Processing math: 0%
Your browser does not support the canvas element.

Jincheng Yang

Institute for Advanced Study

%--Paired Delimiters-- \newcommand{\abs}[1]{\left\lvert #1 \right\rvert} \newcommand{\ang}[1]{\left\langle #1 \right\rangle} \newcommand{\bkt}[1]{\left\lbrack #1 \right\rbrack} \newcommand{\nor}[1]{\left\lVert #1 \right\rVert} \newcommand{\pth}[1]{\left( #1 \right)} \newcommand{\set}[1]{\left\lbrace #1 \right\rbrace} \newcommand{\abset}[1]{\abs{\set{#1}}} \newcommand{\ptset}[1]{\pth{\set{#1}}} %--Operators-- \newcommand{\grad}{\nabla} \newcommand{\La}{\Delta} \renewcommand{\div}{\operatorname{div}} \newcommand{\curl}{\operatorname{curl}} \newcommand{\tr}{\operatorname{tr}} \newcommand{\Hess}{\operatorname{Hess}} \newcommand{\mm}{\mathcal{M}} \newcommand{\inv}{^{-1}} \newcommand{\tensor}{\otimes} \newcommand{\cross}{\times} \newcommand{\weak}[1]{\xrightharpoonup{#1}} %--Notations-- \newcommand{\Vol}{\mathrm{Vol}} \newcommand{\loc}{\mathrm{loc}} \renewcommand{\dim}{\mathrm{dim}} \newcommand{\Span}{\mathrm{Span}} \newcommand{\diam}{\mathrm{diam}} \newcommand{\Id}{\mathrm{Id}} \newcommand{\Lip}{\mathrm{Lip}} \newcommand{\dist}{\mathrm{dist}} \newcommand{\inv}{^{-1}} %--Algebra-- \newcommand{\hfsq}[1]{\frac{\abs{#1} ^2}{2}} %--Sets-- \newcommand{\R}{\mathbb{R}} \newcommand{\RR}[1]{\R ^{#1}} \newcommand{\Rd}{\RR d} \newcommand{\Rn}{\RR n} \renewcommand{\S}{\mathbb{S}} \renewcommand{\Z}{\mathbb{Z}} \newcommand{\T}{\mathbb{T}} \newcommand{\ssubset}{\subset \subset} \newcommand{\spt}{\operatorname{spt}} \newcommand{\supp}{\operatorname{supp}} \newcommand{\diam}{\mathrm{diam}} \newcommand{\diag}{\operatorname{diag}} \renewcommand{\dim}{\mathrm{dim}} \newcommand{\dimH}{\mathrm{dim} _\mathcal{H}} \newcommand{\Sing}{\mathrm{Sing}} \newcommand{\Vol}{\mathrm{Vol}} \newcommand{\inds}[1]{\mathbf1_{\set{#1}}} \newcommand{\ind}[1]{\mathbf1_{#1}} %--Summations-- \newcommand{\SUM}[3]{\sum \cnt{#1}{#2}{#3}} \newcommand{\PROD}[3]{\prod \cnt{#1}{#2}{#3}} \newcommand{\cnt}[3]{ _{#1 = #2} ^{#3} } \newcommand{\seq}[2]{\set{#1 _{#2}} _{#2}} \newcommand{\seqi}[2]{\set{#1 _{#2}} \cnt{#2}1i} %--Superscript and Subscript \newcommand{\pp}[1]{^{(#1)}} \newcommand{\bp}[1]{_{(#1)}} \newcommand{\pps}[2][1]{^{#2 + #1}} \newcommand{\bps}[2][1]{_{#2 + #1}} \newcommand{\pms}[2][1]{^{#2 - #1}} \newcommand{\bms}[2][1]{_{#2 - #1}} %--Notations-- \newcommand{\loc}{\mathrm{loc}} \newcommand{\Span}{\operatorname{Span}} \newcommand{\argmin}{\operatorname{argmin}} \newcommand{\argmax}{\operatorname{argmax}} \newcommand{\osc}{\operatorname{osc}} \newcommand{\Id}{\mathrm{Id}} \newcommand{\Lip}{\operatorname{Lip}} \newcommand{\Leb}{\operatorname{Leb}} \newcommand{\PV}{\mathrm{P.V.}} \newcommand{\dist}{\operatorname{dist}} \newcommand{\at}[1]{\bigr\rvert _{#1}} \newcommand{\At}[1]{\biggr\rvert _{#1}} \newcommand{\half}{\frac12} %--Text-- \newcommand{\inn}{\text{ in }} \newcommand{\onn}{\text{ on }} \renewcommand{\ae}{\text{ a.e. }} \newcommand{\st}{\text{ s.t. }} \newcommand{\forr}{\text{ for }} \newcommand{\as}{\text{ as }} %--Differential-- \newcommand{\d}{\mathop{\kern0pt\mathrm{d}}\!{}} \newcommand{\dt}{\d t} \newcommand{\dx}{\d x} \newcommand{\dy}{\d y} \newcommand{\ptil}{\partial} \newcommand{\pt}{\ptil _t} \newcommand{\pfr}[2]{\frac{\partial #1}{\partial #2}} \newcommand{\dfr}[2]{\frac{\mathrm{d} #1}{\mathrm{d} #2}} \newcommand{\ddt}{\dfr{}t} \newcommand{\pthf}[2]{\pth{\frac{#1}{#2}}} %--Integral-- \newcommand{\intR}{\int _0 ^\infty} \newcommand{\intRn}{\int _{\Rn}} \newcommand{\intRd}{\int _{\Rd}} \newcommand{\fint}{-\!\!\!\!\!\!\int} \newcommand{\intset}[1]{\int _{\set{#1}}} %--Greek-- \newcommand{\e}{\varepsilon} \newcommand{\vp}{\varphi} %--Norm-- \newcommand{\nmL}[2]{\nor{#2} _{L ^#1}}

