2

If a sequence of harmonic functions $u_n \rightharpoonup u$ (converges weakly) in $L^2(\Omega)$, then $\Delta u = 0$ in $\Omega$.

Recall a sequence of functions $f_n$ defined on an open set $\Omega$ is said to converge weakly in $L^2(\Omega)$ to a function $f$ if: $$\int f_n(x)\,g(x)\,dx \to \int f(x)\,g(x)\,dx \hspace{1cm} \forall g \in L^2 (\Omega).$$

My first thought is just to pass the limit using the Mean Value Theorem since if MVT holds that implies $u$ is harmonic. However, I don't think that works with 'weak convergence' with my definition above.

Dragonite
  • 2,388
  • 1
    For $y\in \Omega$, consider the characteristic functions of balls and of spherical shells with centre $y$. – Daniel Fischer Oct 07 '17 at 15:06
  • Hello! I’ve made a slight edit to the MathJax in your post: I’ve introduced a small space \, between the functions of more than one character and before the differentials, as recommended in the MathJax tutorial. If you find this too pedantic, feel free to roll it back! $\ddot\smile$ – gen-ℤ ready to perish Oct 07 '17 at 16:23

2 Answers2

2

$$u_n \rightharpoonup u\implies \int_\Omega u_n(x)\phi(x)dx \to \int_\Omega u(x)\phi(x)dx \hspace{1cm} \forall \phi \in C^\infty_0 (\Omega).$$

in particular $$ \phi \in C^\infty_0 \implies \Delta \phi \in C^\infty_0 $$

Hence $$\int_\Omega u_n(x)\Delta\phi(x)dx \to \int_\Omega u(x)\Delta\phi(x)dx\Longleftrightarrow \int_\Omega (u_n(x)-u(x))\Delta\phi(x)dx \to 0\hspace{1cm} \forall \phi \in C^\infty_0 (\Omega).$$ But using integration by part or By derivative in the sense of distributions we know that $$\int_\Omega (u_n(x)-u(x))\Delta\phi(x)dx =\int_\Omega \Delta(u_n(x)-u(x))\phi(x)dx~~\\ =\int_\Omega -\Delta u(x)\phi(x)dx ~~\to 0~~~\forall~~ \phi \in C^\infty_0 (\Omega).$$ Since $$\Delta u_n =0$$

Hence, $$\int_\Omega -\Delta u(x)\phi(x)dx =0~~\forall~~ \phi \in C^\infty_0 (\Omega)\Longleftrightarrow \Delta u = 0~~\text{a.e on }~~\Omega$$

Guy Fsone
  • 23,903
1

It uses the Wely's Lemma: https://en.wikipedia.org/wiki/Weyl%27s_lemma_(Laplace_equation) as follows:

If $u\in L^1_{loc}(\Omega)$ satisfies $\int_\Omega u \Delta \varphi dx=0$ for all $\varphi\in C_c^\infty(\Omega)$, then $u$ is smooth (possibly after redefining on a set of zero measure) and harmonic.

It suffices to show that $\int_\Omega u \Delta \varphi =0$. For any $\varphi$, we have $$ \int_\Omega u_n\Delta \varphi dx=0, $$ by integration and the property that $u_n$ is harmonic. Then by limiting above and using the weak convergence, we have $$ \int_\Omega u \Delta \varphi dx= \lim_{n\to\infty}\int_\Omega u_n\Delta \varphi dx=0. $$ For the proof of Wely's Lemma, please see:

enter image description here

enter image description here

Ice sea
  • 1,222