I asked this question on math.stackexchange.com, however I didn't get any answers so I'll try it here.

Let $M$ be a compact manifold without boundary. Consider $Lu:=\partial_tu-\Delta u$. Let $f\in C^{0,0,\alpha}((0,T)\times M,\mathbb{R})$, $u_0\in C^{2,\alpha}(M,\mathbb{R})$ and let $u\in C^{1,2,\alpha}((0,T)\times M,\mathbb{R})$ be the unique solution of

$$ \begin{cases} Lu=f \\ u(t,.)\to u_0(.) & \text{as $t\to 0$} \end{cases} $$

Here $C^{k,l,\alpha}$, $C^{k,\alpha}$ are the appropriate (parabolic) Hölder spaces.

Then we have

$$||u||_{C^{1,2,\alpha}((0,T)\times M,\mathbb{R})}\le C(||f||_{C^{0,0,\alpha}((0,T)\times M,\mathbb{R})}+||u_0||_{C^{2,\alpha}(M,\mathbb{R})})$$

For some $C>0$ independent of $f,$ $u$ and $u_0$.

Question: Is it possible to choose $C$ independent of $T$? If C depends on $T$, i.e. $C=C(T)$, is it possible to choose $C(T)$ in a way that $C(T)$ is bounded for $T\to 0$?

If one replaces $M$ by a suitable domain in $\mathbb{R}^n$ the answer of my question seems to be "Yes". However, I need the result for compact manifolds and was wondering if there occur any problems.

  • $\begingroup$ What does mean $\mathbb R$ in ${C^{2,\alpha}(M,\mathbb{R})}$? $\endgroup$ – Andrew Nov 27 '15 at 11:04
  • $\begingroup$ It just refers to funtions $M\to\mathbb{R}$, i.e. the functions have values in $\mathbb{R}$ (and not e.g. another manifold $N$) $\endgroup$ – alerte Nov 27 '15 at 11:08

The constant $C(T)$ depends on $T$, consider $f=1$, $u_0=0$ and the solution $u=t$. But it is bounded for $T\to0$. The rhs $f\in C^{0,\alpha}([0,T]\times M)$ can be continued to $\tilde f\in C^{0,0,\alpha}([0,1]\times M)$, $||\tilde f||_{C^{0,0,\alpha}((0,1)\times M)}= |f||_{C^{0,0,\alpha}((0,T)\times M)}$. Denote $\tilde u$ the solution with rhs $\tilde f$. Then $$ |u||_{C^{2,1,\alpha}((0,T)\times M)}\le |\tilde u||_{C^{2,1,\alpha}((0,1)\times M)}\le $$ $$ \le C(1)(||\tilde f||_{C^{0,0,\alpha}((0,1)\times M)}+||u_0||_{C^{2,\alpha}(M,\mathbb{R})}))\le C(1) (|f||_{C^{0,0,\alpha}((0,T)\times M)}+||u_0||_{C^{2,\alpha}(M,\mathbb{R})})). $$

| cite | improve this answer | |
  • $\begingroup$ Thank you for your answer. I don't understand the last inequality. Why is $||\tilde{f}||_{C^{0,0,\alpha}((0,1)\times M}\le ||\tilde{f}||_{C^{0,0,\alpha}((0,T)\times M)}$ for $T\le 1$ $\endgroup$ – alerte Nov 27 '15 at 11:43
  • $\begingroup$ Consider the following continuation of $f$ to $[0,2T]$, symmetry with respect to the plane $t=T$: $\tilde f(t,x)=f(2T-t,x)$, $T<t\le 2T$. Then $||\tilde f||_{C^{0,0,\alpha}((0,2T)\times M)}=|f||_{C^{0,0,\alpha}((0,T)\times M)}$. То be repeated until $2^nT\ge1$. $\endgroup$ – Andrew Nov 27 '15 at 14:04
  • $\begingroup$ I apologize for asking this seemingly unrelated question here in an old answer of you, I hope you wouldn't mind. Although the original question asked about estimates, I would like to know how one proves an existence result for this equation with the same assumptions on regularity of $u_0$ and $f$? I am not familiar with compact manifold without boundary. $\endgroup$ – BigbearZzz Mar 29 '18 at 8:55
  • $\begingroup$ @BigbearZzz Existence follows from the existence of the fundamental solution: the solution of the Cauchy problem can be written as a sum of two potentials, one for $f$ and another for $u_0$. $\endgroup$ – Andrew Mar 29 '18 at 13:45
  • $\begingroup$ So the fundamental solution for compact manifold without boundary always exists? Is it a well-know result that I can find in a geometric-related PDE book? $\endgroup$ – BigbearZzz Mar 29 '18 at 14:05

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.