Descriptive Thermodynamics

TL;DR

This paper develops a thermodynamic framework for describing complex, empirically accessible systems like communication networks, focusing on data reduction and empirical temperature measurement rather than prediction.

Contribution

It introduces a method to compute an empirical temperature for finite systems, enabling thermodynamics to be applied to complex, data-rich systems such as networks.

Findings

Empirical temperature can be defined for complex systems.

Thermodynamics can be adapted for systems where prediction is infeasible.

Application to TCP/IP networks illustrates practical relevance.

Abstract

Thermodynamics (in concert with its sister discipline, statistical physics) can be regarded as a data reduction scheme based on partitioning a total system into a subsystem and a bath that weakly interact with each other. The ubiquity and applicability of the scheme chiefly derives from that of partitioning protocols in experiments and observations. Whereas conventionally, the systems investigated require this form of data reduction in order to facilitate prediction, a different problem also occurs, in the context of communication networks, markets, etc. Such "empirically accessible" systems typically overwhelm observers with the sort of information that in the case of (say) a gas is effectively unobtainable. What is required for such complex interacting systems is not prediction (this may be impossible when humans besides the observer are responsible for the interactions) but rather,_description_ as a route to understanding. Still, the need for a thermodynamical data reduction scheme remains. In this paper, we show how an empirical temperature can be computed for finite, empirically accessible systems, and further outline how this construction allows the age-old science of thermodynamics to be fruitfully applied to them. The particular example of TCP/IP networks will be briefly discussed.