Summary: Computer simulation is a computational approach for deducing global system properties through composition of dynamics of local entities and systems. The current state of affairs for this area is characterized by little satisfactory formal theory, and a large portion of very specialized knowledge of subsystems. There is a considerable dependence on experience-based practitioners’ art. Since the use of simulations is becoming increasingly more widespread it is very desirable to have a general framework to describe such systems. The advantage of such a framework is that it allows for development of theory and algorithms with a wide range of applicability as well as providing a basis for system validation.
In this paper we describe theoretical foundations and associated modeling principles for simulation of large, biological, and socio-technical systems. Examples of such systems are urban transportation systems, large-scale epidemiology and disease dynamics, gene regulatory networks, and the national electrical power markets. These systems are all composed of large numbers of interacting individuals, physical and technological components or entities. The global dynamics result from sequential composition of local interaction between individual components.
A mathematical theory for these systems must incorporate entity description and interaction properties such as sequential evaluation and composition. The class of Sequential Dynamical Systems (SDS) is a framework specifically designed for modeling and analyzing large classes of computer simulations. SDS allow for mathematical and computational analysis of the underlying systems. We give an overview of the mathematical and computational theory of interaction-based simulations based on SDS. Illustrative problems in functional genomics, urban traffic modeling and social networking are used to show the utility of SDS-based modeling. Next, we discuss engineering methods and principles that allow us to specify, design, and analyze simulations of extremely large systems and implementations of these on high-performance computing architectures. As a practical example, we describe an interaction-based computer modeling framework called ‘Simfrastructure’ that is used to study large interdependent societal infrastructures. Two specific modules are then described in additional detail. The synthetic proto-population module generates synthetic individuals and their daily activities in a built urban environment. The SimDemics modeling framework is useful in studying the spread of infectious diseases. These modules demonstrate how specific models can be integrated in Simfrastructure.
For the entire collection see [Zbl 1119.92001].