Incompatible elasticity and the immersion of non-flat Riemannian manifolds in Euclidean space (Q1760395)

The authors study a variational problem for immersions of a Riemannian manifold into a Euclidean space of the same dimension. It originates from the theory of incompatible elasticity but the discussion is purely geometric. Given a Riemannian \( n \)-dimensional manifold \( (M,\overline{g}) \) with boundary (homeomorphic to a bounded simply-connected subset of \( \mathbb{R}^{n} \) in order to avoid topological obstructions) let \( \mathcal{F} \) denote the set of smooth immersions \( f: M \to \mathbb{R}^{n} \), \( \mathbb{R}^{n} \) equipped with its standard Riemannian metric \( e \). So, for any \( f \), one has on \( M \) besides the Riemannian metric \( \overline{g} \) the pullback \( g := f^{\displaystyle\ast}e \). A natural question is how to measure the extent to which \( \overline{g} \) deviates from flatness. A commonly used quantity is the function \( \left|g-\overline{g}\right|_{\overline{g}}^{2} \), resp., its integral over \( M \) against the volume element of \( \overline{g} \). In this paper, reasons are given to consider another density, namely \[ W_{\text{quad}}(df(x),x) := \min_{Q\in SO(n)} \left|df(x)-Q\overline{q}(x)\right|_{\overline{g},e}^{2}, \] where \( \overline{q} \) is a square root of the metric \( \overline{g} \) (with respect to the Euclidean inner product), i.e., \( \overline{g}(u,v) = e(\overline{q}(u),\overline{q}(v)) \), \( u,v \in T_{p}M \), and \( Q \) runs through the proper isometries of the corresponding image tangent space at \( f(p) \). The functionals finally to be considered have the form of elastic energies \[ E[f] := \int_{M} W(df(x),x)\,\text{vol}_{\overline{g}}(x) \] under coercivity constraints \[ W(df(x),x) \geq c_{1}W_{\text{quad}}(df(x),x), \qquad W(df(x),x) \geq c_{2}\left|(\ast_{\overline{g}}-\ast_{g})df\right|_ {\overline{g},e}^{2}, \] the \( \ast \)'s denoting Hodge-operators of the respective metrics. The authors point out that the second condition is essentially without loss of generality. The main goal is to establish lower bounds for \( E[f] \) in terms of curvature quantities. (Unfortunately, there is a misprint in the definition of the curvature tensor on p. 140.) Using harmonic analysis (see, e.g., [\textit{J. Eells} and \textit{L. Lemaire}, Two reports on harmonic maps. Singapore: World Scientific (1995; Zbl 0836.58012)]) it is shown that the analysis can be restricted to bounding from below the functional \( E_{\text{quad}}[f] := \int_{M} W_{\text{quad}}(df(x),x)\operatorname{vol}_{\overline{g}}(x) \) over the set of harmonic maps \( f: M \to \mathbb{R}^{n} \). As main results, they thus obtain an explicit lower bound depending (besides the constants \( c_{1},c_{2} \)) on the scalar curvature \( s_{\overline{g}} \) together with arbitrary test functions \( \chi \) on \( M \) (case \( n \geq 2 \) and assuming \( s_{\overline{g}} \geq 0 \)) and the Gauss curvature \( K \) (case \( n = 2 \) and assuming no sign condition on \( K \)).