On Set Convergence II

This post comes as a sequel of On Set Convergence I where we introduced some necessary notions that are useful for studying how sets converge in a topological space. Alongside, we provided two non-topological definitions of convergence, namely the limits inferior and superior. In this post we will delve a bit further into the topological properties of the inner and outer limits of sequences of sets.

We start with an important result which characterises the topological nature of the inner limit:

Proposition 4. Let \{C_n\}_{n} be a sequence of sets in a Hausdorff topological space \left(\mathcal{X},\tau\right). Then

\liminf_n C_n = \left\{ x\left| {\begin{array}{c} \forall V\in\mho\left(x\right),\ \exists N\in \mathcal{N}_\infty\\ \forall n\in N: C_n\cap V\neq \varnothing \end{array} } \right.\right\}

or equivalently:

\liminf_n C_n = \left\{ x\left| {\begin{array}{c} \forall V\in\mho\left(x\right),\ \exists N_0\in \mathbb{N},\\ \forall n\geq N_0: C_n\cap V\neq \varnothing \end{array} }\right. \right\}

Proof. (1). If x\in\liminf_n C_n then we can find a sequence \{x_k\}_{k} such that x_k\to x while x_k\in C_{n_k} and \{n_k\}_{k}\subseteq \mathbb{N} is a strictly increasing sequence of indices. For any V\in\mho\left(x\right) there is a N_0\in\mathbb{N} such that for all i\geq N_0 it is: x_i\in V; but also x_i\in C_{n_i}. Thus, C_{n_i}\cap V\neq \varnothing. Therefore x is in the right-hand side set of equation the first equation in Proposition 4.

(2). For the reverse direction assume that x belongs to the right-hand side set of the first equation in Proposition 4. Then, there is a strictly increasing sequence \{n_k\}_{k} such that for every V\in\mho\left(x\right) we can find a x_k\in C_{n_k}\cap V. Hence, x_k\to x (in the topology \tau). ♣

Or course for every result regarding the inner limit, there is a corresponding one for its complementary object: the outer limit. Therefore, following similar steps with the ones in the proof of Proposition 4 we can show that:

Proposition 5. Let \{C_n\}_{n} be a sequence of sets in a Hausdorff
topological space \left(\mathcal{X},\tau\right). Then

\limsup_n C_n = \left\{ x\left|{\begin{array}{c} \forall V\in\mho\left(x\right),\ \exists N\in \mathcal{N}_\infty^\#,\\ \forall n\in N: C_n\cap V\neq \varnothing \end{array} } \right.\right\}

or equivalenty:

\limsup_n C_n = \left\{ x \left| {\begin{array}{c} \forall V\in\mho\left(x\right),\exists \{n_k\}_{k}\subseteq\mathbb{N}\\ \{n_k\}_{k}\uparrow,\forall k\in \mathbb{N}:\\C_{n_k}\cap V\neq \varnothing \end{array} } \right.\right\}

Instead of arbitrary open sets, when \mathcal{X} is a normed space we may use open balls, i.e. sets of the form \mathcal{B}\left(\varepsilon\right)=\left\{x:\ \|x\|<\varepsilon\right\}. We denote the unit ball of \mathcal{X} by \mathcal{B}:=\mathcal{B}\left(1\right). Then \mathcal{B}\left(\varepsilon\right)=\varepsilon\mathcal{B}. This leads us to the following corollary:

Corollary 1. Let \{C_n\}_{n} be a sequence of sets in a normed space
\left(\mathcal{X},\|\cdot\|\right). Then,

\liminf_n C_n = \left\{ {\begin{array}{c} x|\forall \varepsilon>0,\ \exists N\in \mathcal{N}_\infty,\\ \forall n\in N: x\in C_n+\varepsilon\mathcal{B} \end{array} } \right\}

and

\limsup_n C_n = \left\{ {\begin{array}{c} x|\forall \varepsilon>0,\ \exists N\in \mathcal{N}_\infty^\#,\\ \forall n\in N: x\in C_n+\varepsilon\mathcal{B} \end{array} } \right\}

This corollary yields another equivalent characterization for the inner limit. The predicate “\forall\varepsilon>0” translates into the intersection over all \varepsilon>0, the requirement “\exists N\in\mathcal{N}_\infty^\#” will be written as the union over all N\in\mathcal{N}_\infty and then “\forall n\in N” corresponds to an intersection. This allows us to restate the corollary
using set elementary operations as follows:

