On the density for the solution of a Burgers-type SPDE (Q1304921)

Let \(\dot W\) denote a time-space white noise on \([0,T]\times [0,1]\). The author studies existence and regularity of densities for the solution of the stochastic partial differential equation of Burgers type \[ {\partial \over\partial t} X={\partial^2 \over\partial x^2}X+ {\partial \over \partial x}\bigl(g(X)\bigr) +f(X)+ \sigma(X) \dot W(t,x) \] with boundary condition \(X(0,x)=X_0(x)\), \(X(t,0)=X(t,1)=0\). Gyöngy (1998) has proved existence and uniqueness of the solution for this equation by generalizing earlier results of Walsh (1986), Da Prato/Debussche/Temam (1995) and Da Prato/Gatarek (1995). Assuming the smoothness of the coefficients and bounded derivatives as well as the non-degeneracy of \(\sigma\), the author shows by using Malliavin calculus adapted to white-noise driven parabolic SPDE (Bally/Pardoux, 1999) that, for all \(0<x_1<\cdots<x_d<1\), \(t>0\), \((X(t,x_1), \dots, X(t,x_d))\) admits a \(C^\infty\)-density. If \(g\) is of square growth, a localization argument allows him to apply a criterion by Bouleau/Hirsch (1991) to show that \((X(t,x_1),\dots,X(t,x_d))\) has a density. The problem of its regularity remains open. The paper improves results by Lanjri/Nualart (1997) obtained for \(d=1\).