The regularity problem for generalized harmonic maps into homogeneous spaces (Q1347264)

Let \(M\) be a Riemannian surface and consider a map \(u\) from \(M\) into the sphere \(S^ n\) (or into a Riemannian manifold on which a Lie group acts transitively by isometries). For \(u\) of Sobolev class \(H^{1,2}(M,S^ n)\) the harmonic map equation \[ -\Delta u = u| \nabla u |^ 2 \tag{1} \] is easily seen to be equivalent to the system \[ \text{div}(u^ i \nabla u^ j - u^ j \nabla u^ i) = 0, \quad i,j = 1,\dots, n+1 \tag{2} \] and (2) makes sense for functions \(u\) in the space \(H^{1,1}(M,S^ n)\). The paper now gives a careful analysis of the delicate question under which assumptions weak solutions of equation (2) are regular up to a possible singular set of small size. First of all, \(\nabla u \in L^ p\) for some \(p \in [1,2)\) is not sufficient which can be seen by the authors' counterexamples consisting of everywhere discontinuous solutions. In further sections it turns out that the natural condition to demand on \(\nabla u\) is that it belongs to the Lorentz-space \(L(2,\infty)\). In this case one can prove an \(\mathcal E\)- regularity theorem saying that \(u\) is smooth in a neighborhood \(V\) of some point \(x_ 0\) provided the \(L(2,\infty)\)-norm of \(u\) with respect to \(V\) is sufficiently small. If in addition \(\nabla u\) is in \(L(2,q)\) for some \(q \in (2,\infty)\) then \(u\) has a finite number of singularities in \(M\).