From nonlinear Fokker-Planck equations to solutions of distribution dependent SDE (Q782406)

The authors prove the existence of a weak solution to the McKean-Vlasov stochastic differential equation \[ dX(t) =b(t,X(t), L_{X(t)})dt+\sigma (t,X(t), L_{X(t)})dW(t), \] where \(W(t)\) is a Brownian motion and \(L_{X(t)}\) denotes the law of \(X(t)\). The starting point of the equation \(X(0)\) is assumed to be a random variable admitting a density with the respect to Lebesgue measure. The authors derive their conclusion under a weak non-negative definite condition associated with the matrix \(\sigma\), which allows for degeneracy in the equation. The result is obtained by first proving the existence of a weak solution to a related non-linear Fokker-Planck equation (FPE) under similar hypotheses to those imposed in the main theorem, and then appealing to a superposition principle. Along the way existence results are proved for both strongly elliptic and degenerate versions of the FPE.