A problem about details in the proof of the theorem 6.5.3( principal eigenvalue for nonsymmetric elliptic operators) in Evans' PDE

104 Views Asked by At

We consider in this section the boundary-value problem

$$\left\{\begin{aligned} L w &=\lambda w \quad \text { in } U \ \ \ \ \ \ (1)\\ w &=0 \quad \text { on } \partial U \end{aligned}\right.$$ where $U$ is open and bounded, $$L u=-\sum_{i, j=1}^{n} a^{i j} u_{x_{i} x_{j}}+\sum_{i=1}^{n} b^{i} u_{x_{i}}+c u .$$

Let us for simplicity assume that $a^{i j}, b^{i}, c \in C^{\infty}(\bar{U})$ , that $U$ is open, bounded and connected, and that $\partial U$ is smooth. We suppose also $a^{i j}=a^{j i}(i, j= 1, \ldots, n)$ and

$$c \geq 0 \quad \text { in } U$$ THEOREM 3 (Principal eigenvalue for nonsymmetric elliptic operators).
(i) There exists a real eigenvalue $\lambda_{1}$ for the operator $L$ , taken with zero boundary conditions, such that if $\lambda \in \mathbb{C}$ is any other eigenvalue, we have

$$\operatorname{Re}(\lambda) \geq \lambda_{1}$$

(ii) There exists a corresponding eigenfunction $w_{1}$ , which is positive within $U$ .
(iii) The eigenvalue $\lambda_{1}$ is simple; that is, if $u$ is any solution of (1), then $u$ is a multiple of $w_{1}$ .

Proof . 1. Choose $m=\left[\frac{n}{2}\right]+3$ and consider the Banach space $X= H^{m}(U) \cap H_{0}^{1}(U)$ . According to Sobolev Embedding Theorem, $X \subset C^{2}(\bar{U})$ . We define the linear, compact operator $A: X \rightarrow X$ by setting $A f=u$ , where $u$ is the unique solution of

$$\left\{\begin{aligned} L u=f & \text { in } U \\ u=0 & \text { on } \partial U \end{aligned}\right.$$

Next define the cone

$$C=\{u \in X \mid u \geq 0 \text { in } U\} .$$

According to the maximum principle, $A: C \rightarrow C$

  1. Hereafter fix any function $w \in C$, $w \not \equiv 0$ . Employing the strong maximum principle and Hopf's Lemma, we deduce $$v>0 \ \ in \ \ U, \frac{\partial v}{\partial \nu}<0 \ \ on\ \ \partial U \ \ \ (2)$$ for $v=A(w)$ . Remember that $w=0$ on $\partial U$ . So in view of (2) there exists a constant $\mu>0$ so that

$$\mu v \geq w \quad \text { in } U$$

My problem is: why such constant $\mu>0$ exists? I know we only need to consider the point near the boundary of $U$ and notice that $w=0$ on $\partial U$,we may need to apply Tailor's formula, and I believe the conclusion is true, but I just can't prove it clearly. Can someone help me? I would be appreciated if you could help me.

1

There are 1 best solutions below

2
On

Since $v > 0$ (strictly positive) you can just define the function $$ g = \frac{w}{v}, $$ and then take $\mu = \sup_U g$ so that $\mu \geq g$ a.e. in $U$ and therefore $$ \mu \geq \frac{w}{v} \implies \mu v \geq w. $$ Remark that since $w \in C$ and $v > 0$, it holds $g \geq 0$ and therefore $\mu \geq 0$. We can see that $g = 0$ a.e. only if $w=0$ a.e., but in this case since $v > 0$ we can take any $\mu > 0$ and $\mu v > w$ would hold.