3
$\begingroup$

Consider the singular porous medium equation $$u_t - \Delta (|u|^{m-1}u) = 0$$ $$u(0)=u_0$$ given $u_0$ bounded, where $m \in (0,1)$.

When posed on $\mathbb{R}^n$, it is well known that mass is conserved when $m \in (m_c, 1)$ and mass is lost when $m \in (0,m_c)$, where $m_c$ is a number that depends on the dimension $n$.

When we pose this equation on a $n-$dimensional compact closed manifold, does this still hold true? Do we still get a range for which mass is lost, i.e., $$\int_M u(t,x)dx \neq \int_M u_0?$$

I've tried to find papers but there is not much on manifolds.


The solution satisfies $u \in C([0,T);L^2(M))$ with $|u|^{m-1}u \in L^2(0,T;H^1(M))$ and $$-\int_M u_0\phi(0) + \int_0^T \int_M (-u\phi_t + \nabla |u|^{m-1}u \nabla \phi) = 0$$ for all test functions vanishing at $t=T$. So it is not clear how to proceed?

$\endgroup$
5
  • $\begingroup$ Does the manifold have boundary? What about the boundary conditions? No hope of mass conservation if you impose Dirichlet b.c., I suppose. Anyway, a starting point might be §11.5 in J.L. Vázquez' The porous medium equation, Oxford UP 2007 $\endgroup$ Commented Oct 19, 2015 at 11:05
  • $\begingroup$ @DelioMugnolo It is a closed manifold so no boundary. I did try the book by Vazquez, however, there is no discussionn of the singular case there. $\endgroup$
    – AlC
    Commented Oct 19, 2015 at 12:20
  • $\begingroup$ @AIC Sorry, I don't get it. What do you call "singular"? The whole book is about the PME! The quoted theorem deals with an even more general setting. (Sorry, I oversaw "closed") $\endgroup$ Commented Oct 19, 2015 at 13:09
  • $\begingroup$ @DelioMugnolo Singular means the exponent in the nonlineary, which I wrote as $m$ lies in $(0,1)$. The "normal" or "usual" PME has exponent $m \in (1,\infty)$. The PME book focuses only on the $m \in (1,\infty)$ case (see the introduction). $\endgroup$
    – AlC
    Commented Oct 19, 2015 at 13:14
  • $\begingroup$ @AIC: I see. But take a look e.g. at sciencedirect.com/science/article/pii/S0022123605001291 (seemingly only in the range $m>1$) Most results depend on energy estimates (which, given smoothness of the manifolds, will only depend on classical Green formulae or rather on suitably weak definitions of the operators) and suitable Sobolev inequalities. A similar strategy may work in your case, too, and it seems to me that the real challenge is to deduce log-Sobolev inequalities in the manifold case. $\endgroup$ Commented Oct 19, 2015 at 13:53

0

You must log in to answer this question.

Browse other questions tagged .