7
$\begingroup$

I consider a graph $G$ (possibly infinite, but locally finite) embedded in the Euclidean plane $\mathbb{E}^2 \cup \{\infty\}$ such that each local perturbation of the embedding "increases the total length". That is, for any sufficiently small neighborhood $U \subset \mathbb{E}^2$, the total length of $G \cap U$, if not empty, is minimized subject to fixed boundary points $G \cap \partial U$.

So all edges must be straight. If a vertex has degree 3, then it must be the Fermat point of its neighbors. If a vertex has degree 4, the adjacent edges must form two collinear pairs. In general, the outgoing unit vectors along the edges around a vertex must sum up to 0. I consider such embeddings as an analogy to minimal surfaces. Examples include, of course, infinite Steiner trees --- or finite Steiner trees if I use the point at infinity wisely.

I failed to find reasonable literature about such embeddings, hence would like to ask for references. In particular, I wonder if such an embedding may appear as (a nice supgraph of) the graph of some quasi-crystallographic tiling.

$\endgroup$
4
  • 1
    $\begingroup$ Probably you already know this, but it seems relevant to note that the Tutte embedding or barycentric embedding does this except that it does not minimize the total length but the total squared length of the edges. $\endgroup$ Commented Oct 23, 2020 at 11:08
  • $\begingroup$ @TimothyBudd Yes, it's indeed very similar. That's why I believe there must already be some studies. I am only aware of minimal surface theorists that desingularize $\mathbb{R}\times G$ into minimal surfaces (not surprising). $\endgroup$
    – Hao Chen
    Commented Oct 23, 2020 at 11:14
  • $\begingroup$ As written, your definition of local minimization does not make sense. Is U supposed to be a neighborhood of G? Then the intersection of G with the boundary of U will be empty. Maybe you are using some nonstandard notion of a neighborhood? $\endgroup$ Commented Nov 26, 2020 at 17:59
  • $\begingroup$ @MoisheKohan I mean neighborhood of a point. $\endgroup$
    – Hao Chen
    Commented Nov 26, 2020 at 18:23

2 Answers 2

2
$\begingroup$

see https://arxiv.org/pdf/1803.03728.pdf , https://arxiv.org/pdf/1904.00483.pdf and https://arxiv.org/pdf/1902.07872.pdf . Here it is called geodesic nets.

$\endgroup$
1
$\begingroup$

Finally, I find that the works of Ivanov and Tuzhilin seem to be very close to what I'm looking for.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .