This post is under construction.
Let \(G\) be a group acting on \(\RR^3\) through isometries.
This note serves to note that the previous post concerning linking is naturally \(G\)-equivariant:
the (non-equivariant) linking graph, which is weaker than the \(G\)-equivariant linking graph, detects the \(G\)-equivariant unweave.
Our construction of the linking graph was manifestly equivariant;
that is, we supplied a function \(\cW \rightarrow \cbr{\mathrm{Graphs \; in \; } \RR^3}\), which happens to be equivariant for the evident \(\GG\)-action on the target space.
By Elmendorf’s theorem, this data is equivalent to a natural transformation
To provide a grading to this graph is equivalent to upgrading this to a natural transformation over \(\pi_0 \Set^{BG}\).
In fact, this is easy;
the map \(|-|:\cbr{G\mathrm{-equivariant\;graphs\;in\;}\RR^3} \rightarrow \pi_0 \Set^{BG}\) is given by taking the underlying set of vertices and associated \(G\)-action, which is evidently natural and makes the required diagram commute.
(Non-equivariant) linking detects the \(G\)-unweave
Theorem 1. Let \(X\) be the space of \(G\)-weaves of component set \(S\) such that \(L(w)\) is trivial, i.e. no pairs of components are linked.
Then, \(X\) is connected.
The first step is already done:
let \(\iota: X_\varepsilon \hookrightarrow X\) denote the subspace of weave realizations such that the radius \(\rho(w)\) is at most \(\varepsilon d(c,c')\) for all components \(c,c'\).
Then, the non-equivariant proof constructed paths from weave relizations in \(X\) to those in \(X_\varepsilon\);these paths preserve centers and normals, hence they are \(G\)-equivariant.
Together, these form a deformation retract of \(X\) onto \(X_\varepsilon\), so \(\iota\) is a homotopy equivalence, and we can just prove that \(X_\varepsilon\) is connected.
The proof that \(X_\varepsilon\) is connected is not too hard:
first note that, by separately orienting the rings within their respective \(\varepsilon \operatorname{min}_{c,c'} d(c,c')\) balls, we have the following:
Lemma 2. There is an equivalence \(X_\varepsilon \simeq \prn{\RRP^2}^n \times \Map(S,\RR^3)^G\).
Having proved this, the first factor is obviously connected, so we just have to prove the same for the second.
This is the key geometric input:
for an orbit \(G/H \simeq \langle s \rangle \subset S\), write \(R := S \backslash \prn{\langle s \rangle}\), so that
Hence it is sufficient to prove that the complement of a discrete set of points in the 3-manifold \(\RR^3_H\) is (path-)connected.
This is true for any manifold of dimension 2 or higher;
indeed, any path in \(\RR^3_H\) can be isotoped to one which doesn’t intersect \(R_H\), by a routine general position argument.
□