The doubling property of the elliptic measure, for elliptic operators with drifts satisfying an average diverging condition.
Aritro Pathak · 2025
Read The doubling property of the elliptic measure, for elliptic operators with drifts satisfying an average diverging condition. by Aritro Pathak (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-12942.txt.
Introduction
Aritro Pathak proves that the elliptic measure doubles nicely for operators whose drift blows up near the boundary in an averaged sense. The elliptic measure tracks how boundary values influence solutions of an elliptic equation. A doubling property, where enlarging a boundary ball at most multiplies the measure by a controlled factor, is basic to its analysis. Earlier results assumed the drift was pointwise small. Pathak only needs smallness on average over Whitney cubes, a Carleson-type condition on the drift's size.
The results hold in one-sided chord-arc domains and are new even for the half-space. Along the way he proves Hardy inequalities in domains whose boundaries are Ahlfors-David regular, using a stopping time argument. The work connects to recent results on Dirichlet problems in rough domains.
Catalog record
| CREATOR | Aritro Pathak |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 67 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-12942
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The doubling property of the elliptic measure, for elliptic operators with drifts satisfying an average diverging condition. Aritro Pathak Abstract We show doubling of the elliptic measure corresponding to the operator with an elliptic principal term and a drift that diverges, on average on Whitney cubes, like the inverse distance to the boundary, with a small constant. Essentially a small Carleson constant assumption on the drift, this generalizes earlier results with the hypothesis of pointwise smallness of such a drift. This relates to recent perturbative results of rough Dirichlet solvability in domains with drifts or potentials that satisfy a Carleson measure condition, which have also been considered earlier by Hofmann-Lewis and Kenig-Pipher. While we work in 1-sided chord arc domains, these results are new even for the half-space. In the process, we also prove Hardy inequalities in such domains with Alhfors-David regular boundary, using a stopping time argument. 1 Introduction. Consider a 1-sided chord arc domain \Omega\subset\mathbb{R}^{n} with n\geq 3 . Consider the linear operator \displaystyle Lu=-\nabla\cdot(\mathcal{A}\nabla u)+\mathcal{B}\cdot\nabla u, (1) We assume the drift term is small in an averaged sense over each Whitney region, which is to say that there exists a \beta small enough so that for each Whitney cube U_{Q} in the Whitney cube decomposition of \Omega we have, \displaystyle(1):\int_{U_{Q}}|\mathcal{B}|^{2}\delta(X)dV\leq\beta^{2}l(Q)^{n-1}, (2) Here |\mathcal{B}| denotes the magnitude of the drift term. This is a generalization of the case considered in Chapter I of [ HL01 ] , with the pointwise smallness estimate of the form \displaystyle|\mathcal{B}(x)|\leq\frac{\beta}{\delta(x)}, (3) for some small enough \beta . Here we also require \displaystyle(2):\Bigg(\sum\limits_{i=0}^{n}\delta(X)|\mathcal{B}_{i}(X)|+\sum\limits_{i,j=1}^{n}|\mathcal{A}_{ij}(X)|\Bigg)<M<\infty,\ \ \sum\limits_{i,j=1}^{n}\mathcal{A}_{ij}(X)\xi_{i}\xi_{j}\geq\lambda|\xi|^{2}. (4) This enables the use of the Harnack inequality in our arguments, and also is important in establishing the Bourgain type estimate for the elliptic measure for this operator. Also note that the adjoint equation takes the form: \displaystyle L^{*}u=-\nabla\cdot(\mathcal{A}^{*}\nabla(u)+\mathcal{B}u), (5) We note that solutions for the adjoint equation in general are not a-priori known to exist, whereas the Lax-Milgram lemma can be used in a routine way to show existence of solutions for Lu=f for locally integrable f , once one shows coercivity of the corresponding bilinear form with arguments analogous to those used in proving Theorem 10 . One motivation for this work comes from perturbative results such as [ HP24 ] that establishes L^{p} Dirichlet solvability for the operator L when the drift satisfies a smallness assumption in a Carleson measure sense as in Eq. 7 , when we know that the L^{p} Dirichlet problem for the operator \displaystyle L_{0}u=-\nabla\cdot(\mathcal{A}\nabla u) (6) is solvable. In turn, such a question with the drifts is motivated by the question of the rough Dirichlet solvability for the elliptic operator in non-divergence form, since the non-divergence form operator can be written as a divergence form expression along with a drift. This means that for every Q\in\mathbb{D}(\partial\Omega) , we have, \displaystyle\sup\limits_{x\in\partial\Omega,0<r<r_{0}}\frac{1}{\sigma(\Delta(x,r))}\int\limits_{\Omega\cap B(x,r)}\sup\limits_{B(t,\delta(t)/2)}\Big(|\mathcal{B}|^{2}(y)\delta(y)\Big)dt<\infty, (7) and further we also have the pointwise bound of