MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top


I want to prove the existence of a weak solution to:

Find $u:S^1 \times [0,T) \to \mathbb{R}$ such that $$\frac{\partial u}{\partial t} = u^{n_1}\frac{\partial^2 u}{\partial x^2} + u^{n_2}$$ with $u(x, 0) = u_0(x)$ where $u_0 \in C^{1+\alpha}(S^1)$ (and is non-negative) and $n_1$ and $n_2$ are fixed integers (eg. $n_1 = 2$, $n_2 = 3$, which I fix for now).

Also, $u$ should lie in the space $C^{2+\alpha, 1+\alpha}(S^1 \times [0, T-\epsilon])$ for $\epsilon > 0$.

How to prove short-time existence for this PDE? Every book (Michael Taylor's Nonlinear.., Ladyzhenskaya) I look at has existence results that require strong parabolicity (i.e., require $u^2\eta^2 \geq C\eta^2$ for all non-zero $\eta$, which I cannot say I have).

A reference to where this kind of PDE is proven would be appreciated or a detail of how to prove such PDEs.

PS: This question is on Math.SE for a while but it has not received much attention, so I hope it is OK to post it here as otherwise I would be very stuck. Thanks.

share|cite|improve this question
This is weakly parabolic, so the trick is to pretend that it's hyperbolic rather than parabolic. Look up how to prove local time existence of a nonlinear wave equation using energy estimates. The idea is to prove $L^2$ energy estimates. The parabolic term contributes a negative term to the evolution inequality for the $L^2$ energy, which can be tossed away. – Deane Yang Jun 12 '12 at 15:52
up vote 3 down vote accepted

Use the references on strongly parabolic PDE's to show that for each $\epsilon > 0$, you can solve $$ \partial_t u_\epsilon = (\epsilon + |u_\epsilon|^{n_1})\partial_x^2u_\epsilon + |u_\epsilon|^{n_2}. $$ Using energy estimates, get estimates for the time of existence and the $L^2$ Sobolev norms of $u$ that are independent of $\epsilon$. Let $\epsilon \rightarrow 0$.

By the maximum principle, $u$ will be strictly positive for small positive time. Therefore, the equation is strictly parabolic for small positive time. This will imply that $u$ is smooth for small positive time.

share|cite|improve this answer
Thanks a lot. Unfortunately I can't even show that that strongly parabolic PDE can be solved. Michael Taylor's theorem requires $u_0 \in H^2$ while I only have only have $C^1$. The same is true for existence results in Ladyzhenskaya (in addition, the results require a lot of requisities.. mainly horrible inequalities). Do you know of any better references? Also, if I may ask, from where did you learn this "trick?" It's difficult to find much literature on this topic. – user24394 Jun 13 '12 at 20:22
A $C^1$ function on a compact manifold is in $H^2$ (which I presume means the first derivative is in $L^2$). – Deane Yang Jun 13 '12 at 20:49
The model case is not a strongly parabolic equation. The model case is, say, $\partial_t u = a(x, u)\partial_x u + b(x,u)$. – Deane Yang Jun 13 '12 at 20:51
I apologise for any misunderstanding. By $H^2$ I refer to the $L^2$ functions with second derivatives also (and not just the first) in $L^2$. So knowing $u_0 \in C^1$ on a compact set only means $u_0 \in H^1$ (first derivative in $L^2$). I was just referring to the PDE you suggested when I said "strongly parabolic". – user24394 Jun 13 '12 at 21:40
I accepted your answer. Still not sure how to do this question but I'll read up on your comment on the original post. – user24394 Jun 14 '12 at 11:53

Some years ago we wrote a paper on degenerate parabolic equations (and, actually, systems) which you might find of help. We worked in Sobolev classes there, but I think you can adapt the techniques there to work under your assumptions. The paper is here.

share|cite|improve this answer
Thanks for the reply. I will take a look at your paper more closely but on the first glance it appears to use too many things I don't know anything about. – user24394 Jun 13 '12 at 20:23

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.