Singular McKean-Vlasov SDEs: well-posedness, regularities and Wang's Harnack inequality (Q2680398)

Consider the following \(d\)-dimensional McKean-Vlasov stochastic differential equation \[ dX_t=b_t(X_t, \mathscr{L}_{X_t})dt + \sigma_t(X_t) dW_t, \quad t\in [0,T], \] where \(\mathscr{L}_{X_t}\) denotes the probability distribution of the random variable \(X_t\), and \(W\) is an \(m\)-dimensional Brownian motion. In this paper the author establishes the well-posedness of the above equation in the set \(\mathscr{P}_V\) of probability measures \(\mu\) on \(\mathbb{R}^d\) such that \(\int_{\mathbb{R}^d} Vd\mu <\infty\), where \(V\ge 1\) is a twice continuously differentiable function with compact level sets, assuming some technical conditions on the drift and the diffusion coefficient. In particular, \(\sigma\sigma^*\) is supposed to be invertible and the drift contains a locally standard integrable term and a superlinear term in the spacial variable, and it is Lipschitz continuous in the distribution variable with respect to a weighted variation distance. The author also proves regularity estimates in the initial distribution. When the superlinar term is strengthened to be Lipschitz continuous, a dimension-free Wang's Harnack inequality is established. The results are first derived for classical singular stochastic differential equations, extending previous works on this subject.