Large systems of diffusions interacting through their ranks (Q424497)

The author studies the behavior of the weak solutions to the system of stochastic differential equations \[ \mathrm{d} X_i(t) = \mu(F_{\gamma^N(t)}(X_i(t)))\,\mathrm{d} t + \sigma(F_{\gamma^N(t)}(X_i(t))) \,\mathrm{d} B_i(t), \qquad i=1,\dotsc,N, \] on an interval \([0,T]\) as \(N\to\infty\), where \(\gamma^N(t):=\frac{1}{N}\sum_{i=1}^N \delta_{X_i(t)}\) is the empirical measure of the particle system \(X_1(t),\dotsc,X_N(t)\) at time \(t\), \(F_{\gamma^N(t)}\) is its cumulative distribution function, \(\mu\) and \(\sigma\) are measurable functions on \([0,1]\) taking values in \(\mathbb{R}\) and \((0,\infty)\), respectively, and \(B_1,\dotsc, B_N\) are independent standard Brownian motions. Informally, at any time \(t\) the drift and the diffusion coefficients of a fixed particle \(i\) are determined by its rank in the particle configuration \(X_1(t),\dotsc,X_N(t)\) at time \(t\), so, whenever a particle changes its rank, the coefficients change accordingly. Existence and uniqueness of a weak solution to this system for any fixed natural number \(N\) are essentially due to \textit{R. F. Bass} and \textit{E. Pardoux} [Probab. Theory Relat. Fields 76, 557--572 (1987; Zbl 0617.60075)]. Considering the above equation as an equation for the evolution of the empirical measure \(\gamma^N(t)\) of the particle system on \([0,T]\), the author proves that, under suitable assumptions on \(\mu\) and \(\sigma\), and the stationarity assumption on the initial vector of spacings between the particles, the limiting evolution is governed by a McKlean-Vlasov evolution equation. Moreover, it was shown that, under the same assumptions, the cumulative distribution function of the system evolves according the generalized porous medium equation with convention. Possible applications of the results for rank-based models of capital distributions in financial markets are also explained.