# 'Infinitesimal' elements of a topological group

Let $G$ be a topological group, and let $M$ be the intersection of all conjugacy-invariant neighbourhoods of the identity in $G$ (in other words, the set of elements that can be taken arbitarily close to the identity by conjugation). Has this set $M$ been studied/used anywhere? I don't have a precise question here, this is more of a general reference request.

I am particularly interested in the case that $G$ is totally disconnected and locally compact (t.d.l.c.) and in understanding $M \cap U$ where $U$ is an open compact subgroup, but I'd be interested to see what is known for connected groups as well. One can show that $M$ is non-trivial if $G$ is t.d.l.c., compactly generated and has no non-trivial compact normal subgroups, but I don't have any sense of the structure of $M$ in this case, either topologically or algebraically. There has been some work done on contraction subgroups, that is, sets of elements that are sent towards the identity by repeatedly applying a fixed endomorphism of $G$, but as far as I know, $M$ can be non-trivial even if all the (inner) automorphisms have trivial contraction groups.

Perhaps there is some representation-theoretic interpretation? Certainly if $G$ is linear, then $M$ consists of unipotent elements, though I can't see what the conditions would be for $M$ to contain all unipotent elements. What role might $M$ have in infinite-dimensional unitary representations?

-
The more general concept of elements of G that are infinitesimally close to a given subgroup H after conjugation is the key concept that underlies the Mautner phenomenon (ergodicity in one subgroup H implying ergodicity in a larger subgroup), see e.g. projecteuclid.org/DPubS/Repository/1.0/… –  Terry Tao Mar 23 '12 at 5:19
Actually, a later argument of Margulis (found for instance in the proof of Proposition 3.1 in arxiv.org/pdf/math/0603483.pdf ) makes this point a bit more clearly than the older paper of Moore I linked to previously. –  Terry Tao Mar 23 '12 at 5:24
Thinking about it, I do want to use this more general phenomenon of a conjugacy class approaching a subgroup. Essentially I want to know which (compact) subgroups are 'isolated' in the sense that every conjugacy class that approaches the subgroup actually intersects the subgroup. –  Colin Reid Mar 23 '12 at 12:31