Research Blog

Boundary Harnack inequality

2024, April 27

Let e = (1, 0, \dots, 0). Consider a function u satisfying

\begin{cases} \La u = 0 & \inn \Omega \cap B _2 \\ u \ge 0 & \inn \Omega \cap B _2 \\ u = 0 & \onn \partial \Omega \cap B _2 \end{cases}

Theorem. If u, v satisfies above then

  • \max _{\Omega \cap B _1} u \le C u (e)
  • \sup _{\Omega \cap B _1} \frac uv \le C \frac{u(e)}{v(e)}
  • \left[\frac uv\right] _{C ^\alpha (\Omega \cap B _1)} \le C \frac{u (e)}{v (e)}

Recall that:

  1. Interior Harnack inequality: if x can be connected to e by a chain of N balls, then u(x) \le C ^N u (e).
  2. Therefore, if the boundary is Lipschitz (or more generally called nontangentially accessible domains) then c d (x) ^q u (e) \le u (x) \le C d (x) ^{-q} u (e). Here d (x) = d (x, \partial \Omega)
  3. Zero extension is subharmonic, for which weak Harnack inequality holds: if \La u \ge 0 in B _1, u \le 1 in B _1, \abs{\set{u \le 0} \cap B _1} \ge \mu > 0 then \max _{B _\frac12} u \le 1- \theta. \theta can becomes large (close to 1) if \mu is close to \abs{B _1}.
  4. Weak Harnack inequality means: if \La u \ge 0 in B _1, u (x) \ge 1 at some x \in B _\frac12, \abs{\set{u \le 0} \cap B _1} \ge \mu > 0, then \max _{B _\frac12} u \ge 1 + \theta.

Proof of 1.

If it fails, then for arbitrarily large A we can find x _1 \in \Omega \cap B _1 with u (x _1) > A.

  • x _1 must be \delta-close to \partial \Omega, otherwise we can use regular Harnack
  • by weak Harnack, there exists x _2 with d(x _1, x _2) \le 2 d (x _1) such that u (x _2) \ge (1 + \theta) u (x _1). Reason: take r = d (x _1). Then B _{2 r} (x _1) has a lot of zeros.
  • Then d (x _2) \le 3 d (x _1). We can find x _3 such that d (x _2, x _3) \le 2 d (x _2), and u (x _3) \ge u (x _2) (1 + \theta) ^2.

Then

C d (x _j) ^{-q} u (e) \ge u (x _j) \ge u (x _1) (1 + \theta) ^j \ge A (1 + \theta) ^j.

So d (x _j) ^q \le \pthf{C u (e)}A (1 + \theta) ^{-j} \to 0. Then u (x _j) \to 0, contradicting to u (x _j) \ge A.

Proof of 2.

First we show the lemma.

Lemma. There exists \delta, \varepsilon > 0. If

\begin{cases} \La u = 0 & \inn \Omega \cap B _1 \\ u \ge 1 & \inn \Omega _\delta \cap B _1 \\ u \ge -\varepsilon & \inn \Omega \cap B _1 \\ u = 0 & \onn \partial \Omega \cap B _1 \end{cases}

then u \ge 0 in \Omega \cap B _{\frac12}.

From Lemma to 2: WLOG u (e) = v (e) and u, v \le 1 in B _2. Apply to u - \varepsilon _0 v. Regular Harnack implies u - \varepsilon _0 v \ge c.

Lelemmamma. There exists \delta, a > 0. If

\begin{cases} \La u = 0 & \inn \Omega \cap B _1 \\ u \ge 1 & \inn \Omega _\delta \cap B _1 \\ u \ge -\varepsilon & \inn \Omega \cap B _1 \\ u = 0 & \onn \partial \Omega \cap B _1 \end{cases}

then u \ge a in \Omega _{\frac\delta2} \cap B _{\frac12}, and u \ge -a \varepsilon in \Omega \cap B _\frac12.

Lemma is an extreme version of Lelemmamma with a = 0.

From Lelemmamma to Lemma: Iterate it to get a cone of positivity, which is the union \Omega _{\delta / 2 ^k} \cap B _{2 ^{-k}}.

Proof of Lelemmamma. Let v = u + \varepsilon _0. v \ge 0 and v \ge 1 in \Omega _\delta \cap B _\frac12. So using Harnack once we can get v \ge 2 a in \Omega _{\delta/2} \cap B _\frac12. Apply weak Harnack inequality to u _- = \max (0, -u). Recall the \theta can be large: \max _{B _\frac12} u _- \le C \nor{u _-} _{L ^1 (B _1)}. Since u _- is only nonzero in \Omega \setminus \Omega _\delta, it is a thin strip with very little measure.

Proof of 3.

Harnack inequality implies oscillation decay, which implies Hölder.

By part 2, \frac1A \le \frac uv \le A in B _1. So

\frac1A u \le v \le A u.

Now A u - v and A v - u are both harmonic and nonnegative in B _1 and vanish on \partial \Omega, so by boundary Harnack

\frac1C \le \frac{A u - v}{A v - u} \le C \qquad \text{ in } B _{\frac12} \cap \Omega.

So

\frac{C + A}{C + \frac1A} \cdot \frac1A u \le \frac{A + C}{A C + 1} u \le v \le \frac{A C + 1}{A + C} u = \frac{C + \frac1A}{C + A} \cdot A u.

We have oscillation decay!

Recent Posts

22 Jan 2021

De Giorgi

14 Apr 2020

Lorentz Space