Infimum of two sets

870 Views Asked by At

$\newcommand{\card}{\operatorname{card}}$

Let $A$ be a nonempty countable set of real numbers, and $0< a\leq b$. Is the following true:

$$ \tag{*} \inf_{x\in \mathbb R} \card(A\cap [x,x+a]) \geq \inf_{x\in \mathbb R} \card(A\cap [x,x+b]) $$

where $\card$ means number of elements in the set.

As far as I know, since $a\leq b$ then $$A\cap [x,x+a] \; \subseteq \; A\cap [x,x+b] $$

so $$ \card(A\cap [x,x+a]) \; \leq\; \card(A\cap [x,x+b]) $$ for all $x\in \mathbb R$. How I can complete the proof (if the result is correct)?

(I know that if we have two sets $B\subseteq C$ then $\inf B \geq\inf C$. But how I can use this in terms of cardinality of sets?)

2

There are 2 best solutions below

2
On

The inequality displayed in ($*$) is incorrect.

Consider $A=\mathbb{Z}$, $a=\frac{1}{2}$, $b=\frac{3}{2}$. An interval of length $\frac{1}{2}$ contains at most one integer, so for every $x\in\mathbb{R}$, $A\cap[x,x+a]$ is either empty, or has cardinality $1$. In particular, since there are $x$ for which the intersection is empty, then it follows that $$\inf_{x\in\mathbb{R}}\mathrm{card}(A\cap [x,x+a]) = 0.$$ On the other hand, an interval of length $\frac{3}{2}$ contains always at least one integer, and sometimes two; so $$\inf_{x\in\mathbb{R}}\mathrm{card}(A\cap[x,x+b]) = 1.$$ Therefore $$\inf_{x\in\mathbb{R}}\mathrm{card}(A\cap [x,x+a]) = 0 \not\geq 1 = \inf_{x\in\mathbb{R}}\mathrm{card}(A\cap[x,x+b]).$$

On the other hand, if you meant whether $$\inf_{x\in\mathbb{R}}\mathrm{card}(A\cap[x,x+a])\leq\inf_{x\in\mathbb{R}}\mathrm{card}(A\cap [x,x+b]),$$ then that inequality does hold.

Since $A$ is countable, the infimum on the left is either infinite, or a natural number.

If it is infinite, then $A\cap[x,x+a]$ is infinite for every $x$, and therefore $A\cap[ x,x+b]$ is infinite for every $b$, so both infima are infinite, hence equal.

Now suppose that the infimum on the left is finite, equal to $n$. For every $x$, $A\cap [x,x+a]\subseteq A\cap [x,x+b]$, so $\mathrm{card}(A\cap [x,x+a])\leq\mathrm{card}(A\cap [x,x+b])$. Since the infimum is less than or equal to every element of the set, we have $$n\leq \mathrm{card}(A\cap [x,x+a])\leq\mathrm{card}(A\cap [x,x+b]),$$ hence $n$ is a lower bound for the set $$\{\mathrm{card}(A\cap[x,x+b])\mid b\in\mathbb{R}\},$$ and hence we have $$n\leq \inf_{x\in \mathbb{R}}\mathrm{card}(A\cap [x,x+b]).$$ Thus, the desired inequality holds.

1
On

$\newcommand{\card}{\operatorname{card}}$The correct is $$ \inf_{x \in \mathbb{R}} \card(A\cap [x,x+a]) \le \inf_{x \in \mathbb{R}}\card(A\cap [x,x+b]).$$ By definition of $\inf$ there exists $x_n \in \mathbb{R}$ such that $$\card(A\cap [x_n,x_n+b]) \le \inf_{x \in \mathbb{R}}\card(A\cap [x,x+b]) + 1/n.$$ Hence, $$ \inf_{x \in \mathbb{R}} \card(A\cap [x,x+a]) \le \card(A\cap [x_n,x_n+a]) \le \card(A \cap [x_n,x_n+b]) $$ $$\le \inf_{x \in \mathbb{R}}\card(A\cap [x,x+b]) +1/n.$$ Do $n \rightarrow \infty.$