Abstract:
This paper presents a study of an algorithm that implements the evolution of the macroscopic state of an arbitrary system in the direction of reproducing and further preserving
the distribution defined on the configuration space of this system. Through methodological analysis and conceptual modifications, we show that this algorithm, originally proposed for physical modeling in thermodynamics and statistical physics, represents a universal mechanism that produces quasi-continuous dynamics of the structured state of a
system with controlled directionality. Meanwhile, all changes, being macroscopic in their
essence, are determined at the microlevel of single “agents”, whose behavior can be regulated within the framework of the heuristic approach. It was demonstrated that this dynamics retains effectiveness and numerical stability in different modes of functional application. As such, we considered a problem for a complete rearrangement of the system
state from one structure to another, a problem to synthesize competing structures within a single state, and a maze solving problem defined at the level of structural topology. The key advantage of the algorithm is the almost complete independence of its performance
and accuracy from both the information volume of the system states and their structural complexity. This allows us to go beyond the known limitations of standard approaches to
numerical modelling and provides new opportunities and perspectives for the digital
twinning applied to the large-scale multifactor systems of different nature.
Keywords:macroscopic modelling, heuristic algorithms, digital twinning.
Fulltext:
PDF file (1017 kB)
First page:
PDF file