A dynamical characterization of evolutionarily stable states (Q684773)

The paper discusses and compares various definitions of stability in evolutionary game theory. In the simplest setting it is assumed that an individual of a given population can display one out of \(n\) different types of behaviour. If \(x_ i\) represents the relative frequency of individuals displaying behaviour \(i\) then the state space evolution is given by a so-called replicator dynamics \(\dot x= f(x)\), where \(x= (x_ 1,\dots,x_ n)\) belongs to the simplex \(S^ n\). Evolutionary stability is defined in terms of Nash equilibrium like conditions on \(x\in S^ n\). Dynamic stability refers to local stability around a fixed point of \(\dot x= f(x)\) as \(t\to\infty\). These two definitions do not coincide; one reason being that \(\dot x= f(x)\) can have many fixed-points. A purely dynamical characterization of evolutionary stability is not available in a straightforward manner and a main objective of the paper is to provide such a characterization. Two other definitions of stability are discussed within this context as well: global and weak dynamical stability. This characterization is achieved by the investigation of so-called ``derived games'' which consist of a game played with respect to the successful states (fixed-points) of the original game. Biological interpretation of the various mathematical results are given. One of the aims of the author's was to provide a unified presentation of both notation and terminology.