Nonlinear reinforcement learning of dynamic nash equilibria

Abstract

Various firms in a market compete to gain market share or increase revenues or both. [1] analyzed the competition in an oligopolistic market of personal computer microprocessor industry, in which Intel (the incumbent firm) and AMD (the entrant/fringe firm) competed in terms of price, technological innovation and vertical integration. [2], [3] and [4] modelled market duopoly using random game models. Market share and revenue in a competitive market do not remain constant and thus it is highly likely that, in this dynamic environment, the real loss of one rm is not the immediate real gain of the other firm.