Corollary 2. Let \{C_n\}_{n} be a sequence of sets in a normed space \left(\mathcal{X},\|\cdot\|\right). Then,

\liminf_n C_n = \bigcap_{\varepsilon>0}\bigcup_{N\in\mathcal{N}_\infty}\bigcap_{v\in N} C_v + \varepsilon\mathcal{B}

and

\limsup_n C_n = \bigcap_{\varepsilon>0}\bigcup_{N\in\mathcal{N}_\infty^\#}\bigcap_{v\in N} C_v + \varepsilon\mathcal{B}

Before we state and prove the following result (Proposition 7) we need to recall the following description of the closure of a set. Firstly, if \left(\mathcal{X},\tau\right) is a topological space and C\subseteq\mathcal{X}, its closure is defined as:

\mathrm{cl} C := \bigcap\left\{F:\ F\supseteq C,\ F^c \in \tau \right\}

Then, we have the following useful proposition:

Proposition 6. \mathrm{cl} C=\left\{x: \forall V\in \mho\left(x\right),\ V\cap C\neq \varnothing \right\}

We may then show the following result:

Proposition 7. Let \left(\mathcal{X},\tau\right) be a Hausdorff topological space and \{C_n\}_{n} be a sequence of sets in \mathcal{X}. Then,

\liminf_n C_n = \bigcap_{N\in\mathcal{N}_\infty^\#}\mathrm{cl}\bigcup_{n\in N} C_n

Proof. (1). Let x\in\liminf_n C_n and let \Sigma\in\mathcal{N}_\infty^\#. Let W be a neighbourhood of x. There is a N_0\in\mathbb{N} sucht that for all n\geq N_0 such that n\in\Sigma:

W\cap C_n \neq \varnothing

Thus,

x\in \mathrm{cl} \bigcup_{n\in\Sigma}C_n

(2). Assume that x\notin \liminf_n C_n. Then, there is an  open neighbourhood of x, let W\ni x, such that  \Sigma_0:=\left\{n\in\mathbb{N}| W\cap C_n = \varnothing\right\}. Therefore, x\notin \mathrm{cl} \bigcup_{n\in\Sigma_0}C_n. This completes the proof. ♣

Proposition 6 reveals a very important property of the inner limit:

Corollary 3. For any sequence of sets \{C_n\}_{n} in a Hausdorff topological space \left( \mathcal{X},\tau\right) the limit \liminf_n C_n is closed.

Indeed, the inner limit is given as an arbitrary intersection of closed sets \liminf_n C_n = \bigcap_{N\in\mathcal{N}_\infty^\#}\mathrm{cl}\bigcup_{n\in N} C_n which is closed.

The following corollary is another immediate consequence of Proposition 6.

Corollary 4. Let \{C_n\}_{n} be a sequence of sets in \mathcal{X}. Then,

\bigcap_{n=1}^\infty C_n \subseteq \mathrm{cl} \bigcap_{n=1}^\infty C_n \subseteq \liminf_n C_n

The following result is preliminary to what follows. I put it here just because the proof has some interesting steps. It is expedient to know beforehand that the interior limit of a sequence of sets in normed spaces is described by:

\liminf_n C_n = \left\{x\in \mathcal{X} | \lim_{n} d\left(x,C_n\right)=0\right\}

Proposition 8. Let \left(\mathcal{X},\|\cdot\|\right) be a normed space and \{C_n\}_{n} be a sequence of sets in \mathcal{X}. If  x\in\liminf_n C_n then \liminf_{n} d\left(x,C_n\right)=0.

Proof. Let x\in\liminf_{n} C_n. We will construct a sequence \{x_k\}_{k}\subset X and a strictly increasing sequence of indices \{n_k\}_{k}\subseteq \mathbb{N} such that \|x-x_k\|\leq \frac{1}{k} while x_k \in C_{n_k}.

