The implicit function theorem for maps that are only differentiable: an elementary proof (Q1664077)

In this paper, the author presents an elementary proof of a version of the implicit function theorem that is stronger than the classical one. One proves the implicit function theorem for differentiable maps \(F(x,y)\), defined on a finite-dimensional Euclidean space, assuming that the leading principal minors of the Jacobian matrix \(\frac{\partial F}{\partial y} (x,y)\) are nowhere vanishing; there are no hypotheses on the continuity of the partial derivatives of the map \(F\). The case of one equation with several real variables and a differentiable function, that is, \(F: \mathbb{R}^{n+1}\to \mathbb{R}\) has been considered by the author in a previous paper. For the general case the result is the following: Let \(F: \Omega\to \mathbb{R}^{m}\) be differentiable, with \(\Omega\) a non-degenerate open ball within \(\mathbb{R}^{n}\to \mathbb{R}^{m}\) and centered at \((a,b)\). Let us suppose that \(F(a,b)=0\) and that all the \(m\) leading principal minors of the matrix \(\frac{\partial F}{\partial y}\) are nowhere vanishing. The following are true. \(\bullet\) There exists an open set \(X\times Y\), within \(\Omega\) and containing \((a,b)\), and a differentiable function \(g: X\to Y\) satisfying \[ F(x,g(x))=0,\text{ for all }x\in X,\quad \text{and }g(a)=b. \] \(\bullet\) We have \[ Jg(x)=-\left[\frac{\partial F}{\partial y} (x,g(x))\right]^{-1}_{m\times m}\left[ \frac{\partial F}{\partial x} (x,g(x))\right]_{m\times n},\text{ for all \(x\) in \(X\)}. \] Let us suppose that we also have \(\det\bigl(\frac{\partial F_{j}}{\partial y_{j}}(\xi_{i})\bigr)_{1\leq i,j\leq m}\neq 0\), for every point \((\xi_{1},\dotsc,\xi_{m})\) in \(\Omega^{m}\). Then, the following uniqueness is true. \(\bullet\) If \(h: X\to Y\) satisfies \(F(x,h(x))=0\) for all \(x\in X\), then we have \(h=g\). From this, one can deduce the corresponding inverse function theorem: Let \(F: \Omega\to \mathbb{R}^{n}\) be a differentiable map, with \(\Omega\) a non-degenerate open ball within \(\mathbb{R}^{n}\) and centered at the point \(x_{0}\). Let us suppose that all the \(n\) leading principal minors of \(JF(x)\) are nowhere vanishing. We also suppose \(\det\bigl(\frac{\partial F_{j}}{\partial x_{j}}(\xi_{i})\bigr)_{1\leq i,j\leq n}\neq 0\) for every point \((\xi_{1},\dotsc,\xi_{n})\) inside \(\Omega^{n}\). Under such conditions, there exist an open set \(X\) containing \(x_{0}\), an open set \(Y\) containing \(y_{0}=F(x_{0})\), and a differentiable map \(G: Y\to X\) satisfying \[ F(G(y))=y\text{ for all }y\in Y,\text{ and } G(F(x))=x\text{ for all }x\in X. \] In addition, we have \[ JG(y)=JF(G(y))^{-1}\text{ for all }y\text{ in }X. \] The author also provides some illustrative examples.