“…As well in the classical situation [Riv07], as also in the case of fractional harmonic maps, the argument relies on transforming the equation with an orthogonal matrix P (in a similar way as Hèlein's moving frame technique, cf. [Sch10a]). That is, one computes the respective equation P ∇u instead of ∇u, or P ∆ n 4 u instead of ∆ n 4 u and obtains a transformed Ω P , which for the right choice of P exhibits better properties than the original Ω: In the classical case, div(Ω P ) = 0, while in the fractional case, Ω P ∈ L 2,1 (where L 2,1 L 2 is the Lorentz space dual to the weak L 2 , denoted by L 2,∞ ).…”