Let $D \subset \Bbb C$ be a bounded domain, and let $h$ be a continuous function on $\partial D$.

We remove one point from the domain to obtain $D^*=D \setminus \{z_0\}$, and extend $h$ to a function $h^*$ on $\partial D^*$ by choosing some value $h^*(z_0)$.

The Perron solution on $D$ is the harmonic function defined by $\tilde h = \sup \mathcal{F}_h$, where $$\mathcal{F}_h = \{ u \in \mathcal{SH}(D) : \limsup_{z \to \zeta} u(z) \leq h(\zeta) \}, $$

and we similarly define the Perron solution for $h^*$.


In a past post, it is proved that $\tilde h = \tilde h_0$ when we remove the center of a disk. Does it also hold true when removing a point from an arbitrary domain? In other words,

I want to know whether the Perron solution of the Dirichlet problem on $D^*$ is always equal to the Perron solution on $D$, regardless of the value of $h(z_0)$.

The insight of the proof in that linked post was to provide a subharmonic function $\lambda$ that vanishes on $\partial D$ but is equal to $-\infty$ on $z_0$. I don't know if such a function always exists, when the domain is arbitrary. Is the result true nevertheless?

Added: The proof of Daniel Fischer for a disk

Copied here for ease of reference and to show where it seems not to work.

Assume $D=\{ |z-z_0| < r \}$. Letting $\lambda(z) = \ln \frac{|z-z_0|}{r}$, we will need the following three properties:

  1. $\lambda$ is subharmonic on $D$
  2. $\lambda(z) \to 0$ on $\partial D$
  3. $\lambda(z_0) = -\infty$

First direction. Let $c>0$. The function $\tilde h+c\lambda \in \mathcal F_{h^*}$ (subharmonic, agrees with $h$ on all limits on $\partial D$, is $\infty$ on $z_0$). Letting $c \to 0$ gives $\tilde h \leq \tilde h^*$.

Second direction.

Let $c>0$. By continuity of $\lambda$ at $z_0$ there is a neighborhood of $z_0$ where $\widetilde {h^*} \leq \tilde h-c\lambda$.

Since $D$ is a soluble domain for the Dirichlet problem, we have on $\partial D$ that $\limsup \widetilde {h^*} \leq h^*=h=\tilde h \leq \tilde h-c\lambda$, so $\tilde h-c\lambda$ is a superharmonic function that has boundary limits greater than the harmonic function $\widetilde{h^*}$ on all $\partial D^*$. By the minimum principle, we then have $\tilde h-c\lambda \geq \tilde h^*$. Tending again $c \to 0$ finishes the proof.

  • $\begingroup$ You want to know if it is possible to adapt DanielFischer's solution for $D = \{ 0 < |z| < 1\}$ but you don't copy and explain this solution, not even the notations. $\endgroup$ – reuns Aug 29 '17 at 17:18
  • 1
    $\begingroup$ I see, good idea. Edited this in. $\endgroup$ – Pachirisu Aug 29 '17 at 19:47

Note that property 2 is not necessary. Instead you can have $\limsup\lambda\leq0$ at the boundary. This generalizes the proof to arbitrary bounded domain.

  • $\begingroup$ Hi, thanks for answering. However, I suspect the proof for the disk uses the solvability of the Dirichlet problem, otherwise the inequality $\tilde h \leq \tilde h -c\lambda$ says little about $\widetilde {h^*}$. I edited the presentation of the proof in the post to explain what I mean by that. $\endgroup$ – Pachirisu Sep 9 '17 at 22:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.