Two-sided estimates on the density of Brownian motion with singular drift (Q2505482)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Two-sided estimates on the density of Brownian motion with singular drift
scientific article

    Statements

    Two-sided estimates on the density of Brownian motion with singular drift (English)
    0 references
    0 references
    0 references
    26 September 2006
    0 references
    Let \(\mu=(\mu^1,\dots,\mu^d)\) be a family of signed measures on \({\mathbb R}^d\) belonging to the Kato class \({\mathbf K}_{d,1}\). The Brownian motion on \({\mathbb R}^d\) with drift \(\mu\) is formally defined as the diffusion process \((X_t,{\mathbf P}_x)\) with generator \({1\over 2}\Delta+ \mu\cdot \nabla\). Rigorous definition and the proof of existence of \(X_t\) are shown by \textit{R.~F.~Bass} and \textit{Z.~Q.~Chen} [Ann. Probab. 31, No. 2, 791--817 (2003; Zbl 1029.60044)]. Denote by \(q^\mu(t,x,y)\) the transition desity of \(X_t\). In this paper, two-sided Gaussian estimates of \(q^\mu\) are given, that is, it is shown that \(q^\mu(t,x,y)\) is bounded from above and below by \(c_1 e^{-c_2 t} t^{-d/2}\exp(-c_3| x-y| ^2/2t)\) for some constants \(c_i\). For a bounded smooth domain \(D\), similar estimates of the transition function \(q^{\mu,D}(t,x,y)\) of the part process \(X^D_t\) of \(X\) in \(D\) are given. The proof is performed by approximating \(\mu^i\) by \(U_n^i(x)dx\) for smooth functions \(U^n_i\) and showing that the transition function \(q_n(t,x,y)\) of the diffusion process \(X^n_t\) corresponding to \({1\over 2}\Delta+U^n\cdot \nabla\) converges uniformly to \(q^\mu(t,x,y)\). These estimates are used to prove the Harnack principle for positive harmonic functions. For a bounded \(C^{1,1}\) domain, a boundary Harnack principle of harmonic functions in \(D\) is also given by using the corresponding estimates of \(q^{\mu,D}\). In this case, the coincidence of the Martin boundary of \(X^D\) with the Euclidean boundary is also given.
    0 references
    0 references
    Gaussian estimate
    0 references
    Harnack principle
    0 references
    Martin boundary
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references