The upper half-space $H^n$ in $\Bbb R^n$ is the set of those $x\in\Bbb R^n$ such that $x_n\ge 0$.

So I ask if it is true that any not empty and open set $U$ in $H^n$ has interior (in $\Bbb R^n$) not empty. Probably this is a consequence that the interior of open set in a convex set is not empty? Is my this last statement true? So could someone help me, please?

  • 1
    $\begingroup$ The empty set is open and its interior is empty. $\endgroup$ Sep 17, 2020 at 7:02
  • $\begingroup$ Okay, and if $U$ is not empty? $\endgroup$ Sep 17, 2020 at 7:03
  • $\begingroup$ Then it's true, but you assertion that $X\subseteq Y$ convex, $U$ open in $X$ $\Rightarrow$ interior of $U$ in $Y$ not empty, is wrong. For instance, take the $x$-axis as a convex subset of $\mathbb R^2$. Any of its subsets, including open ones, have empty interior as subsets of $\mathbb R^2$. The interior of the convex set should also be non-empty for your assertion to hold. $\endgroup$ Sep 17, 2020 at 7:11

2 Answers 2


$\newcommand{gae}[1]{\newcommand{#1}{\operatorname{#1}}}\gae{cl}\gae{int}$ The topological property you are looking for is that $H^n$ is contained in the closure of its interior. In fact, this property is necessary and sufficient for your condition to hold. Let $X$ be a topological space and $S\subseteq X$. If $\cl_X\int_X S\supseteq S$, then every open set $U$ such that $U\cap S\ne\emptyset$ satisfies $U\cap\int_X S\ne\emptyset$ as well. Since $U\cap \int_X S$ is open in $X$ and $U\cap \int_X S\subseteq U\cap S$, we have that $U\cap \int_X S\subseteq \int_X(U\cap S)$. Vice versa, suppose that $S\setminus \cl_X \int_X S\ne \emptyset$. Then, $S\setminus\cl_X\int_X S$ is a non-empty subset of $S$ which is open in $S$ and which cannot contain open subsets of $X$ (because $(S\setminus \cl_X\int_X S)\cap\int_X S=\emptyset$). Therefore $\int_X(S\setminus\cl_X\int_X S)=\emptyset$.

  • $\begingroup$ Well, I read you wrote. So if $S\subseteq\text{cl}_X\big(\text{int}(S)\big)$ then you state that if $U\cap S\neq\emptyset$ for a open set $U$ in $\Bbb R^n$ then $U\cap\text{int}_X(S)\neq\emptyset$ but I don't understand completely your proof. Could you write it better? $\endgroup$ Sep 17, 2020 at 7:42
  • $\begingroup$ I say that for all topological spaces $X$ and for all $S\subseteq X$, the interior of $S$ is dense in $S$ if and only if every non-empty subset which is open in $S$ has non-empty interior in $X$. $\endgroup$
    – user239203
    Sep 17, 2020 at 7:45
  • $\begingroup$ Okay, perhaps I understand!!! if $S\subseteq\text{cl}_X\big(\text{int}_X(S)\big)$ then any $x\in S$ is an adherent poin of $\text{int}(S)$ so if $U$ is an open set containing $x$ then by definition of adherent point there must necessarly be $U\cap\text{int}(S)\neq\emptyset$, right? $\endgroup$ Sep 17, 2020 at 7:47
  • $\begingroup$ Yes.${}{}{}{}{}$ $\endgroup$
    – user239203
    Sep 17, 2020 at 7:48

By definition of the relative topology, if $U$ is open in $H^n$ it exists an open subset $V$ of $\mathbb R^n$ such that $U = V \cap H^n$. If $U$ is not empty, it exists $x \in U$. Therefore $x \in V$ and it exists an open ball $B(x,r)$ centered on $x$ with $B(x,r) \subseteq V$.

If $x=(x_1, \dots, x_{n-1}, 0)$ then $B(\bar x, r/4) \subseteq B(x,r) \subseteq U $ where $\bar x = (x_1,\dots,x_{n-1}, r/2)$. And if $x=(x_1, \dots, x_{n-1}, x_n)$ with $x_n >0$ then $B(x, \bar r) \subseteq B(x,r) \subseteq U $ where $\bar r = \min(r, x_n/2)$. Proving that the interior of $U$ is not empty.

Here, the main argument is that for an open ball $B(x,r) \subseteq \mathbb R^n$ we have $B(x,r) \cap H^n = B(x,r)$ for $r$ small enough and $x_n >0$.

This is not related to the fact that $H^n$ is convex. For example $L= \{(x,0) \mid x \in \mathbb R\}$ is a convex subset of the plane $\mathbb R^2$. $I= \{(x,0) \mid x \in (0,1)\}$ is an open subset of $L$. However, the interior of $I$ is empty in $\mathbb R^2$.

  • $\begingroup$ Excuse me but I don't fully understand: if $U$ is open in $H^n$ it is not necessarly open in $\Bbb R^n$ so why for any $x\in U$ there exist $r>0$ such that $B(x,r)\subseteq H^n$? Could you explain better, please? $\endgroup$ Sep 17, 2020 at 7:16
  • $\begingroup$ For example what happens if $x\in U\cap\text{Bd}(H^n)$? In this case it looks to me that there not exist $r>0$ such that $B(x,r)\subseteq H^n$!!! $\endgroup$ Sep 17, 2020 at 7:20
  • $\begingroup$ @AntonioMariaDiMauro I added additional elements in my answer. Let me know if this clarify. $\endgroup$ Sep 17, 2020 at 7:20
  • $\begingroup$ Unfortunately what you write don't say to me what happens when $x\in\text{Bd}(H^n)$. $\endgroup$ Sep 17, 2020 at 7:22
  • $\begingroup$ Perhaps you implicitely state that $U\cap\text{int}(H^n)$ is not empty??? $\endgroup$ Sep 17, 2020 at 7:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.