Kuratowski convergence

In mathematics, Kuratowski convergence or Painlevé-Kuratowski convergence is a notion of convergence for subsets of a topological space. First introduced by Paul Painlevé in lectures on mathematical analysis in 1902,This is reported in the Commentary section of Chapter 4 of Rockafellar and Wets' text. the concept was popularized in texts by Felix Hausdorff{{Cite book |last=Hausdorff |first=Felix |title=Mengenlehre |publisher=Walter de Gruyter & Co. |year=1927 |edition=2nd |location=Berlin |language=de}} and Kazimierz Kuratowski.{{Cite book |last=Kuratowski |first=Kazimierz |title=Topologie, I & II |publisher=Panstowowe Wyd Nauk |year=1933 |location=Warsaw |language=fr}} Intuitively, the Kuratowski limit of a sequence of sets is where the sets "accumulate".

Definitions

For a given sequence \{x_n\}_{n=1}^{\infty} of points in a space X, a limit point of the sequence can be understood as any point x \in X where the sequence eventually becomes arbitrarily close to x. On the other hand, a cluster point of the sequence can be thought of as a point x \in X where the sequence frequently becomes arbitrarily close to x. The Kuratowski limits inferior and superior generalize this intuition of limit and cluster points to subsets of the given space X.

= Metric Spaces =

Let (X,d) be a metric space, where X is a given set. For any point x and any non-empty subset A \subset X, define the distance between the point and the subset:

:d(x, A) := \inf_{y \in A} d(x, y), \qquad x \in X.

For any sequence of subsets \{A_n\}_{n=1}^{\infty} of X, the Kuratowski limit inferior (or lower closed limit) of A_n as n \to \infty; is\begin{align}

\mathop{\mathrm{Li}} A_{n} :=&

\left\{ x \in X : \begin{matrix} \mbox{for all open neighbourhoods } U \mbox{ of } x, U \cap A_{n} \neq \varnothing \mbox{ for large enough } n \end{matrix} \right\} \\

=&\left\{ x \in X : \limsup_{n \to \infty} d(x, A_{n}) = 0 \right\};

\end{align}the Kuratowski limit superior (or upper closed limit) of A_n as n \to \infty; is\begin{align}

\mathop{\mathrm{Ls}} A_{n} :=&

\left\{ x \in X : \begin{matrix} \mbox{for all open neighbourhoods } U \mbox{ of } x, U \cap A_{n} \neq \varnothing \mbox{ for infinitely many } n \end{matrix} \right\} \\

=&\left\{ x \in X : \liminf_{n \to \infty} d(x, A_{n}) = 0 \right\};

\end{align}If the Kuratowski limits inferior and superior agree, then the common set is called the Kuratowski limit of A_n and is denoted \mathop{\mathrm{Lim}}_{n \to \infty} A_n.

= Topological Spaces =

If (X, \tau) is a topological space, and \{A_i\}_{i \in I} are a net of subsets of X, the limits inferior and superior follow a similar construction. For a given point x \in X denote \mathcal{N}(x) the collection of open neighborhoods of x. The Kuratowski limit inferior of \{A_i\}_{i \in I} is the set\mathop{\mathrm{Li}} A_i := \left\{

x \in X :

\forall U \in \mathcal{N}(x) \, \exists i_0 \in I \colon \forall i \geq i_0 \Rightarrow U \cap A_i \neq \varnothing

\right\},and the Kuratowski limit superior is the set\mathop{\mathrm{Ls}} A_i := \left\{

x \in X :

\forall U \in \mathcal{N}(x) \, \forall i_0 \in I \, \exists i \geq i_0 \colon U \cap A_i \neq \varnothing

\right\}.Elements of \mathop{\mathrm{Li}} A_i are called limit points of \{A_i\}_{i \in I} and elements of \mathop{\mathrm{Ls}} A_i are called cluster points of \{A_i\}_{i \in I}. In other words, x is a limit point of \{A_i\}_{i \in I} if each of its neighborhoods intersects A_i for all i in a "residual" subset of I, while x is a cluster point of \{A_i\}_{i \in I} if each of its neighborhoods intersects A_i for all i in a cofinal subset of I.

