The mining game: a brief introduction to the Stochastic Diffusion Search metaheuristic

In recent years studies of social agents have suggested several new metaheuristics for use in search and optimisation; Stochastic Diffusion Search (SDS) [1] is one such ‘Swarm Intelligence’ algorithm. SDS is a distributed population based search algorithm utilising interaction between simple agents to locate a global optimum; such ‘communicating agents’ have recently been suggested as a potential metaphor for some cognitive processes [6]. SDS is most easily applied to discrete search and optimisation problems where the task is to identify the hypothesis, h, which maximises the value of a decomposable objective function. Unlike many nature inspired search algorithms SDS has a solid mathematical framework which fully describes the behaviour of the algorithm, investigating its: resource allocation [4], convergence to global minima [5], robustness and minimal convergence criteria [2] and time complexity [3]. In the following brief summary we deploy a simple new metaphor the mining game to introduce SDS to readers of the AISBQ. The mining game provides a high-level description of a search to identify the best hill, Hbest, in a large mountain range on which a group of miners should prospect for gold; each hill is quantised into a fixed set of regions, R, where each region yields a specific ‘rate of return’ Rj of gold (concentration). Thus the ‘best’ hill for the miners is the hill, Hi, which maximises the value of the [decomposable] objective function F = Hi ∑ j Rj .