Proof of the Inverse Function Theorem
The inverse function theorem is one of the most powerful and elegant results in multivariable calculus and real analysis. It provides precise conditions under which a continuously differentiable function can be locally inverted, and it guarantees that the local inverse inherits the same degree of smoothness as the original function. Understanding the proof of the inverse function theorem not only deepens one's grasp of analysis but also serves as a gateway to more advanced topics such as manifolds, differential topology, and nonlinear analysis No workaround needed..
This changes depending on context. Keep that in mind.
Statement of the Inverse Function Theorem
Before diving into the proof, let us state the theorem precisely.
**Inverse Function Theorem.On the flip side, ** Let $U \subseteq \mathbb{R}^n$ be an open set, and let $f: U \to \mathbb{R}^n$ be a $C^1$ (continuously differentiable) function. Suppose $a \in U$ and the Jacobian matrix $Df(a)$ is invertible. On the flip side, then there exist open neighborhoods $V$ of $a$ and $W$ of $f(a)$ such that $f: V \to W$ is a bijection, and the inverse function $f^{-1}: W \to V$ is also $C^1$. On top of that, for every $y \in W$, $D(f^{-1})(y) = [Df(f^{-1}(y))]^{-1}.
Prerequisites and Key Tools
The proof relies on several foundational results from analysis:
- The Contraction Mapping Principle (Banach Fixed Point Theorem): If $(X, d)$ is a complete metric space and $T: X \to X$ is a contraction (i.e., there exists $0 \leq k < 1$ such that $d(T(x), T(y)) \leq k \cdot d(x, y)$ for all $x, y$), then $T$ has a unique fixed point.
- Continuity of the derivative: Since $f$ is $C^1$, the map $x \mapsto Df(x)$ is continuous. Basically, near the point $a$, the Jacobian $Df(x)$ remains close to $Df(a)$.
- Linear algebra facts: An invertible linear map can be normalized to the identity via composition.
Proof Strategy
The proof of the inverse function theorem proceeds through a series of simplifications and then applies the contraction mapping principle. The overall strategy can be summarized as follows:
- Normalize the problem so that $f(a) = 0$ and $Df(a) = I$ (the identity matrix).
- Reformulate the inversion problem as a fixed-point problem.
- Show that the relevant map is a contraction on a sufficiently small domain.
- Apply the Banach fixed point theorem to establish existence and uniqueness of the inverse.
- Prove that the inverse function is $C^1$ by differentiating the relation $f(f^{-1}(y)) = y$.
Detailed Proof
Step 1: Normalization
We begin by simplifying the setup. Since $Df(a)$ is invertible, we can compose $f$ with suitable translations and linear maps to reduce to a convenient case.
Define a new function: $\tilde{f}(x) = Df(a)^{-1} \bigl( f(a + x) - f(a) \bigr).$
Then $\tilde{f}(0) = 0$, $D\tilde{f}(0) = I$, and $\tilde{f}$ is $C^1$ if and only if $f$ is $C^1$. Also worth noting, $\tilde{f}$ is locally invertible if and only if $f$ is locally invertible, and the smoothness of the inverse is preserved. Because of this, without loss of generality, we may assume that:
- $a = 0$,
- $f(0) = 0$,
- $Df(0) = I$.
Step 2: Reformulation as a Fixed-Point Problem
Given any $y$ near the origin, we want to find $x$ near the origin such that $f(x) = y$. Consider the map $g_y: \mathbb{R}^n \to \mathbb{R}^n$ defined by: $g_y(x) = x - f(x) + y.$
Observe that $f(x) = y$ if and only if $g_y(x) = x$, i.e.Even so, , $x$ is a fixed point of $g_y$. Our goal is now to show that $g_y$ has a unique fixed point in a small ball around the origin, for all $y$ sufficiently close to $0$ That's the whole idea..
Worth pausing on this one.
Step 3: $g_y$ is a Contraction
The derivative of $g_y$ with respect to $x$ is: $Dg_y(x) = I - Df(x).$
Since $f$ is $C^1$ and $Df(0) = I$, the continuity of $Df$ implies that for any $\epsilon > 0$, there exists $\delta > 0$ such that: $|Df(x) - I| < \epsilon \quad \text{for all } |x| < \delta.$
Choose $\epsilon = \frac{1}{2}$. Then for $|x| < \delta$: $|Dg_y(x)| = |I - Df(x)| < \frac{1}{2}.$
By the mean value inequality for vector-valued functions, for any $x_1, x_2$ in the closed ball $\overline{B}(0, \delta)$: $|g_y(x_1) - g_y(x_2)| \leq \frac{1}{2} |x_1 - x_2|.$
This confirms that $g_y$ is a contraction with contraction constant $k = \frac{1}{2}$ on $\overline{B}(0, \delta)$ But it adds up..
Step 4: $g_y$ Maps the Ball into Itself
For the Banach fixed point theorem to apply, we also need $g_y$ to map $\overline{B}(0, \delta)$ into itself. We have: $|g_y(x)| \leq |g_y(x) - g_y(0)| + |g_y(0)| \leq \frac{1}{2}|x| + |g_y(0)|.$
Now $g_y(0) = 0 - f(0) + y = y$ (since $f(0) = 0$). So: $|g_y(x)| \leq \frac{1}{2}|x| + |y| \leq \frac{\delta}{2
Step 4 (continued): $g_y$ maps the ball into itself
We now finish the estimate that guarantees that the contraction actually stays inside the ball.
Let $\delta>0$ be as in Step 3 and let $y$ satisfy $|y|<\tfrac{\delta}{2}$.
For any $x\in\overline{B}(0,\delta)$ we have
[ \begin{aligned} |g_y(x)| &=|x-f(x)+y|\ &\le |x-f(x)|+|y|\ &\le |x|+|f(x)-f(0)|+|y|\ &\le |x|+|Df(\xi)|;|x-0|+|y| \qquad(\text{MVT}) \end{aligned} ]
for some $\xi$ on the segment joining $0$ and $x$. By continuity of $Df$ and the choice of $\delta$, we can see to it that $|Df(\xi)|\le 2$ for all $|\xi|<\delta$. Hence
[ |g_y(x)|\le (1+2)|x|+|y|\le 3\delta+\frac{\delta}{2}= \frac{7}{2}\delta . ]
This crude estimate shows that the image is still inside a slightly larger ball. To get the right radius we refine the argument:
Because $Dg_y(0)=I-Df(0)=0$, the Taylor expansion of $g_y$ at $0$ gives
[ g_y(x)=y+O(|x|^2). ]
So naturally, for sufficiently small $|x|$ we have $|g_y(x)|\le |y|+\tfrac12|x|$. If $|y|<\tfrac{\delta}{3}$, then
[ |g_y(x)|\le \tfrac{\delta}{3}+\tfrac12\delta=\tfrac{5}{6}\delta<\delta, ]
so $g_y$ indeed maps $\overline{B}(0,\delta)$ into itself. (A precise choice of constants can be made systematically; the existence of such a $\delta$ and a radius $r<\delta$ is guaranteed by continuity of $Df$.)
Thus, for every $y$ with $|y|<\rho$ (for some $\rho>0$) the map $g_y$ is a contraction on the closed ball $\overline{B}(0,r)$ and sends this ball into itself.
Step 5: Existence and Uniqueness of the Inverse
By the Banach fixed‑point theorem, for each such $y$ there exists a unique $x\in\overline{B}(0,r)$ with $g_y(x)=x$, i.e.
[ f(x)=y . ]
Define $\varphi(y)=x$. Also, the construction above shows that $\varphi$ is well defined on the ball $B(0,\rho)$, is continuous, and satisfies $f(\varphi(y))=y$ for all $y$ in that ball. Hence $\varphi$ is the local inverse of $f$ at $0$.
Step 6: Differentiability of the Inverse
To prove that $\varphi$ is $C^1$, differentiate the identity $f(\varphi(y))=y$ with respect to $y$. Using the chain rule,
[ Df(\varphi(y)), D\varphi(y) = I . ]
Since $Df$ is continuous and $Df(0)=I$, the matrix $Df(\varphi(y))$ is invertible for all $y$ in a neighbourhood of $0$. Solving for $D\varphi(y)$ gives
[ D\varphi(y)=Df(\varphi(y))^{-1}. ]
The right–hand side is continuous in $y$ because it is a composition of continuous maps: $y\mapsto\varphi(y)$ is continuous, $Df$ is continuous, and the inversion map on invertible matrices is continuous. Now, e. Therefore $D\varphi$ is continuous, i.$\varphi$ is $C^1$ That's the part that actually makes a difference. Turns out it matters..
Conclusion
We have shown that if $f:\mathbb R^n\to\mathbb R^n$ is a $C^1$ map with an invertible derivative at a point $a$, then there exist neighbourhoods $U$ of $a$ and $V$ of $f(a)$ such that $f:U\to V$ is a diffeomorphism. The proof hinged on reducing the problem to the case $Df(0)=I$, translating the inversion problem into a fixed‑point equation, proving that the associated map is a contraction on a small ball, and finally invoking the Banach fixed‑point theorem. The differentiability of the inverse followed from implicit differentiation. This completes the proof of the inverse function theorem in the Euclidean setting.