When these sets agree, the common set is the Kuratowski limit of \{A_i\}_{i \in I}, denoted \mathop{\mathrm{Lim}} A_i.

Examples

  • Suppose (X, d) is separable where X is a perfect set, and let D = \{d_1, d_2, \dots\} be an enumeration of a countable dense subset of X. Then the sequence \{A_n\}_{n=1}^{\infty} defined by A_n := \{d_1, d_2, \dots, d_n\} has \mathop{\mathrm{Lim}} A_n = X.
  • Given two closed subsets B, C \subset X, defining A_{2n-1} := B and A_{2n} := C for each n=1,2,\dots yields \mathop{\mathrm{Li}} A_n = B \cap C and \mathop{\mathrm{Ls}} A_n = B \cup C.
  • The sequence of closed balls A_n := \{y \in X: d(x_n,y) \leq r_n\}converges in the sense of Kuratowski when x_n \to x in X and r_n \to r in [0, +\infty), and in particular, \mathop{\mathrm{Lim}}(A_n) = \{y \in X : d(x,y) \leq r\}. If r_n \to +\infty, then \mathop{\mathrm{Lim}} A_n = X while \mathop{\mathrm{Lim}}(X \setminus A_n) = \emptyset.
  • Let A_{n} := \{ x \in \mathbb{R} : \sin (n x) = 0 \}. Then A_n converges in the Kuratowski sense to the entire line.
  • In a topological vector space, if \{A_n\}_{n=1}^{\infty} is a sequence of cones, then so are the Kuratowski limits superior and inferior. For example, the sets A_n := \{(x,y) \in \mathbb{R}^2 : y \geq n|x|\} converge to \{(0,y) \in \mathbb{R}^2 : y \geq 0\}.

Properties

The following properties hold for the limits inferior and superior in both the metric and topological contexts, but are stated in the metric formulation for ease of reading.The interested reader may consult Beer's text, in particular Chapter 5, Section 2, for these and more technical results in the topological setting. For Euclidean spaces, Rockafellar and Wets report similar facts in Chapter 4.

  • Both \mathop{\mathrm{Li}} A_n and \mathop{\mathrm{Ls}} A_n are closed subsets of X, and \mathop{\mathrm{Li}} A_n \subset \mathop{\mathrm{Ls}} A_n always holds.
  • The upper and lower limits do not distinguish between sets and their closures: \mathop{\mathrm{Li}} A_n = \mathop{\mathrm{Li}} \mathop{\mathrm{cl}}(A_n) and \mathop{\mathrm{Ls}} A_n = \mathop{\mathrm{L s}} \mathop{\mathrm{cl}}(A_n).
  • If A_n := A is a constant sequence, then \mathop{\mathrm{Lim}} A_n = \mathop{\mathrm{cl}} A.
  • If

A_n := \{x_n\}

is a sequence of singletons, then

\mathop{\mathrm{Li}} A_n

and

\mathop{\mathrm{Ls}} A_n

consist of the limit points and cluster points, respectively, of the sequence

\{x_n\}_{n=1}^{\infty} \subset X

.

