The long-run behavior of the stochastic replicator dynamics (Q1774207): Difference between revisions

The author considers the normalized stochastic process \(X_j(t)= Z_j(t)\sum^n_{k=1} Z_k(t)\) of \(Z_j(t)\), where \(Z_j(t)\), \(j= 1,\dots, n\), is the solution of the stochastic equation \[ dZ_j(t)= Z_j(t)\sum^n_{k=1} a_{jk} Z_k(t)\,dt+ Z_j(t)\sigma_j dW_j(t), \] modelling a game theory type population dynamics, and investigates the behaviour of \(\vec X(t)= (X_1(t),\dots, X_n(t))\) with respect to the evolutionarily stable strategies (ESS). Some estimates of frequency of \(\vec x(t)\) in a neighbourhood of an ESS \(\vec p\) or of distance between \(\vec x(t)\) and \(\vec p\) are established. Also an estimate for \(X_k(t)\) is given in the case that the ``strategy'' \(k\) is dominated by a ``mixed strategy''. If \(\sigma_j\) are small enough and \(\vec p_k= (0,\dots, 0,1,0\dots, 0)\) is an ESS, then the convergence in some sense of \(\vec x(t)\) to \(\vec p_k\) is proved. An example is illustrated with details.