the form |\mathcal{B}(x)|\leq M/\delta(x) coming from Eq. 4 . The Carleson measure condition of Eq. 7 implies with a standard pigeonholing argument that, for a given small \epsilon>0 , for all but a finite number of Whitney cubes contained in B(x,r) , \displaystyle\int\limits_{U_{Q}}\sup\limits_{B(t,\delta(t)/2)}\Big(|\mathcal{B}|^{2}(y)\delta(y)\Big)dt<\epsilon l(Q)^{n-1}. (8) Further, the supremum condition in the integrand forces a pointwise smallness estimate in each Whitney cube of this type. Such a condition on the drift has been considered first in [ KP01 ] . There, the authors also proved a result for Dirichlet solvability for operators with only the principal elliptic term, where the elliptic matrices satisfied the ’DKP’ type condition: \displaystyle\sup\limits_{x\in\partial\Omega,0<r<r_{0}}\frac{1}{\sigma(\Delta(x,r))}\int\limits_{\Omega\cap B(x,r)}\sup\limits_{B(t,\delta(t)/2)}\Big(|\nabla\mathcal{A}|^{2}(y)\delta(y)\Big)dt<\infty, (9) In other results, one establishes pointwise estimates on the Green function, found most recently in [ HP24 ] where one has the pointwise smallness condition of Eq. 3 on the drift. Further, the so called ‘Bourgain’ property’, for the operator with the Eq. 3 assumption, is also found in Chapter 1 of [ HL01 ] . These two ingredients combine to establish the doubling of the elliptic measure, by an argument found in [ Ai06 ] , as mentioned in Theorem 2 of this paper. We note that these two ingredients separately are of independent interest. For example, in [ HL18 ] , the Bourgain property for the elliptic measure corresponding to the elliptic operator without any lower order terms, is essentially used to prove the fact that an a-priori assumption of BMO-solvability implies L^{p} Dirichlet solvability. In a separate manuscript, we are studying the operator Eq. 1 with the conditions of Eqs. 7 and 4 , with the hypothesis of BMO-solvability in the half space. The pointwise estimates on the Green function with lower order terms is itself of interest in other areas of partial differential equations [ Mour23 , Sak21 , KS19 ] . Although the lower pointwise estimates are proved by adopting the techniques of [ GW82 ] , the upper pointwise estimates require more work, and crucially needs a Calderon Zygmund type decomposition of the Whitney cubes. The Bourgain property in this instance also needs a new argument, in controlling an upper bound on terms of the form \int|\mathcal{B}|^{2}u^{2}\psi^{2}dV for local solutions to Lu=0 and \psi some radially symmetric test function centered on the boundary. In presence of pointwise estimates of Eq. 3 , as in [ HL01 ] , this is somewhat routine. However, with the hypothesis of only Eq. 2 on the drift, the arguments become more delicate, in the general setting of the 1- sided chord arc domains. In [ HMT17 ] , it is pointed out that Dirchlet solvability holds by weakening the hypothesis of Eq. 9 in (ii) below, along with the local Lipschitz condition on \mathcal{A} as well as a pointwise estimate on the gradient of |\nabla\mathcal{A}| : \displaystyle(i)\mathcal{A}\ \text{in Lip}_{loc}(\Omega),\ \ \ |\nabla\mathcal{A}(x)|\leq\frac{M}{\delta(x)} (10) \displaystyle(ii)\sup\limits_{x\in\partial\Omega,0<r<r_{0}}\frac{1}{\sigma(\Delta(x,r))}\int\limits_{\Omega\cap B(x,r)}\Big(|\nabla\mathcal{A}|^{2}(y)\delta(y)\Big)dt<\infty, (11) Thus it is natural to ask whether the results of [ HP24 ] can be proved for more general operators with a Carleson measure condition on the drift, where the supremum condition within the integrand is removed, \displaystyle\sup\limits_{x\in\partial\Omega,0<r<r_{0}}\frac{1}{\sigma(\Delta(x,r))}\int\limits_{\Omega\cap B(x,r)}\Big(|\mathcal{B}|^{2}(y)\delta(y)\Big)dt<\infty. (12) This means that for Whitney cubes that typically satisfy the smallness assumption, instead of Eq. 8 and thus the pointwise smallness estimate, we only have the average smallness assumption of Eq. 2 considered in this paper, along with Eq. 4 . Because of the Carleson measure condition, by an elementary counting argument, depending on the Carleson measure constant and \beta , one sees that all but a finite number of Whitney cubes contained in any \Omega\cap B(x,r) have to satisfy the average smallness condition of Eq. 4 . This paper establishes doubling of the elliptic measure for this more general situation of drifts that have a smallness condition Eq. 2 on average on every Whitney cube in a chord arc domain, instead of having a pointwise smallness assumption, along with a background pointwise bound with the large constant as in Eq. 4 . This result is new even for the half space. We also note that the Carleson measure involves a square of the drift term. If one worked with a Carleson measure condition of the form \displaystyle\sup\limits_{x\in\partial\Omega,0<r<r_{0}}\frac{1}{\sigma(\Delta(x,r))}\int\limits_{\Omega\cap B(x,r)}|\mathcal{B}|(y)dt<\infty, (13) that has a linear drift term in the integrand, the point-wise estimate on the drift in Eq. 4 along with Eq.