I am reading a paper and have trouble understanding the following step on page 581:

Assume we know that $w: \mathbb{R}^n \to \mathbb{R}$ is smooth and bounded. Furthermore $\Vert w \Vert_{L^\infty}>1$ and $w(x) \to 1$ whenever $\vert x \vert \to \infty.$ Finally the following inequality holds \begin{equation} \Delta w + 2 w(1+\frac{c^2}{4}-w) \geq 0. \end{equation}The author then applys the weak maximum principle to obtain \begin{equation} w \leq 1+\frac{c^2}{4}. \end{equation}

While I found several versions of a "weak maximum principle" (e.g. here), I can't figure out how to apply them to this situation. They all either assume that the domain is bounded and/or there are only derivatives of $w$ (but not $w$ itself) in the PDE.

So my question is:

Can someone give me a hint which version of the weak maximum principle is applicable in this situation and how to apply it?

All help is much appreciated!

  • $\begingroup$ Hint: you can't have strictly positive Laplacian at a point of local maximum. $\endgroup$ – ˈjuː.zɚ79365 Jun 7 '13 at 8:40
  • $\begingroup$ Aside: not everyone has access to MathSciNet. A direct link to a paper (even if it's behind paywall) is better than a link to a reference to a paper which is itself behind a paywall. The first kind of link at least tells the reader what the paper is, so they can try to find it elsewhere. $\endgroup$ – ˈjuː.zɚ79365 Jun 7 '13 at 10:39
  • 1
    $\begingroup$ @user79365: the better solution actually is for the OP to copy the link from the "Make link" button on MathSciNet. That link automatically resolves to the MathSciNet review if a subscription is detected, or also to a bibliographic entry that links to the original paper if not. Compare this link with the one provided by the OP. See also this FAQ item. $\endgroup$ – Willie Wong Jun 7 '13 at 11:02

Suppose $w(x)>1+c^2/4$ for some $x$. Since the set $\{x: w(x)\ge 1+c^2/4 \}$ is bounded and closed, it is compact. Let $x_0$ be a point of this set where $w$ attains its maximum. Since it is a global maximum of $w$, we have $w(x_0)>1+c^2/4$.

On the other hand, $$\Delta w(x_0)\ge 2w(w-1-c^2/4)>0 \tag1$$ Therefore there is $j\in \{1,\dots,n\}$ such that $$\frac{\partial^2}{\partial x_j^2}w(x_0)>0 \tag2$$ Inequality (2) is incompatible with $w$ (considered as a function of $j$th variable only) having a maximum at $x_0$.


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.