Non-quasicontinuous Newtonian functions and outer capacities based on Banach function spaces

Anders Björn · 2025

Read Non-quasicontinuous Newtonian functions and outer capacities based on Banach function spaces by Anders Björn (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-21665.txt.

Introduction

Anders Björn, Jana Björn, and Lukáš Malý constructed in 2025 examples of Sobolev-type functions on metric spaces that fail a continuity property long assumed in the theory. For the classical function spaces, such functions are known to be quasicontinuous on well-behaved spaces. Their work shows this breaks down for more general Banach function spaces.

The counterexamples live on classic fractal sets in the plane, including the snowflake curve, the Cantor set, and the Sierpiński carpet. Related questions, such as whether the associated capacity is an outer capacity, are tied to a missing regularity property of the function space. The results mark the boundary of a widely used theory.

Catalog record

CREATORAnders Björn
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE99 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-21665

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Non-quasicontinuous Newtonian functions and outer capacities based on Banach function spaces Anders Björn Abstract Abstract. We construct various examples of Sobolev-type functions, defined via upper gradients in metric spaces, that fail to be quasicontinuous or weakly quasicontinuous. This is done with quasi-Banach function lattices X as the function spaces defining the smoothness of the Sobolev-type functions. These results are in contrast to the case X=L^{p} with 1\leq p<\infty , where all Sobolev-type functions in N^{1,p} are known to be quasicontinuous, provided that the underlying metric space \mathcal{P} is locally complete. In most of our examples, \mathcal{P} is a compact subset of \mathbf{R}^{2} and X=L^{\infty} . Four particular examples are the damped topologist’s sine curve, the von Koch snowflake curve, the Cantor ternary set and the Sierpiński carpet. We also discuss several related properties, such as whether the Sobolev capacity is an outer capacity, and how these properties are related. A fundamental role in these considerations is played by the lack of the Vitali–Carathéodory property. Key words and phrases : Banach function lattice, Banach function space, metric space, Newtonian space, outer capacity, quasicontinuity, Sobolev capacity, upper gradient, Vitali–Carathéodory property, weak quasicontinuity. Mathematics Subject Classification (2020): Primary: 46E36, Secondary: 30L99, 31C15, 31E05, 46B42. 1 Introduction We assume throughout the paper that \mathcal{P}=(\mathcal{P},d,\mu) is a metric measure space equipped with a positive complete Borel measure \mu such that 0<\mu(B)<\infty for every ball B\subset\mathcal{P} . Such spaces allow themselves to first order analysis, including Sobolev type spaces, geometric analysis and potential theory as in Hajłasz [ 13 ] , [ 14 ] , Heinonen–Koskela [ 18 ] , Cheeger [ 10 ] , Shanmugalingam [ 28 ] , Heinonen [ 17 ] and many subsequent papers, as well as the books by Hajłasz–Koskela [ 15 ] , Heinonen [ 16 ] , Björn–Björn [ 5 ] and Heinonen–Koskela–Shanmugalingam–Tyson [ 19 ] . Since the late 1990s it has been an open question whether all functions in the (Sobolev) Newtonian space N^{1,p}(\mathcal{P}) , 1\leq p<\infty , are quasicontinuous, without additional assumptions on \mathcal{P} . It has recently been answered in the affirmative in Eriksson-Bique–Poggi-Corradini [ 12 , Theorem 1.3] when \mathcal{P} is