Skip to content

Formalization of the proof for partial Lipschitz map extensions written in Lean4

Notifications You must be signed in to change notification settings

chriscamano/Formal_Lipschitz

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

48 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

This code was written as part of the 2023 MSRI "Formalization of Mathematics" workshop hosted in Berkeley California

Formal_Lipschitz

Implementation of Theorem 2.2 of "Metric Embeddings and Lipschitz Extentensions by Assaf Naor" page 5

Link: https://web.math.princeton.edu/~naor/homepage%20files/embeddings_extensions.pdf

Code now integrated in Mathlib4 under :

Mathlib.Topology.MetricSpace.Kuratowski

Mathlib.Analysis.lpSpace

Mathlib.Topology.MetricSpace.Lipschitz

Theorem 2.2.

Suppose $\left(X, d_X\right)$ is a metric space, $A \subseteq X$ and let $f: A \rightarrow \ell_{\infty}(\Gamma)$ a Lipschitz function. Then, there is an

extension $\tilde{f}: X \rightarrow \ell_{\infty}(\Gamma)$ of $f$, i.e. with $\tilde{f}|_A=f$, such that $|\tilde{f}|_{\text {Lip }}=|f|_{\text {Lip }}$.

Proof.

Let $|f|_{\text {Lip }}=L$ and write

$$ f(x)=\left(f_\gamma(x)\right)_{\gamma \in \Gamma}, \text { for } x \in X \text { and } f_\gamma(x) \in \mathbb{R} $$

Then, we see that

$$ |f|_{\text {Lip }}=\sup _{d_X(x, y) \leqslant 1} \sup _{\gamma \in \Gamma}\left|f_\gamma(x)-f_\gamma(y)\right|=\sup _{\gamma \in \Gamma}\left|f_\gamma\right|_{\text {Lip }}, $$

thus each $f_\gamma$ is also $L$-Lipschitz. Thus, it is enough to extend all the $f_\gamma$ isometrically, that is prove our theorem with $\mathbb{R}$ replacing $\ell_{\infty}(\Gamma)$. This will be done in the next important lemma.

Lemma 2.3

(Nonlinear Hahn-Banach theorem). Suppose $\left(X, d_X\right)$ is a metric space, $A \subseteq X$ and let $f: A \rightarrow \mathbb{R}$ a Lipschitz function.

Then, there is an extension $\tilde{f}: X \rightarrow \mathbb{R}$ of $f$, i.e. with $\left.\tilde{f}\right|_A=f$, such that $|\tilde{f}|_{\text {Lip }}=|f|_{\text {Lip }}$

First-direct proof. Call again $L=|f|_{\text {Lip }}$ and define the function $\tilde{f}: X \rightarrow \mathbb{R}$ by the formula

$$ \tilde{f}(x)=\inf _{a \in A}\left\{f(a)+L d_X(x, a)\right\}, \quad x \in X $$

To see that this function satisfies the results, fix an arbitrary $a_0 \in A$. Then, for any $a \in A$ :

$$ f(a)+L d_X(x, a) \geqslant f\left(a_0\right)-L d\left(a, a_0\right)+L d(x, a) \geqslant f\left(a_0\right)-L d\left(x, a_0\right)>-\infty, $$

so $\tilde{f}(x)$ is well-defined. Also, if $x \in A$, the above shows that $\tilde{f}(x)=f(x)$. Finally, for $x, y \in X$ and $\varepsilon>0$, choose $a_x \in A$ such that

$$ \tilde{f}(x) \geqslant f\left(a_x\right)+L d\left(x, a_x\right)-\varepsilon $$

Then,

$$ \tilde{f}(y)-\tilde{f}(x) \leqslant f\left(a_x\right)+L d\left(y, a_x\right)-f\left(a_x\right)-L d\left(x, a_x\right)+\varepsilon \leqslant L d(x, y)+\varepsilon . $$

Thus, we see that $\tilde{f}$ is indeed $L$-Lipschitz.

About

Formalization of the proof for partial Lipschitz map extensions written in Lean4

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages