Recall that a regular open set is an open set $U$ for which $U = \operatorname{int} \operatorname{cl} U$.
A fundamental region for a finite group $G\leq \mathcal{O}(V)$ is an open set $F\subseteq V$ such that nonidentity elements transport $F$ to a disjoint set and the closures $\{\operatorname{cl}(TF)\mid T\in G\}$ cover $V$.
In general, fundamental regions are not unique (up to conjugation): If we remove any point from $F$, the closure is not affected by this, so the covering property is still met. Unfortunately, $F\setminus \{f\}$ is not regularly open, which I used in an attempted proof that connected fundamental regions of Coxeter groups are unique up to conjugation.
Since many of the „nice“ fundamental regions are
- connected
- contractible
- convex
(ordered by strength), and we can find fundamental regions for cases 1 and 2 which are not regularly open, my question would be:
Is every convex fundamental region regularly open?
Counterexamples
$\mathbb R^2 \setminus \left( \{0\} \times [0,\infty) \right)$ is a fundamental region for the trivial group $\{\mathrm{id}_{\mathbb R^2}\}$ which is contractible, but not regularly open, since it is a dense proper subset.
Every convex open subset of $\mathbb R^n$ is regularly open. The non-obvious part of the proof is the inclusion $\text{int}(\text{cl}(U)) \subset U$. I'll give the proof in $\mathbb R^2$; it's not too hard to generalize to $\mathbb R^n$.
Consider a point $p = (x,y) \in \text{int}(\text{cl}(U))$. Since open boxes are a basis for the topology on $\mathbb R^n$, there exists $r > 0$ such that $(x-r,x+r) \times (y-r,y+r) \subset \text{cl}(U)$. Slice this box in half horizontally and vertically to obtain four quadrants, each of which is a subset of $\text{cl}(U)$:
Since each of these quadrants is an open set that contains a point of $\text{cl}(U)$, each also contains a point of $U$, say $q_i \in Q_i \cap U$. The line segment $\overline{q_1 q_2}$ is therefore in the convex set $U$, and it crosses the "positive $y$-axis", i.e. it crosses the segment $\{x\} \times (y,y+r)$ at some point contained in $U$, call that point $p_+$. Also, the line segment $\overline{q_3 q_4}$ is in the convex set $U$, and it crosses the "negative $y$-axis", i.e. the segment $\{x\} \times (y-r,y)$, at some point contained in $U$, call that point $p_-$. The whole segment $\overline{p_- p_+}$ is therefore contained in $U$, and that segment contains $p$, proving that $p \in U$.