The fundamental theorem of surface theory for surfaces with little regularity (Q702343)

If \(\omega\) is a connected, simply-connected region in \(\mathbb R^2\) on which a symmetric positive definite matrix field \((a_{\alpha\beta})\) of order two and class \(C^2\) and a symmetric matrix field \((b_{\alpha\beta})\) of order two and class \(C^1\) are defined, satisfying the Gauss and Codazzzi-Mainardi equations, viz. \[ \partial_\gamma \Gamma^\tau _{\alpha\beta} - \partial_\beta \Gamma^\tau_{\alpha\gamma} + \Gamma^\sigma_{\alpha\beta} \Gamma^\tau_{\sigma\gamma}- \Gamma^\sigma_{\alpha\gamma} \Gamma^\tau_{\sigma\beta} = b_{\alpha\beta} b^\tau_\gamma - b_{\alpha\gamma}b^\tau_\beta,\tag \(*\) \] \[ \partial_\gamma b_{\alpha\beta} - \partial_\beta b_{\alpha\gamma} + \Gamma^\sigma_{\alpha\beta} b_{\sigma\gamma} - \Gamma^\sigma_{\alpha\gamma} b_{\sigma\beta} = 0 \tag \(**\) \] (\(\Gamma\) being the usual Christoffel symbols), then the fundamental theorem of surface theory guarantees the existence of a mapping \(\theta\: \omega \to \mathbb R^3\) of class \( C^3\) so that the first and the second fundamental forms of the surface \(S = \theta (\omega)\) are precisely the given matrix fields \((a_{\alpha\beta})\) and \((b_{\alpha\beta}).\) More precisely, this means \(\partial_\alpha \theta \cdot \partial_\beta \theta = a_{\alpha\beta}\) and \(\partial_{\alpha\beta} \theta \cdot \xi = b_{\alpha\beta},\) where \(\xi\) is the normalized \(| \partial_1\theta \wedge \partial_2\theta| \) (the unit normal to \(S\)). Although for the definition of \((a_{\alpha\beta})\) and \((b_{\alpha\beta})\) it suffices to assume \(\theta\) to belong only to \(C^2(\omega , \mathbb R^3),\) in which case \(a_{\alpha\beta}\) and \(b_{\alpha\beta}\) are respectively in \(C^1(\omega)\) and \(C^0(\omega),\) for the recovery of \(\theta\) from matrix fields \((a_{\alpha\beta})\) and \((b_{\alpha\beta})\) extra regularity is needed and one has to assume that these fields belong to \(C^2(\omega)\) respectively \(C^1(\omega),\) in which case \(\theta \in C^3(\omega, \mathbb R^3).\) To get rid of this extra regularity Hartman and Wintner assume \(a_{\alpha\beta}\) and \(b_{\alpha\beta}\) to be in \(C^1(\omega)\) and \(C^0(\omega)\) respectively, and assume that these fields satisfy the Gauss and Codazzi-Mainardi equations in the ``integrated'' form, in which the appropriate forms of \((*)\) and \((**)\) are integrated along Jordan curves of class \(C^1\) or over regions bounded by such curves and contained in \(\omega\) [\textit{P. Hartman} and \textit{A. Wintner}, Am. J. Math. 72, 757--774 (1950; Zbl 0039.16802)]. In the present paper the author assumes that the Gauss and the Codazzi-Mainardi equations are satisfied in a distributional sense where a positive definite symmetric matrix field \((a_{\alpha\beta})\) belongs to the Sobolev space \(W^{1, \infty}_{\text{loc}}(\omega)\) and a symmetric matrix field \((b_{\alpha\beta})\) is of class \(L^\infty_{\text{loc}}(\omega).\) Under these assumptions he proves the existence of a mapping \(\theta : \omega \to \mathbb R^3\) of class \(W^{2, \infty}_{\text{loc}}(\omega)\) by which \((a_{\alpha\beta})\) and \((b_{\alpha\beta})\) are the first and the second fundamental form of the surface \(S = \theta (\omega).\) In this context, Hartman and Wintner's result becomes a corollary of this more general result.