  • If A_n \subset B_n \subset C_n and B := \mathop{\mathrm{Lim}} A_n = \mathop{\mathrm{Lim}} C_n, then \mathop{\mathrm{Lim}} B_n = B.
  • (Hit and miss criteria) For a closed subset A \subset X, one has
  • A \subset \mathop{\mathrm{Li}} A_n, if and only if for every open set U \subset X with A \cap U \ne \emptyset there exists n_0 such that A_n \cap U \ne \emptyset for all n_0 \leq n,
  • \mathop{\mathrm{Ls}} A_n \subset A, if and only if for every compact set K \subset X with A \cap K = \emptyset there exists n_0 such that A_n \cap K = \emptyset for all n_0 \leq n.
  • If A_1 \subset A_2 \subset A_3 \subset \cdots then the Kuratowski limit exists, and \mathop{\mathrm{Lim}} A_n = \mathop{\mathrm{cl}} \left( \bigcup_{n = 1}^{\infty} A_n \right). Conversely, if A_1 \supset A_2 \supset A_3 \supset \cdots then the Kuratowski limit exists, and \mathop{\mathrm{Lim}} A_n = \bigcap_{n = 1}^{\infty} \mathop{\mathrm{cl}}(A_n).
  • If d_H denotes Hausdorff metric, then d_H(A_n, A) \to 0 implies \mathop{\mathrm{cl}}A = \mathop{\mathrm{Lim}} A_n. However, noncompact closed sets may converge in the sense of Kuratowski while d_H(A_n, \mathop{\mathrm{Lim}} A_n) = +\infty for each n=1,2,\dotsFor an example, consider the sequence of cones in the previous section.
  • Convergence in the sense of Kuratowski is weaker than convergence in the sense of Vietoris but equivalent to convergence in the sense of Fell. If X is compact, then these are all equivalent and agree with convergence in Hausdorff metric.

Kuratowski Continuity of Set-Valued Functions

Let S : X \rightrightarrows Y be a set-valued function between the spaces X and Y; namely, S(x) \subset Y for all x \in X. Denote S^{-1}(y) = \{x \in X : y \in S(x)\}. We can define the operators\begin{align}

\mathop{\mathrm{Li}}_{x' \to x} S(x') :=& \bigcap_{x' \to x} \mathop{\mathrm{Li}} S(x'), \qquad x \in X \\

\mathop{\mathrm{Ls}}_{x' \to x} S(x') :=& \bigcup_{x' \to x} \mathop{\mathrm{Ls}} S(x'), \qquad x \in X\\

\end{align}where x' \to x means convergence in sequences when X is metrizable and convergence in nets otherwise. Then,

  • S is inner semi-continuous at x \in X if S(x) \subset \mathop{\mathrm{Li}}_{x' \to x} S(x');
  • S is outer semi-continuous at x \in X if \mathop{\mathrm{Ls}}_{x' \to x} S(x') \subset S(x).

When S is both inner and outer semi-continuous at x \in X, we say that S is continuous (or continuous in the sense of Kuratowski).

Continuity of set-valued functions is commonly defined in terms of lower- and upper-hemicontinuity popularized by Berge.Rockafellar and Wets write in the Commentary to Chapter 6 of their text: "The terminology of 'inner' and 'outer' semicontinuity, instead of 'lower' and 'upper', has been forced on us by the fact that the prevailing definition of 'upper semicontinuity' in the literature is out of step with developments in set convergence and the scope of applications that must be handled, now that mappings S with unbounded range and even unbounded value sets S(x) are so important... Despite the historical justification, the tide can no longer be turned in the meaning of 'upper semicontinuity', yet the concept of 'continuity' is too crucial for applications to be left in the poorly usable form that rests on such an unfortunately restrictive property [of upper semicontinuity]"; see pages 192-193. Note also that authors differ on whether "semi-continuity" or "hemi-continuity" is the preferred language for Vietoris-Berge continuity concepts. In this sense, a set-valued function is continuous if and only if the function f_S : X \to 2^Y defined by f(x) = S(x) is continuous with respect to the Vietoris hyperspace topology of 2^Y. For set-valued functions with closed values, continuity in the sense of Vietoris-Berge is stronger than continuity in the sense of Kuratowski.

= Examples =

  • The set-valued function B(x,r) = \{y \in X : d(x,y) \leq r \} is continuous X \times [0,+\infty) \rightrightarrows X.
  • Given a function f : X \to [-\infty, +\infty], the superlevel set mapping S_f(x) := \{\lambda \in \mathbb{R} : f(x) \leq \lambda\} is outer semi-continuous at x, if and only if f is lower semi-continuous at x. Similarly, S_f is inner semi-continuous at x, if and only if f is upper semi-continuous at x.

