1
$\begingroup$

I am reading the paper Large global solutions for energy supercritical nonlinear wave equations on $\mathbb{R}^{3+1}$ by Krieger and Schlag and am confused by one of their steps.

For context, $v(t,r)$ is radial and solves a septic nonlinear wave equation $$-\partial_{tt}v + \Delta v = F(v),$$ on $(t,x)\in[T, T_1]\times \mathbb{R}^3$ where $F(v) = G(v) + g(r,t)$, where $G(0)=0$ and $g(r,t)$ is supported on $\{r=|x|<t+C\}$. We know that at time $T$, $v$ is supported in the ball $\{r=|x|\le T+C\}$ with $C>0$. The authors then claim, on page 34, that by Huygens' principle, it follows $v(r,t)$ is supported in the ball $\{r=|x|<t+C\}$ for each fixed $t\ge T$

However, I believe this simply follows from the finite propagation speed property of the wave equation. Is this correct, or is the Huygens' principle genuinely needed here?

$\endgroup$
1
  • $\begingroup$ I apologize, I have fixed the title. $\endgroup$
    – Dispersion
    Commented 14 hours ago

2 Answers 2

3
$\begingroup$

As a remark, invoquing the finite propagation speed property for the wave equation is hand waving. How do one make this statement a rigorous mathematical concept ?

For a general hyperbolic PDE (or system of PDEs), you can make a definition, by looking at the velocity of planar waves. But then you have to transform this definition into a theorem and this requires pseudo-differential calculus.

For the wave equation, or more generally for so-called symmetric hyperbolic (systems of) PDEs, a much simpler proof occurs, which consists in integrating the energy equation $$\partial_t\frac12(v_t^2+|\nabla v|^2)-{\rm div}(v_t\nabla v)=-v_tF(v)$$ over the characteristic cone $t\in(0,T)$, $|x|<t+c$.

Mind that you need of course $F(0)=0$.

$\endgroup$
1
  • $\begingroup$ To be clear then, Huygen's principle is not needed here and thus this type of result holds in even dimensions too. $\endgroup$
    – Dispersion
    Commented 49 mins ago
0
$\begingroup$

Q: Huygens' principle or finite speed of propagation?

It's the same thing, Huygens principle is a statement of causality, which means finite speed of wave front propagation.

Without Huygens principle you might imagine that the right-hand-side of the wave equation is a source term that allows a nonzero $v(x,t)$ to appear at $|x|>t+C$, without any relationship to the wavefronts at earlier times. The notion that a nonzero $v$ at some later time is causally related to a wavefront at an earlier time is the content of Huygens principle.

$\endgroup$
4
  • $\begingroup$ According to en.wikipedia.org/wiki/Huygens%E2%80%93Fresnel_principle : In 1900, Jacques Hadamard observed that Huygens' principle was broken when the number of spatial dimensions is even. Does this mean that for even number of spatial dimensions the speed of wave front propagation is infinite ? $\endgroup$
    – jjcale
    Commented 2 hours ago
  • $\begingroup$ Indeed, this will not work in two spatial dimensions (the question is for $\mathbb{R}^3$). $\endgroup$ Commented 2 hours ago
  • $\begingroup$ It seems that the answer by Denis Serre works in any number of dimensions, whereas your answer implies that such a result only works in odd dimensions. What is the discrepancy? $\endgroup$
    – Dispersion
    Commented 1 hour ago
  • $\begingroup$ I added more details on the source term $v$ in my question: it is the sum of $G(v)$, where $G(0)=0$ and a term $g(r,t)$ supported on $\{|x|<t+C\}$ for each fixed $t$. Does this form preclude the type of source term you mention above? $\endgroup$
    – Dispersion
    Commented 54 mins ago

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .