McKean-Vlasov diffusions: from the asynchronization to the synchronization (Q640897): Difference between revisions
From MaRDI portal
Latest revision as of 13:03, 4 July 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | McKean-Vlasov diffusions: from the asynchronization to the synchronization |
scientific article |
Statements
McKean-Vlasov diffusions: from the asynchronization to the synchronization (English)
0 references
21 October 2011
0 references
Given a random variable \(X_0\), consider the real valued diffusion process \[ X_t= X_0+ \sqrt{\varepsilon} B_t- \int^t_0 W^1_s(X_s)\,ds,\tag{1} \] for a potential \(W_s\) of the form \(V+ F* L(X_s)\), where J\(*\) denotes the convolution product and \(L(X_s)\) the law of \(X_s\). The functions \(V\) and \(F\) considered in this note are \({x^4\over 4}- {x^2\over 2}\) and \({\alpha\over 2} x^2\), for some \(\alpha> 0\), respectively. It is known that for \(\alpha\neq 1\) the order of the speed of convergence of the unique symmetric stationary measure for (1) when \(\varepsilon\) goes to \(0\) is \(\varepsilon\). This is in contrast with the case \(\alpha=1\), for which it is of order \(\varepsilon^{1/2}\). The author shows in this note that this difference of bihaviour is removed by the introduction of the parameter \(\varepsilon\) in the potential after a suitable change of scales.
0 references
McKean-Vlasov diffusions
0 references
symmetric stationary measures
0 references