Harmonic type approximation lemmas (Q1012188)

This survey article summarizes the results on techniques related to harmonic approximation, which the two authors, among others, have pushed forward to apply to a vast variety of PDE problems. The classical harmonic approximation lemma, first used by De Giorgi and formulated by L. Simon in its current form, says that for every \(\varepsilon>0\) there is \(\delta=\delta(\varepsilon,n,N) \in(0,1]\), such that the following holds: Whenever \(u\in W^{1,2}(B^n,R^N)\) satisfying \(\int_{B^n}|Du|^2\,dx\leq1\) is approximately harmonic in the sense that \[ \Big|\int_{B^n}Du\cdot D\varphi\,dx\Big|\leq\delta\sup_{B^n}|D\varphi| \] holds for every \(\varphi\in C^\infty_0(B^n,R^N)\), then there exists a harmonic map \(h\in W^{1,2}(B^n,R^N)\), i.e.\ \(\Delta h=0\) in \(B^n\), such that \(\int_{B^n}|Dh|^2\,dx\leq1\) and \[ \int_{B^n}|h-u|^2\,dx\leq\varepsilon^2. \] This lemma can be modified in many ways, all of which are discussed in the paper: {\parindent6mm \begin{itemize}\item[(1)] Harmonic maps can be replaced by \(A\)-harmonic maps solving an elliptic system with constant coefficients. This gives the \textit{\(A\)-harmonic approximation lemma}. It can be used to simplify many things in elliptic regularity therory, as first shown by Duzaar, Steffen and Grotowski. \item[(2)] A similar lemma can be proven in a \(W^{1,p}\)-setting, with \(\Delta\) replaced by the \(p\)-Laplacian \(div(|Du|^{p-2}Du)\). There is also an \(A\)-\(p\)-harmonic version of the lemma, which is proven here for the first time. \item[(3)] Even a parabolic version holds, proven by Duzaar and Mingione (like the previous one). Here suitable functions are approximated by solutions to heat flows, which is stated in the \textit{caloric approximation lemma.} \item[(4)] A \(p\)-caloric version of the lemma has not been proven. A conjectured version is formulated in the article. \item[(5)] Several variants, e.g.\ higher order versions, are mentioned briefly; the reader is referred to the carefully collected list of references. \end{itemize}} A powerful gain of these lemmas for regularity theory is that they often allow to avoid technical questions concerning higher integrability and the Gehring Lemma. This may help organizing more direct and transparent blowup proofs of known results, and it has also inspired new regularity theorems. More than other survey articles, this one gives proofs of the lemmas in quite some detail. Their applications are discussed with many references, and in the case of degenerate problems even a new partial regularity theorem is proven, applying the new lemma mentioned above. The mentioned features make the paper an excellent introduction to a recent and active area of PDE theory.