining all their commonalities and unifying principles exists. We define the class containing these instances as that of complex multiscale systems (CMSS), including naturally evolved and artificially designed systems. Due to the increasing interconnectedness of the human experience and the need to understand a wider range of phenomena whose description is much richer than that provided by current models, we claim that the development of effective CMSS research methods represents a scientific program of a new kind. This work provides evidence underlining why CMSS constitutes a novel and fruitful scientific domain. We concern ourselves with improving the existing understanding of CMSS by finding causal mappings between the behavior of parts of the system at a small scale (microscale) giving rise to larger components at a larger scale (macroscale) through theory development, cyberinfrastructure and case studies. Three studies contained here suggest that placing interactions as a central construct --instead of objects and laws- simplifies the analysis of CMSS and brings clarity across the formal structure of related scientific theories. We present a Generalized Theory of Interactions (GToI) that aims to provide a universal and efficient description of CMSS in general; the theory can be conveniently adjusted to match specific phenomena across a wide variety of disciplines. We evaluate existing agent-based systems and discuss the ongoing development of a new agent-based framework benefiting from the perspective of the interaction. We demonstrate how explicit interaction models of CMSS can efficiently increase information gain about these systems, including their scaling laws, self-organization properties and collective behavior with two examples: one on social perception, and another one on modeling the COVID- 19 epidemic for the cities of Urbana and Champaign. Finally, we explore philosophical, conceptual and pragmatic externalities of interactions in computer-assisted epistemology, theoretical biology, global studies and artificial intelligence.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Informatics

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2021

Author dc:creator

• Nunez-Corrales, Santiago

Contributors dc:contributor

• Jakobsson, Eric
• Caetano-Anollés, Gustavo
• Ludäscher, Bertram
• Kirkpatrick, Kay
• Turk, Matt

• complex multiscale stochastic systems
• generalized theory of interactions
• statistical physics
• agent-based modeling and simulation
• stochastic phenomena
• thermodynamic irreversibility

Statement dc:rights

• © 2020 by Santiago Núñez-Corrales. All rights reserved.

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/109481

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/109481