locally complete. In this paper we shall see that this is not so in the wider generality of Newtonian spaces {N^{1}X}(\mathcal{P}) based on (quasi)-Banach function lattices X , and in particular not for N^{1,\infty}(\mathcal{P}) . A simple example of such a situation is when the metric space \mathcal{P}\subset[0,1] is the classical ternary Cantor set, equipped with the \tfrac{\log 2}{\log 3} -dimensional Hausdorff measure. In this case, the function \chi_{[0,a]\cap\mathcal{P}}\in N^{1,\infty}(\mathcal{P})=L^{\infty}(\mathcal{P}) is not quasicontinuous when a\in\mathcal{P} is not an end point of any of the generating intervals. The following result gives a more complete picture and shows the striking difference between N^{1,\infty}(\mathcal{P}) and N^{1,p}(\mathcal{P}) with 1\leq p<\infty . See Theorem 6.2 for a more refined version and Section 2 for the definitions. Theorem 1.1 . Assume that there are L^{\infty} -almost no nonconstant rectifiable curves in the metric space \mathcal{P} and that there is a point x_{0}\in\mathcal{P} with \mu(\{x_{0}\})=0 . Then N^{1,\infty}(\mathcal{P})=L^{\infty}(\mathcal{P}) and the following are true : 1. \chi_{\{x_{0}\}}\in N^{1,\infty}(\mathcal{P}) is weakly quasicontinuous, but not quasicontinuous. 2. The capacity {C_{\infty}^{\mathcal{P}}} is not an outer capacity, not even for sets of zero capacity. 3. If there is a closed set F with empty interior and \mu(F)>0 , then \chi_{F}\in N^{1,\infty}(\mathcal{P}) is not weakly quasicontinuous. In particular, all of these facts hold if \mathcal{P} is the von Koch snowflake curve equipped with the \frac{\log 4}{\log 3} -dimensional Hausdorff measure \mu . To appreciate the statement of Theorem 1.1 , let us turn back to N^{1,p}(\mathcal{P}) , with 1\leq p<\infty . Unlike the classical Sobolev functions on \mathbf{R}^{n} , the functions in N^{1,p}(\mathcal{P}) are by definition better than arbitrary a.e.-representatives. Namely, two N^{1,p} -functions which are equal a.e. are equal up to a set of zero p\mspace{1mu} -capacity (that is, quasieverywhere (q.e.)), because the other a.e.-representatives do not possess p\mspace{1mu} -integrable upper gradients. For metric spaces \mathcal{P} without (or with L^{p} -almost no) rectifiable curves and 1\leq p<\infty , the Newtonian space N^{1,p}(\mathcal{P}) coincides with L^{p}(\mathcal{P}) and the p\mspace{1mu} -capacity coincides with the underlying measure on \mathcal{P} . All u\in N^{1,p}(\mathcal{P}) are thus quasicontinuous by virtue of Luzin’s theorem. We mention already here that in this context, one crucial difference between N^{1,p} and N^{1,\infty} is that L^{p} has an absolutely continuous norm and satisfies the Vitali–Carathéodory property (VC, Definition 2.7 ), while L^{\infty} usually lacks these properties, see Section 6 . Clearly, without rectifiable curves, {N^{1}X}(\mathcal{P})=X(\mathcal{P}) for every quasi-Banach function lattice X , but this is also the case for the well-connected Sierpiński carpet S_{1/3} , see Example 9.1 . As a consequence, Theorem 1.1 applies also to \mathcal{P}=S_{1/3} . When \mathcal{P} is complete and has enough rectifiable curves (so that a p\mspace{1mu} -Poincaré inequality holds, and also assuming that \mu is doubling), it was shown in Björn–Björn–Shanmugalingam [ 8 , Theorem 1.1 and p. 1198] that all functions in N^{1,p}(\mathcal{P}) , 1\leq p<\infty , are quasicontinuous. (This had been conjectured already in the late 1990s by Shanmugalingam [ 27 , Corollary 3.1.7] , [ 28 , Remark 4.4] , see also [ 8 , p. 1198] .) This result has later been generalized in various ways, see [ 19 , Theorem 