0

Consider these two sets:

$$ A\equiv \{x\in X: \forall \xi>0 \text{ }\exists N_{\xi, x} \text{ s.t. } \forall N\geq N_{\xi, x} \text{ } d(p_N, I(x))\leq \xi)\}, $$ $$ B\equiv \bigcap_{\xi>0} \bigcup_{N=1}^\infty\bigcap_{K=N}^\infty \{x\in X: d(p_K, I(x))\leq \xi\}, $$ where

  • $A$ and $B$ are non-empty.
  • $(p_N)_N$ is a sequence of reals taking values in $[0,1]$.
  • $I(x)$ is an interval in $(0,1)$.
  • $d\big(p_N, I(x) \big):= \inf \big\{|p_N - y| : y \in I(x) \big\}$.

Question: Is $A=B$, $A\subseteq B$, $A\supseteq B$?


Some thoughts:

  1. $A$ is equal to: $$ A= \{x\in X: \lim_{N\rightarrow \infty}d(p_N, I(x))=0)\}, $$ where I use the definition of limit.

  2. $B$ is equal to: $$ B=\bigcap_{\xi>0} \liminf_{N\rightarrow \infty}\{x\in X: d(p_N, I(x))\leq \xi \}= \liminf_{N\rightarrow \infty}\{x\in X: d(p_N, I(x))=0 \} $$ where the first equality comes from the definition of $\liminf$ applied to sets (see here) and the second equality comes from my understanding of some comments given to another questions of mine. However, from the comments below, it seems that this second equality is wrong (any clarification would be welcome here).

So, $A$ has the limit inside, while $B$ outside. Is there any obvious relation between $A$ and $B$, e.g., $B$ is a superset/subset of $A$?

Star
  • 414
  • $B=A\ne\bigcup_{N=1}^\infty\bigcap_{K=N}^\infty {x\in X: d(p_K, I(x))= 0}$. – Anne Bauval Apr 18 '24 at 11:14
  • Related (from the same user): https://math.stackexchange.com/questions/4898969 and https://math.stackexchange.com/questions/4880218. – Anne Bauval Apr 18 '24 at 11:16
  • Thanks, Anne. (1) Can you help me to show that $B=A$? – Star Apr 18 '24 at 11:32
  • (2.a) is it correct to write that $$B\equiv \bigcap_{\xi>0}\bigcup_{N=1}^\infty \bigcap_{K=N}^\infty {x\in X: d(p_K, I(x))\leq \xi}= \bigcap_{\xi>0} \liminf_{N\rightarrow \infty} {x\in X: d(p_K, I(x))\leq \xi}\quad ? $$ It is taken from here https://math.stackexchange.com/questions/107931/lim-sup-and-lim-inf-of-sequence-of-sets – Star Apr 18 '24 at 12:32
  • (2.b) in a related question https://math.stackexchange.com/questions/4898969/distance-between-two-sets-different-from-zero one of the users has suggested that $$\bigcap_{\xi>0}{x\in X: d(p_N, I(x))\leq \xi}= {x\in X: d(p_N, I(x))=0}.$$ (See in the comments) (2.c) Hence, my claim comes from combining (2.a) and (2.b). – Star Apr 18 '24 at 12:32
  • @AnneBauval My use of $\liminf$ in 2a comes from https://math.stackexchange.com/questions/107931/lim-sup-and-lim-inf-of-sequence-of-sets I don't understand what is wrong – Star Apr 18 '24 at 12:33
  • (2.a) yes, (2.b) yes, (2.c) no. – Anne Bauval Apr 18 '24 at 16:51

1 Answers1

1

I will limit myself to your question 1 and your thoughts about it. Maybe you should ask your question 2 later in a new post, more clearly and with some attempts.

  • $B=A$ by definition of (infinite) unions and intersections: $$\begin{align}B&=\bigcap_{\xi>0}\bigcup_{N=1}^\infty\{x\in X:\forall K\ge N, d(p_K, I(x))\le\xi\}\\&=\bigcap_{\xi>0}\{x\in X:\exists N\ge1\;\forall K\ge N\;d(p_K, I(x))\le\xi\}\\&=\{x\in X:\forall\xi>0\;\exists N\ge1\;\forall K\ge N\;d(p_K, I(x))\le\xi\}\\&=A. \end{align}$$
  • $B=\bigcap_{\xi>0} \liminf_{N\rightarrow \infty}\{x\in X: d(p_N, I(x))\leq \xi \}$ indeed, by definition of $\liminf$ of a sequence of sets.
  • This $\liminf$ cannot be pulled out. More precisely:
    • the following inclusion holds: $$ B\supseteq\liminf_{N\rightarrow \infty}\{x\in X: d(p_N, I(x))=0 \} $$ because if $x\in\liminf_{N\rightarrow\infty}\{x\in X:d(p_N,I(x))=0\}$, i.e. if $\exists N\;\forall K\ge N\;d(p_K, I(x))=0$, then obviously $\lim_{N\rightarrow \infty}d(p_N, I(x))=0$, i.e. $x\in A$, or equivalently $x\in B$.
    • the reverse inclusion generally doesn't hold. For instance if $I(x)=(1/2,1/4)$ ($\forall x\in X$) and $p_n=\frac12-\frac1n$, then $$ B=A=X\not\subseteq\varnothing=\liminf_{N\rightarrow \infty}\{x\in X: d(p_N, I(x))=0 \} $$
Anne Bauval
  • 49,005
  • Thanks. Regarding the proof that $B$ is not a subset of $\liminf_{N\rightarrow \infty} {x\in : d(p_N, I(x))=0}$: your counterexample considers the case where $\liminf_{N\rightarrow \infty} {x\in : d(p_N, I(x))=0}=\emptyset$. Does the claim hold even if we assume $\liminf_{N\rightarrow \infty} {x\in : d(p_N, I(x))=0}\neq \emptyset$? – Star Apr 19 '24 at 09:47
  • 1
    You will easily find an analogous counterexample with $\liminf_{N\rightarrow \infty} {x\in : d(p_N, I(x))=0}\neq \emptyset$. – Anne Bauval Apr 19 '24 at 10:43
  • You have shown that $B\supseteq \liminf_{N\rightarrow \infty} {x\in X: d(p_N, I(x))=0}$. Does this imply $d_H(B, \liminf_{N\rightarrow \infty} {x\in X: d(p_N, I(x))=0})=0$? – Star Apr 22 '24 at 09:19
  • 1
    No. Use the counterexample you are supposed to have found since the previous comment. – Anne Bauval Apr 22 '24 at 09:56