4
$\begingroup$

Denote by $Lip_0(X)$ the set of all Lipschitz functions on a metric space $X$ vanishing at some base point $e \in X$. The norm in $Lip_0$ is defined as fololows $$ \|f\|_{Lip_0} := Lip(f), $$ where $Lip(f)$ denotes the Lipschitz constant. With pointwise operations $f \vee g := \max\{f,g\}$ and $f \wedge g := \min\{f,g\}$ the space $Lip_0$ becomes a Lipschitz lattice, in which the following condition holds $$ \|f \vee g\|_{Lip_0} \leq \max\{\|f\|_{Lip_0},\|g\|_{Lip_0}\}. $$ The Banach lattice condition $|f| \leq |g| \implies \|f\| \leq \|g\|$, however, fails. (Nik Weaver. Lipschitz Algebras, 2nd ed.)

Question. Are operations $f_+ := f \vee 0$, $f_- := (-f) \vee 0$ and $|f| := f \vee (-f)$ continuous in the $Lip_0$ norm, i.e. does, e.g., $$ \|f_+ - g_+\|_{Lip_0} \leq C\|f - g\|_{Lip_0} $$ hold?

I have searched a lot for either a proof or a counterexample, but couldn't find anything. Any help will be appreciated.

$\endgroup$
2
$\begingroup$

No, the lattice operations are not continuous, in general.

An easy way to see this is to consider the space of Lipschitz continuous functions on $[0,1]$ (without base point, and the norm being the supremum of the infinity norm and the Lipschitz constant).

Set $f_\varepsilon(x) := x - \varepsilon$ for all $x \in [0,1]$. Then $f_\varepsilon \to f_0$ as $\varepsilon \to 0$, but convergence of the absolute values does not hold.

If you insist on using a base point, add the point $-1$ and set all functions to $0$ there.

$\endgroup$
3
  • 1
    $\begingroup$ thanks! there is a typo that I'm not allowed to correct: the limit is x, not 0 $\endgroup$ – Yury Aug 28 '20 at 15:43
  • 1
    $\begingroup$ @Yury: Thank you for the correction! I actually intended to write $f_0$ instead of $0$ :-). Corrected. Wunderfull question, by the way - for several years I have been wondering now and then whether the lattice operations in a lattice ordered Banach space with non-normal cone are automatically continuous; but I never thought thoroughly enough about it - until I read your question. $\endgroup$ – Jochen Glueck Aug 28 '20 at 15:51
  • 1
    $\begingroup$ Thank you! :) by the way, just as a matter of reference, I asked a similar question about sequential weak-* convergence on MO. The answer in this case is yes. $\endgroup$ – Yury Jan 5 at 16:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.