Inverse and Implicit Function Theorems

The two structural theorems of multivariable calculus (analysis spine). Both say the same thing from different angles: a $C^1$ map looks, locally, like its derivative. Where the Jacobian is invertible, the map is locally invertible (inverse function theorem); where it has full rank, level sets are smooth submanifolds (implicit function theorem). These are what license parametrizing the curves and surfaces that the integral theorems integrate over.

References: Rudin, Principles of Mathematical Analysis, ch. 9; Spivak, Calculus on Manifolds, ch. 2; Lee, Introduction to Smooth Manifolds.

1. The Inverse Function Theorem

Inverse Function Theorem (IFT). Let $f:U\to {\mathbb{R}}^n$ be $C^1$ on an open $U\subseteq {\mathbb{R}}^n$, and suppose $Df(a)$ is invertible at some $a\in U$. Then there are open neighbourhoods $V\ni a$ and $W\ni f(a)$ such that $f:V\to W$ is a bijection with a $C^1$ inverse, and $$D(f^{-1})(f(x))=(Df(x){)}^{-1}.$$ If $f$ is $C^k$ then so is $f^{-1}$.

The derivative formula is forced by the chain rule (multivariable.md §3) applied to $f^{-1}\circ f=\mathrm{id}$: $D(f^{-1})Df=I$.

1.1 Proof via the contraction mapping theorem

The existence half is the marquee application of the Banach fixed point theorem (metric-spaces.md §7). After translating and composing with the linear map $Df(a{)}^{-1}$ we may assume $a=0$, $f(0)=0$, $Df(0)=I$. To solve $f(x)=y$, define

$$T_y(x)=x-f(x)+y,\text{so }f(x)=y⟺T_y(x)=x.$$

Because $D{T}_y(x)=I-Df(x)$ is small near $0$ (as $Df$ is continuous and $Df(0)=I$), the mean value inequality (multivariable.md §6) makes $T_y$ a contraction on a small closed ball, which is complete. Banach's theorem yields a unique fixed point $x=g(y)$ for each nearby $y$, defining the inverse $g=f^{-1}$; a further estimate shows $g$ is $C^1$ with the stated derivative. $\square$

This is the payoff promised back in metric-spaces.md: completeness $\Rightarrow$ contraction has a fixed point $\Rightarrow$ smooth maps are locally invertible.

2. The Implicit Function Theorem

Often we cannot solve $F(x,y)=0$ explicitly for $y$, but want to know that $y$ is some smooth function of $x$ near a solution.

Implicit Function Theorem. Let $F:U\to {\mathbb{R}}^m$ be $C^1$ on an open set in ${\mathbb{R}}^n\times {\mathbb{R}}^m$, and let $(a,b)$ satisfy $F(a,b)=0$. Split the derivative into blocks $DF=[D_{x}F\mid D_{y}F]$. If the $m\times m$ block $D_{y}F(a,b)$ is invertible, then there are neighbourhoods of $a$ and $b$ and a unique $C^1$ map $g$ with $g(a)=b$ and $$F(x,g(x))=0,$$ whose derivative is $Dg(a)=-(D_{y}F(a,b){)}^{-1}{D}_{x}F(a,b).$

Proof. Apply the IFT to $\Phi (x,y)=(x,F(x,y))$, whose Jacobian is block-triangular with invertible diagonal blocks $I$ and $D_{y}F$, hence invertible. The second component of ${\Phi }^{-1}$ is $g$. $\square$

The derivative formula comes from differentiating $F(x,g(x))=0$ with the chain rule — the rigorous version of "implicit differentiation."

3. Level Sets Are Submanifolds

A point $a$ is a regular point of $F:U\to {\mathbb{R}}^m$ if $DF(a)$ has full rank $m$; $c$ is a regular value if every point of $F^{-1}(c)$ is regular.

Regular Value Theorem. If $c$ is a regular value of a $C^k$ map $F:U\subseteq {\mathbb{R}}^n\to {\mathbb{R}}^m$, then the level set $M=F^{-1}(c)$ is a $C^k$ submanifold of dimension $n-m$, with tangent space $T_{a}M=\ker DF(a)$ at each point.

This is the implicit function theorem read geometrically: near a regular point the $m$ constraints can be solved for $m$ of the coordinates, giving a local graph — which is exactly a chart (topology-manifolds.md §2). Examples: the sphere $S^{n-1}=\{|x{|}^2=1\}$ ($F(x)=|x{|}^2$, regular value $1$); matrix groups like $SO(n)=\{A:A^{\top }A=I\}$. The tangent-space identity $T_{a}M=\ker DF(a)$ says the gradient(s) of the constraints are normal to the level set — the basis of Lagrange multipliers and of orienting surfaces by their normal in vector-calculus.md.

4. The Rank Theorem (constant rank)

More generally, if $Df$ has constant rank $r$ near $a$, then in suitable local $C^k$ coordinates $f$ becomes the linear projection $(x_1,\dots ,x_n)\mapsto (x_1,\dots ,x_r,0,\dots ,0)$. The IFT ($r=n=m$) and the regular-value theorem ($r=m$) are special cases. This is the analytic foundation for immersions, submersions, and embeddings in differential geometry.

5. Application: Smooth Parametrizations

The integral theorems require curves, surfaces, and solid regions described by smooth parametrizations or as regular level sets. The theorems of this page guarantee these descriptions are consistent and locally equivalent:

• A regular level surface $\{g=c\}\subseteq {\mathbb{R}}^3$ is locally a graph $z=h(x,y)$, hence smoothly parametrizable — needed to define surface integrals (vector-calculus.md §4).
• The normal direction $\nabla g$ orients the surface, fixing the sign conventions in Stokes' and the divergence theorem (integral-theorems.md).

6. Where this page is used

• Smooth parametrization / orientation of level sets $\to$ line and surface integrals (vector-calculus.md).
• Submanifolds with boundary $\to$ the domains of integration in integral-theorems.md and stokes-general.md.
• Charts from graphs connect directly to the manifold definition in topology-manifolds.md §2.

Next: Multiple Integrals.