MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

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


Let $(M,g)$ be a smooth compact riemannian manifold without boundary. Let $L$ be a linear elliptic operator on $M$ of order $2k$ with smooth coefficients. Suppose i have $u\in W^{2k,2}(M)$ and $f\in C^{0,\alpha}(M)$ such that $$L(u)=f$$ Do i have Schauder estimates of type

$$\left\|u\right\|_{C^{2k,\alpha}\left(M\right)}\leq C\left(L\right)\left\|f\right\|_{C^{0,\alpha}\left( M \right)}$$

I can assume also that $L$ (it would be better for $L$ of general type) is self adjoint and $u$ is $L^2$-orthogonal to $\ker\left(L \right)$.

If yes is there a reference for this kind of result?

Thank you in advance.

share|cite|improve this question
Dear Italo I have tried to edit your question, it is ok for you? – Giuseppe Tortorella Jan 7 '13 at 15:23
Perfect! What was the problem? – Italo Jan 7 '13 at 15:24
When MathJax seems having problems, the basic solution is to enclose between backtips any TeX code containing underscores or asterisks. – Giuseppe Tortorella Jan 7 '13 at 15:38
Thank you very much! – Italo Jan 7 '13 at 15:42
up vote 4 down vote accepted

The result is true with some caveats. Under your assumptions we have the following results.

1. If $u\in W^{2k,2}(M)$ and $Lu\in C^{j,\alpha}(M)$, then $u\in C^{2k+j,\alpha}(M)$.

2. There exists $C>0$ depending only on $M$, $L$, $j$, and $\alpha$ such that, for any $u\in C^{2k+j,\alpha}(M)$ we have

$$\Vert u\Vert_{C^{2k+j,\alpha}} \leq C\Bigl(\; \Vert Lu\Vert_{C^{j,\alpha}}+ \Vert u\Vert\_{C^{0,\alpha}}\;\Bigr) $$

3. There exists $C>0$ depending only on $M$, $L$, $j$, and $\alpha$ such that, for any $u\in C^{2k+j,\alpha}(M)\cap (\ker L)^\perp $ (the $\perp$ refers to the $L^2$-inner product) we have

$$\Vert u\Vert_{C^{2k+j,\alpha}} \leq C \Vert Lu\Vert_{C^{j,\alpha}} $$

For proofs and more details see Chapter 10 of these notes and the references therein.

share|cite|improve this answer
+1 Great answer! – András Bátkai Jan 7 '13 at 16:29
Thank you very much Liviu! I'm reading your notes, once i work out the details i'll accept the answer! If i need, can i ask you some details? – Italo Jan 8 '13 at 15:33
Sure, you can ask me any questions. – Liviu Nicolaescu Jan 8 '13 at 16:13
Dear Liviu i have some questions: 1) how do you get the local $C^{2k,\alpha}$-regularity of $u$ if you only know that $u\in W^{2k,2}(M)$? I had this idea but it brought me nowhere: i can't use the elliptic estimates (thm 10.3.1 part b) of your notes as they are because you assume the $C^{k+j}$-regularity for $u$ so i mollify $u$ with a smoothing kernel $\rho_{\frac{1}{n}}$ obtaining a sequence of smooth functions $u_n=u\star \rho_{\frac{1}{n}}$ and i use cutoffs to come to the euclidean case. – Italo Jan 9 '13 at 15:46
So i get elliptic estimates of type ` $$\left\|u_n\right\|_{C^{2k,\alpha}(M)}\leq C(L,M,g)\left(\left\|L(u_n)\right\|_{C^{0,\alpha}(M)} +\left\|u_n \right\|_{C^{0,\alpha}(M)} \right)$$ ` but now i can't get the $C^{0,\alpha}(M)$-convergence for $u_n$. – Italo Jan 9 '13 at 15:50

This is probably more of a comment: Section 3.2 in Lunardi's book contains a broad overview on higher order parabolic problems with lots of references, including the most important Hölder estimates.

On manifolds, you should be able to extend these results using a finite number of coordinate charts (by compactness) as in these notes. But, this is not a reference...

share|cite|improve this answer

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.