Modeling of complex systems II: A minimalist and unified semantics for heterogeneous integrated systems
Introduction
The concept of complex systems has led to various definitions in numerous disciplines (biology, physics, engineering, mathematics, computer science, etc). One speaks for instance of dynamical, mechanical, Hamiltonian, hybrid, holonomic, embedded, concurrent or distributed systems (cf. [2], [4], [21], [24], [28], [32]). A minimalist fuzzy definition consistent with (almost) all those of the literature is that a ‘system’ is “a set of interconnected parts forming an integrated whole”, and the adjective ‘complex’ implies that a system has “properties that are not easily understandable from the properties of its parts”. In the mathematical formalization of “complex systems”, there are today two major approaches: the first one is centered on understanding how very simple, but numerous, elementary components can lead to complex overall behaviors (e.g. cellular automatas), the second one (that will also be ours) is centered on giving a precise semantics to the notion of system and to the integration of systems to build greater overall systems.
When mathematically apprehended, the concept of system (in the sense of this second approach) is classically defined with models coming from:
- •
control theory and physics, that deal with systems as partial functions (dynamical systems may also be rewritten in this way), called transfer functions, of the form:where and y are inputs, states and outputs dataflows, and where T stands for time (usually considered in these approaches as continuous (see [32], [1], [12]).
- •
theoretical computer sciences and software engineering, with systems that can be depicted by models equivalent to timed Turing machines with input and output, evolving on discrete times generally considered as a universal predefined sequence of steps (see for instance [19], [5], [16]).
The theory of hybrid systems was developed jointly in control theory (see [32], [34]) and in computer science (see [2], [3], [22]) to address this challenge. A serious issue with this theory is however that the underlying formalism has some troubling properties such as the Zeno effect which corresponds to the fact that an hybrid system can change of state an infinite number of times within a finite time (because of the convergence of series of durations) that one usually prefers to avoid in a robust modeling approach. Moreover, it does not allow to consider various time scales of heterogeneous granularity (which will be the central point of our approach). Other interesting and slightly different attempts in the same direction can also be found in Rabinovitch and Trakhtenbrot (see [27], [33]) who tried to reconstruct a finite automata theory on the basis of a real time framework, or in [35].
In the literature about (complex) systems, the real object and its model are often confused and both called “system”. We will call a real system any object of the real world which transforms flows of data. We will call system the mathematical object introduced to model real systems. In this paper, we are interested in modeling the functional behavior of real systems, and their integration. Thus, we will model real systems as functional blackboxes (with an internal state), whose structure and behaviors can be described by the recursive integration of heterogeneous smaller subsystems (thus considering complex systems as heterogeneous integrated systems). We will thus focus on two aspects of the complexity of systems:
- •
the heterogeneity of systems (modeled following continuous or discrete time, and exchanging data of different types, - informational, material or energetic).1
- •
the integration of systems, i.e. the mechanism to construct a system resulting from the composition of smaller systems, whose behaviors may be described at a more concrete level (i.e. a finer grain).
This paper is the second of a series on “Modeling of Complex Systems”. Indeed, we generalize the approach of the first paper [7] (where a unified framework for continuous and discrete systems was defined by using non-standard infinitesimal and finite time steps) by dealing with time, data, and synchronization axiomatically, and by introducing integration operators. The purpose of this second paper is to give a unified and minimalist semantics for heterogeneous integrated systems and their integration. By “unified”, we mean that we propose a unified model of real systems that can describe the functional behavior of heterogeneous systems and that is closed under integration. By “minimalist” we mean that our formalization intends to provide a small number of concepts and operators to model the behaviors and the integration of (complex) real systems. We believe that our work allows to give a relevant formal semantics for concepts and models typically used in Systems Engineering, where semi-formal modeling is well-spread. The paper is organized as follows:
- •
in Section 2 Time, 3 Data, we introduce unified definitions of time (both continuous and discrete) and data (with various behaviors) to handle heterogeneous components and encompass classical approaches. We also define a generic synchronization for dataflows,
- •
in Section 4, we introduce a formal definition of systems as unified functional objects modeling heterogeneous real systems,
- •
in Section 5, we introduce minimalist operators for integrating systems (with closure of the definition of system) and prove that they are consistent with classical concepts of integration formalized on transfer functions.
Section snippets
Time
Most of the challenges raised by the unified definition of heterogeneous integrated systems are coming from time. Indeed, real systems are naturally modeled according to various time scales (modeling discrete or continuous time), and we must therefore be able to define:
- •
a unified model of time encompassing continuous and discrete times to later introduce a unified definition of heterogeneous systems,
- •
the mixture of various time scales for integrating systems.
Unifying both discrete and continuous
Data
Another challenge to address to model complex systems is the heterogeneity of data (modeling any element that can be exchanged between real systems) and of their synchronization between different time scales. We introduce datasets that will be used for defining data transmitted by dataflows. The dataflows will be used to describe variables of systems (inputs, outputs and states), and we define the synchronization of dataflows between time scales.
Systems
We introduce the definition of a system as a timed Mealy machine, and show that it can be represented by a transfer function.
Integration operators
We propose three elementary operators allowing to model systems integration, i.e. to build greater systems from a set of elementary systems by recursive application of composition operators and abstraction operator.
Conclusion
We have introduced a minimalist and unified semantics for heterogeneous integrated systems. This semantics allows us to capture two very important properties of complex systems: heterogeneity (being able to deal with various types of systems through rich time & data) and recursive integration (taking into account the integrative dimension of complex systems that are build recursively with multiple levels of components).
This work is the theoretical part of a broader project aiming at building an
References (35)
- et al.
The algorithmic analysis of hybrid systems
Theor. Comput. Sci.
(1995) - et al.
The algebra of stream processing functions
Theor. Comput. Sci.
(2001) - et al.
Timed rewriting logic with an application to object-based specification
Sci. Comput. Progr.
(1997) - et al.
Systems Engineering
(1992) - et al.
The algorithmic analysis of hybrid systems
- et al.
Concurrent systems: an integrated approach to operating systems
(2002) The Foundations of Esterel
(2000)- et al.
Systems Engineering and Analysis
(1998) - et al.
Modelling of complex systems: systems as dataow machines
Fundam. Inform.
(2009) - I.S. Board, IEEE Standard VHDL Language Reference Manual (IEEE Std. 1076-1993), IEEE...
Specification and development of interactive systems: focus on streams, interfaces, and refinement
Fundamentals of Modeling and Analysing Engineering Systems
The Algebraic Theory of Semigroups, Math. Surveys 7
Amer. Math. Soc., R. I.
Analyse Non Standard
Cited by (12)
Constraint based approach for the steady-state simulation of complex systems: Application to ship control
2013, Engineering Applications of Artificial IntelligenceA modular formal semantics for Ptolemy
2013, Mathematical Structures in Computer ScienceModel-based Systems Architecting: Using CESAM to Architect Complex Systems
2022, Model-based Systems Architecting: Using CESAM to Architect Complex SystemsSocial media, cyberculture, blockchains, and education: A new strategy for Brazilian higher education
2019, Blockchain Technology Applications in EducationDynamic networks of heterogeneous timed machines
2018, Mathematical Structures in Computer ScienceIterative prototyping methodology for the development of innovative and dependable complex embedded systems through SPC&KPI techniques
2018, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)