By Proposition 6, if x\in\liminf_n C_n, then for any \Sigma_1\in \mathcal{N}_\infty^\#:

x\in \mathrm{cl}\bigcup_{i\in\Sigma_1}C_i \Rightarrow \exists x_1 \in\bigcup_{i\in\Sigma_1}C_i : \|x-x_1\|<1

Choose n_1\in\Sigma_1 to be the smallest integer such that x\in C_{n_1}. Set \Sigma_2=\left\{\sigma\in\Sigma_1,\ \sigma> n_1\right\}.
Note: \Sigma_2\subseteq \Sigma_1 is cofinal for \Sigma_1. Then,

x\in\mathrm\bigcup_{i\in\Sigma_2}C_i \Rightarrow \exists x_2 \in\bigcup_{i\in\Sigma_2}C_i : \|x-x_2\|<\frac{1}{2}

and take n_2\in\Sigma_2 to the be smallest such that x_2\in C_{n_2}.
Eventually, we construct the aforementioned sequences. We now have that:

\|x_k-x\|<\frac{1}{k},\ x_k \in C_{n_k}\Rightarrow 0\leq d\left(x,C_{n_k}\right)<\frac{1}{k}

Therefore we have proven that \lim_k d\left(x,C_{n_k}\right)=0, i.e. \liminf_n d\left(x,C_n\right)=0. ♣

The previous result is extended to provide a particularly important description of the notion of the inner limit on normed spaces. We use the point-to-set distance function to describe the inner limit of a sequence of sets:

Proposition 9. Let \left(\mathcal{X},\|\cdot\|\right) be a normed space and \{C_n\}_{n} be a sequence of sets in \mathcal{X}. The inner limit of a sequence of sets is:

\liminf_n C_n = \left\{x\in \mathcal{X} | \lim_{n} d\left(x,C_n\right)=0\right\}

Proof. (1).  We firstly need to show that for any x\in\mathcal{X} it is \limsup_n d\left(x,C_n\right)=0. Let us assume that \limsup_n d\left(x,C_n\right)>0, i.e. there exists an increasing sequence of indices \{n_k\}_{k} so that d\left(x,C_{n_k}\right)\to_k a > 0. This suggests that there is a \varepsilon_0>0 such that for all k\in\mathbb{N} one has that d\left(x,C_{n_k}\right)>\varepsilon_0. However, according to Proposition 6,

x\in\mathrm{cl}\bigcup_{k\in\mathbb{N}}C_{n_k}

whereas

d\left(x,\mathrm{cl}\bigcup_{k\in\mathbb{N}}C_{n_k}\right)\geq\varepsilon_0

which contradicts our initial assumption. Hence, \limsup_n d\left(x,C_n\right)=0, i.e. \lim_n d\left(x,C_n\right)=0 and this way we have proven that x is in the right-hand side set.

(2). Conversely, assume that x in the right-hand side of the equation in the statement of Proposition 8, that is \lim_n d\left(x,C_n\right)=0. For any \varepsilon>0; we can find n_0\in\mathbb{N} such that d\left(x,C_{n}\right)\leq \frac{\varepsilon}{2} for all n\geq n_0. By definition, we have that

d\left(x,C_{n}\right)=\inf\left\{\|x-y\|,\ y\in C_{n}\right\},

thus we can find a y_n\in C_{n} such that

\|y_n-x\|<d\left(x,C_{n}\right)+\frac{\varepsilon}{2}=\varepsilon

That is:

\exists\ y_n\in C_{n}:\ \|y_n-x\|<\varepsilon

Therefore, x\in C_{n} + \varepsilon \mathcal{B} from which it follows that x\in\liminf_n C_n (see Corollary 1). ♣

Note. For a sequence \{a_n\}_{n}, in order to show that \liminf_n a_n=a, it suffices to find a subsequence of it that converges to a, i.e. it suffices to determine a strictly increasing sequence \{n_k\}_{k}\subseteq \mathbb{N} such that a_{n_k} \to_k a, i.e. \lim_k a_{n_k}=a. Using this fact we can prove the following:

Proposition 10.  Let \left(\mathcal{X},\|\cdot\|\right) be a normed space and \{C_n\}_{n} be a sequence of sets in \mathcal{X}. The outer limit of a sequence of sets is:

\limsup_n C_n = \left\{x\in \mathcal{X} | \liminf_{n} d\left(x,C_n\right)=0\right\}

Advertisements

One response

  1. […] Where in this, third and final, post on the topic of set convergence we introduce the notion of Painleve-Kuratowski convergence of sequences of sets in topological spaces and we discuss about the unique properties of this notion of converges. We see that the K-limit of a constant sequence may not be equal to the elements of that sequence and many other interesting properties. * Read first the previous posts: On Set Convergence I and II. […]

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

mathbabe

Exploring and venting about quantitative issues

Look at the corners!

The math blog of Dmitry Ostrovsky

The Unapologetic Mathematician

Mathematics for the interested outsider

Almost Sure

A random mathematical blog

Mathematix

Mathematix is the mathematician of the village of Asterix

%d bloggers like this: