14

What is minimal triangulation of Klein bottle? А triangulation is a subdivision of a geometric object into simplices. Minimal in sense of vertex count.

So, I know that minimal count of vertices in the shortest triangulation must be greater than $7$, because the shortest triangulation of torus consists of $7$ vertices and Euler characteristic is equal to $0$.

I would be cool if you can show me the picture.

Gleb
  • 1,055
  • 15
    The minimal number is 8. Section 4 of this paper has a proof of that. Section 5 of same paper derive all six distinct 8-vertex triangulations of the Klein bottle and has picture for them. – achille hui Jul 27 '13 at 13:02
  • Thanks, it`s very good paper. – Gleb Jul 27 '13 at 13:21
  • 2
    These papers might be interesting for you: https://arxiv.org/pdf/math/0407008v2.pdf , http://www.sciencedirect.com/science/article/pii/S0095895697999998 – taper Dec 22 '16 at 14:22
  • 1
    Sadly, OP did not give a correct definition of a triangulation here (instead he referred to the correct definition in a comment to an answer). – Moishe Kohan Nov 02 '23 at 19:11

3 Answers3

1

The standard definition of a triangulation of a space $X$ consists of a simplicial complex $T$ and a homeomorphism $f: |T| \to X$. (n.b. As pointed out by M.K. in the comments, the OP gave an incorrect definition in the question, but referred to the correct definition in a reply to an answer).

The other two answers give CW decompositions of the Klein bottle $K$, but since they are not simplicial complexes, they are not triangulations of $K$ in the standard sense. A minimal CW decomposition of the Klein bottle requires one $0$-cell, two $1$-cells, and one $2$-cell. See here.

The paper Vertex-Minimal Simplicial Immersions of the Klein Bottle in Three Space linked in the first comment on OP shows that every triangulation of $K$ has at least $8$ vertices. I will summarize the argument here.

To get a lower bound they use Euler's formula for polyhedra

$$\chi(X) = V - E + F$$

where $\chi(X)$ is the Euler characteristic of $X$, and $V, E, F$ are the numbers of vertices, edges, and faces respectively. Some simple combinatorics and algebra shows that

$$V \geq \frac{7 + \sqrt{49 - 24 \chi(X)}}{2}.$$

Since $\chi(K)= 0$ (see, e.g., here) then one immediately obtains $V \geq 7$.

However, there are in fact no $7$-vertex triangulations of $K$. To see this, if $V=7$ then $E=21$ by Euler's formula, but since $21$ is the number of edges in the complete graph $K_7$ on $7$ vertices, therefore every vertex is connected to every other vertex by an edge. By a result of Franklin, every graph embeddable in the Klein bottle is $6$-colorable, but $K_7$ has chromatic number $7$, a contradiction. Therefore $V \geq 8$.

The authors go on to classify all combinatorially distinct $8$-vertex triangulations of $K$, of which there are six. Here they are:

Triangulations of Klein Bottle

0

The Klein bottle can be seen as the square $I^2$ with the boundaries identified in a specific way. Thus some triangulations of the square induce a triangulation of the Klein bottle. In particular you have a triangulation with exactly two faces, three edges and one vertex induced by the triangulation of the square obtained by cutting along the diagonal.

  • 2
    I think that for a triangulation it is required that the closures of all the triangles are mapped injectively to the triangulated space (here the Klein bottle). Your simple idea fails, because all the vertices of the two triangles are mapped on the same point on the Klein bottle. – Jyrki Lahtonen Jul 27 '13 at 12:32
  • Unfortunately, this partition isnt triangulation, because we have a edge, which incidence only one vertex. So, it isnt simlices. – Gleb Jul 27 '13 at 12:35
  • @JyrkiLahtonen Yes, if you require this of a triangulation then my triangulation doesn't work. I still think that working on the square is the best way to obtain something, though. – Daniel Robert-Nicoud Jul 27 '13 at 12:35
  • 1
    IOW, this is a fine structure as a CW-complex but not as a simplicial complex. – Jyrki Lahtonen Jul 27 '13 at 12:41
  • So this 1-vertex object doesn't work, but the minimal has 8 vertices? Surely there has to be something inbetween 2, and 7 which is equal to a number of vertices of some klein bottle triangulation... This 1 vertex example is so close to working I don't see why we couldn't achieve, say, one with 7 vertices fixing the edge issue. – Snared May 28 '23 at 04:13
  • @Snared It is proven in the paper linked in the first comment on OP that a minimal triangulation of the Klein bottle has 8 vertices. – pyridoxal_trigeminus Aug 14 '25 at 21:33
0

Here is a obvious triangulation into two triangles:

Gluing diagram for Klein Bottle

Glue the red arrows together, and the black arrows together to get a Klein Bottle from the gray square.

The other way round, we can cut at the arrows and also cut at the yellow line to get 2 triangles from a Klein Bottle.

However, you could also glue a single triangle to a Klein Bottle, if the condition is dropped that complete edges shall be glued to complete edges.

FYI, I make the plot here (Desmos)

  • Similar to the other answer, this is a CW structure (in fact, a $\Delta$-complex structure in Hatcher's language), but not a triangulation (in the standard sense of being a simplicial complex). See also this post. – pyridoxal_trigeminus Aug 14 '25 at 21:35