8.2.1] , [ 7 , p. 1190] and [ 6 , Theorem 9.1] . The assumption of Poincaré inequality was (in complete doubling spaces and for p>1 ) removed by Ambrosio et al. in [ 1 ] and [ 2 ] (cf. [ 6 , Remark 8.9] ), while Eriksson-Bique–Poggi-Corradini [ 12 , Theorem 1.3] recently removed most of the earlier assumptions and only require that \mathcal{P} is locally complete. The assumption that \mu(B)<\infty for every ball B is however crucial, as our Example 7.2 (b) shows. In the world of (quasi)-Banach function lattices, the letter X is usually reserved for the function lattice itself and corresponds in our setting to the L^{p} or L^{\infty} spaces, which control the Sobolev functions and their gradients. For the underlying metric space carrying the Sobolev functions, we therefore use \mathcal{P} rather than X , which is otherwise the usual notation in analysis on metric spaces based on L^{p} as the function space. Quasicontinuity is closely related to the outer property of the Sobolev capacity {C_{X}^{\mathcal{P}}} associated with {N^{1}X(\mathcal{P})} . We also study this property and how it is related to quasicontinuity, as well as some variants of these properties. The following theorem summarizes quite well what can happen in the general case. Here {\overline{\mathbf{R}}}:=[-\infty,\infty] . Theorem 1.2 . Consider the following statements for general quasi-Banach function lattices X and metric spaces \mathcal{P} : (A). Every u\in{N^{1}X}(\mathcal{P}) is quasicontinuous. (B). {C_{X}^{\mathcal{P}}} is an outer capacity. (C). {C_{X}^{\mathcal{P}}} is a quasiouter capacity. (D). Every weakly quasicontinuous u:\mathcal{P}\to{\overline{\mathbf{R}}} is quasicontinuous. (E). {C_{X}^{\mathcal{P}}} is an outer capacity for sets of zero capacity. (F). Every u\in{N^{1}X(\mathcal{P})} has a quasicontinuous representative v , i.e. v=u q.e. (G). Every u\in{N^{1}X(\mathcal{P})} is weakly quasicontinuous. Then (\ref{b-weak-qcont}\text{ and }\ref{b-wqcont})\mathchoice{\quad\Longleftrightarrow\quad}{\Leftrightarrow}{\Leftrightarrow}{\Leftrightarrow}(\ref{b-outer-zero}\text{ and }\ref{b-repr})\mathchoice{\quad\Longleftrightarrow\quad}{\Leftrightarrow}{\Leftrightarrow}{\Leftrightarrow}\ref{b-qcont}. (1.1) and \begin{split}\lx@xy@svg{\hbox{\raise 2.55554pt\hbox{\kern 6.8889pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\halign{\entry@#!@&&\entry@@#!@\cr&&&&\cr&\cr&\crcr}}}\ignorespaces{\hbox{\kern-6.8889pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise-2.55554pt\hbox{$\textstyle{\ref{b-qcont}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\hbox{\kern 0.0pt\raise 1.0pt\hbox{\lx@xy@droprule}}\hbox{\kern 0.0pt\raise-1.0pt\hbox{\lx@xy@droprule}}}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 8.3889pt\raise 7.5pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.5pt\hbox{$\scriptstyle{{\textstyle\kern 5.0pt\not\kern-5.0pt\Longleftarrow}}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 30.8889pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1.5}\lx@xy@tip{-1.5}}}}}}{\hbox{\hbox{\kern 0.0pt\raise 1.0pt\hbox{\lx@xy@droprule}}\hbox{\kern 0.0pt\raise-1.0pt\hbox{\lx@xy@droprule}}}}{\hbox{\hbox{\kern 0.0pt\raise 1.0pt\hbox{\lx@xy@droprule}}\hbox{\kern 0.0pt\raise-1.0pt\hbox{\lx@xy@droprule}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\hbox{\kern 0.86096pt\raise 0.50867pt\hbox{\lx@xy@drawline@}}\hbox{\kern-0.86096pt\raise-0.50867pt\hbox{\lx@xy@drawline@}}}}\ignorespaces{\hbox{\kern 33.09183pt\raise-55.94446pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1.5}\lx@xy@tip{-1.5}}}}}}\ignorespaces\ignorespaces{\hbox{\hbox{\kern

Read the complete text