= Properties =

  • If S is continuous at x, then S(x) is closed.
  • S is outer semi-continuous at x, if and only if for every y \notin S(x) there are neighborhoods V \in \mathcal{N}(y) and U \in \mathcal{N}(x) such that U \cap S^{-1}(V) = \emptyset.
  • S is inner semi-continuous at x, if and only if for every y \in S(x) and neighborhood V \in \mathcal{N}(y) there is a neighborhood U \in \mathcal{N}(x) such that V \cap S(x') \ne \emptyset for all x' \in U.
  • S is (globally) outer semi-continuous, if and only if its graph \{(x,y) \in X \times Y : y \in S(x)\} is closed.
  • (Relations to Vietoris-Berge continuity). Suppose S(x) is closed.
  • S is inner semi-continuous at x, if and only if S is lower hemi-continuous at x in the sense of Vietoris-Berge.
  • If S is upper hemi-continuous at x, then S is outer semi-continuous at x. The converse is false in general, but holds when Y is a compact space.
  • If S : \mathbb{R}^n \to \mathbb{R}^mhas a convex graph, then S is inner semi-continuous at each point of the interior of the domain of S. Conversely, given any inner semi-continuous set-valued function S, the convex hull mapping T(x) := \mathop{\mathrm{conv}} S(x) is also inner semi-continuous.

Epi-convergence and Γ-convergence

{{Main|Epi-convergence|Γ-convergence}}

For the metric space

(X, d)

a sequence of functions f_n : X \to [-\infty, +\infty], the epi-limit inferior (or lower epi-limit) is the function \mathop{\mathrm{e}\liminf} f_n defined by the epigraph equation

\mathop{\mathrm{epi}} \left( \mathop{\mathrm{e}\liminf} f_n\right) := \mathop{\mathrm{Ls}} \left(\mathop{\mathrm{epi}} f_n\right),

and similarly the epi-limit superior (or upper epi-limit) is the function \mathop{\mathrm{e}\limsup} f_n defined by the epigraph equation

\mathop{\mathrm{epi}} \left( \mathop{\mathrm{e}\limsup} f_n\right)

:= \mathop{\mathrm{Li}} \left(\mathop{\mathrm{epi}} f_n\right).

Since Kuratowski upper and lower limits are closed sets, it follows that both \mathop{\mathrm{e}\liminf} f_n and \mathop{\mathrm{e}\limsup} f_n are lower semi-continuous functions. Similarly, since \mathop{\mathrm{Li}} \mathop{\mathrm{epi}} f_n \subset

\mathop{\mathrm{Ls}} \mathop{\mathrm{epi}} f_n, it follows that \mathop{\mathrm{e}\liminf} f_n \leq \mathop{\mathrm{e}\liminf} f_n uniformly. These functions agree, if and only if \mathop{\mathrm{Lim}} \mathop{\mathrm{epi}} f_n exists, and the associated function is called the epi-limit of \{f_n\}_{n=1}^{\infty}.

When (X, \tau) is a topological space, epi-convergence of the sequence \{f_n\}_{n=1}^{\infty} is called Γ-convergence. From the perspective of Kuratowski convergence there is no distinction between epi-limits and Γ-limits. The concepts are usually studied separately, because epi-convergence admits special characterizations that rely on the metric space structure of X, which does not hold in topological spaces generally.

See also

Notes

References

  • {{cite book

| last = Beer

| first = Gerald

| title = Topologies on closed and closed convex sets

| series = Mathematics and its Applications

| publisher = Kluwer Academic Publishers Group

| location = Dordrecht

| year = 1993

| pages = xii+340

}}

  • {{cite book |last=Kuratowski |first=Kazimierz |title=Topology. Volumes I and II |series=New edition, revised and augmented. Translated from the French by J. Jaworowski |publisher=Academic Press |location=New York |year=1966 |pages=xx+560}} {{MathSciNet|id=0217751}}
  • {{cite book |last1=Rockafellar |first1=R. Tyrrell |url=https://www.worldcat.org/oclc/883392544 |title=Variational analysis |last2=Wets |first2=Roger J.-B. |date=1998 |isbn=978-3-642-02431-3 |edition= |location=Berlin |oclc=883392544}}

{{Metric spaces}}

Category:Metric geometry