(Originally posted at m.se without answers.)

Let $T$ be a set of triangles in an abstract simplicial complex, with orientation of the triangles chosen such that

$$\partial \left( \sum \limits_{t \in T} t \right) = 0$$

where the boundary operator is as usual, and we consider coefficients in $\mathbb{Z}$. Let $S$ be the support of $T$, i.e. the set of nodes in the complex that appear in some $t \in T$.

Does there necessarily exist a $3$-dimensional geometric realization of $T$, in the following sense:

Is there a polyhedron $P$ with vertices $N$ and all triangular faces $F$, as well as a map $\phi : N \to S$ such that $\phi(F) = T$?

(where $\phi(F) = \bigcup \limits_{(a, b, c) \in F} (\phi(a), \phi(b), \phi(c))$)

I mean "polyhedron" here in the classical sense: a 3d solid with finitely many faces, all of which are "flat," such that only adjacent faces intersect and they do so only at a boundary point or a line. The polyhedron can be non-convex (although I suspect that this question is equivalent if we restrict to the convex case).

  • $\begingroup$ What do you mean by a polyhedron? A geometric realization of the complex $T$ such that two triangles intersect only along a side or a vertex and only if they do intersect in the abstract simplicial complex? Also, what do you mean by "possibly with repeats"? $\endgroup$ – Ivan Izmestiev Jun 6 '17 at 11:16
  • $\begingroup$ @IvanIzmestiev Thanks -- I edited the post with clarifications/more precise language that will hopefully clear up these points. $\endgroup$ – GMB Jun 6 '17 at 14:54
  • $\begingroup$ Any $2D$ homology class can be realized by a map of a compact surface into the space. (This is not true in all dimensions.) Then, if you want a 3D polyhedron, just think of your surface as the boundary of a cube with several holes drilled through the middle. $\endgroup$ – Jim Conant Jun 6 '17 at 16:05
  • 1
    $\begingroup$ A special case of your question is: given an abstract triangulation of a surface, can it be realized as a polyhedral surface in $\mathbb{R}^3$? The answer is "no", although counterexamples are complicated (in genus 6 if I remember correctly). $\endgroup$ – Ivan Izmestiev Jun 6 '17 at 16:24
  • 1
    $\begingroup$ If I understand your question correctly, then counterexamples are discussed on the top of page 4 of this article. $\endgroup$ – Ivan Izmestiev Jun 16 '17 at 6:23

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.