A two-layer identification strategy for the development of stochastic models of the travelling traders’ exchange problem

Abstract

The travelling traders’ exchange problem (TTEP) is a general mathematical problem arising in a number of applications where the purpose is to characterise the distribution of money over time related to a population of traders which can move in space and interact with each other. Results from stochastic simulations of TTEP models can be analysed over time in terms of i) standard deviation (STD); ii) probability density function (PDF) of the observations in time. A two-layer model identification strategy is proposed in this paper for the development of time-dependent nonlinear regression models from the results of TTEP computational stochastic simulations. The models are capable of representing the money distribution as a function of the TTEP operating parameters, paving the way to a new framework for model identification.