Let $G$ be an arbitrary group and we construct the classifying space $BG$ as quotient of $EG$ where the latter one is considered in this discussion to be constructed in natural way as $\Delta$-complex by method from Hatcher's AT book on page 89. That means all $n$-simplices of $EG$ are the ordered $(n + 1)$ tuples $[g_0, ..., g_n]$ of elements of $G$ glued together along faces $[g_0, ..., \hat{g_i},..., g_n]$ in canonical way. For concrete geometrical realization of such formally defined $n$-simplex $[g_0, ..., g_n]$ we identify $g_0,..., g_n$ with $(n+1)$ real points in $ \mathbb{R}^n$ which are affinely independent, which means $g_1-g_0,... g_n-g_0$ are linearly independent. Then we can write $[g_0, ..., g_n]$ as
$$ \{ \theta_0 g_0 + ... \theta_n g_n \ \vert \
\sum_{i=0}^n \theta_i=1 \text{ and } \theta_i \ge 0\} $$
and endow it naturally with Euklidian topology.
The group $G$ acts on $EG$ by left multiplication,
an element $g \in G$ takes the
simplex $[g_0, ..., g_n]$
linearly onto the simplex $[gg_0, ..., gg_n]$ and by definition
$BG= EG/G$.
My goal is to understand locally the topology
of $BG$ better. Is there a standard method known
to construct open subsets in $BG$?
Note that since $BG$ inherits a $\Delta$-complex
structure from $EG$, an subset $O \subset BG$
is open iff the intersection of $U$ with any
simplex of $BG$ is open with respect to the induced standard
Euklidian topology.
It seems naturally for me instead to work with complicated
$BG$ to try to construct open
subsets in $EG$, but assure
that these are $G$-invariant with respect
the described action.
Is there a most "natural, not too exotic" way (ways?) known to construct
such open $G$-invariant subsets $O \in EG$. Moreover can these
be arranged to form a cover of $EG$ and therefore
that their images under $EG \to BG$ to form a cover
of $BG$.
Contain such constructions also some geometric intuition or are these done pure axiomatically?