The numbers of edges of 5-polytopes with a given number of vertices
Takuya Kusunoki, Satoshi Murai

TL;DR
This paper characterizes the possible pairs of the number of vertices and edges in 5-dimensional convex polytopes, extending known results from lower dimensions and confirming recent independent findings.
Contribution
It provides a complete characterization of the first two entries of the $f$-vector for 5-polytopes, filling a gap in the combinatorial understanding of higher-dimensional polytopes.
Findings
Characterization of vertex-edge pairs in 5-polytopes.
Extension of Steinitz and Grünbaum's results to 5 dimensions.
Independent confirmation of the characterization by other researchers.
Abstract
A basic combinatorial invariant of a convex polytope is its -vector , where is the number of -dimensional faces of . Steinitz characterized all possible -vectors of -polytopes and Gr\"unbaum characterized the pairs given by the first two entries of the -vectors of -polytopes. In this paper, we characterize the pairs given by the first two entries of the -vectors of -polytopes. The same result was also proved by Pineda-Villavicencio, Ugon and Yost independently.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Combinatorial Mathematics · Point processes and geometric inequalities · Mathematical Inequalities and Applications
The numbers of edges of 5-polytopes
with a given number of vertices
Takuya Kusunoki
Takuya Kusunoki, Department of Pure and Applied Mathematics, Graduate School of Information Science and Technology, Osaka University, Suita, Osaka, 565-0871, Japan
and
Satoshi Murai
Satoshi Murai, Department of Mathematics, Faculty of Education Waseda University 1-6-1 Nishi-Waseda, Shinjuku, Tokyo 169-8050, Japan
Abstract.
A basic combinatorial invariant of a convex polytope is its -vector , where is the number of -dimensional faces of . Steinitz characterized all possible -vectors of -polytopes and Grünbaum characterized the pairs given by the first two entries of the -vectors of -polytopes. In this paper, we characterize the pairs given by the first two entries of the -vectors of -polytopes. The same result was also proved by Pineda-Villavicencio, Ugon and Yost independently.
The second author was partially supported by KAKENHI16K05102.
1. Introduction
The study of -vectors of convex polytopes is one of the central research topic in convex geometry. We call a -dimensional convex polytope a -polytope. For a convex polytope (or a polyhedral complex) , we write for the number of -dimensional faces of . The -vector of a -polytope is the vector . In 1906, Steinitz characterized all possible -vectors of -polytopes (see [Gr, §10.3]). While a characterization of -vectors of -polytopes is a big open problem in convex geometry, for any , the following set was characterized by Grünbaum [Gr], Barnette [Ba] and Barnette–Reay [BR] (see also [BL, Theorem 3.9])
[TABLE]
Moreover, Sjöberg and Ziegler [SZ] recently characterize all possible values of the pairs of flag face numbers of -polytopes. In this paper, we characterize all possible pairs of -polytopes.
Let
[TABLE]
The set was determined by Steinitz in 1906 who shows that
[TABLE]
Note that, by Euler’s relation, this actually determines all possible -vectors of -polytopes. In higher dimensions, it is easy to see that any -polytope satisfies
[TABLE]
Indeed, the first inequality follows since equals to times the sum of degrees of the vertices of and since each vertex of has degree . Grünbaum [Gr, §10.4] proved that the inequality (1) characterizes , with four exceptions. More precisely, he proved the following statement.
Theorem 1.1** (Grünbaum).**
[TABLE]
In dimension , the situation is more complicated. The set is close to the set of integer points satisfying (1), but there are not only a finite list of exceptions but also an infinite family of exceptions. Indeed, our main result is the following.
Theorem 1.2**.**
Let and . Then
[TABLE]
Here denotes the integer part of a rational number . Note that it is not hard to see since a -polytope with must be a simple polytope. Also, was proved in [PUY1] recently.
The following table illustrates the shape of .
In the table, black dots represent points in , white circles and triangles represent points in and respectively. For example, on the line , is presented by a white circle, is presented by a triangle, and the possible numbers of edges are .
Theorem 1.2 was also independently proved by Pineda-Villavicencio, Ugon and Yost [PUY2] by a different method.
It would be interesting to determine for , and more generally to characterize the set \{(f_{i}(P),f_{j}(P)):\mbox{Pd-polytope}\} for any . About the latter problem, Sjöberg and Ziegler [SZ] recently study the case when and .
2. sufficiency
In this section, we prove the sufficiency part of Theorem 1.2. If a polytope is the pyramid over a polytope , then we have
[TABLE]
This simple fact and Theorem 1.1 prove the next lemma.
Lemma 2.1**.**
[TABLE]
Let be a -polytope. The degree of a vertex of is the number of edges of that contain . We say that a vertex is simple if . Let be the vertex set of .
Lemma 2.2**.**
If is a -polytope such that , then has a simple vertex.
Proof.
Observe for any . Since , there must exist a vertex of degree . ∎
Let
[TABLE]
be the right-hand side of Theorem 1.2, and let
[TABLE]
for . We want to prove for all . To prove this, we use truncations. For a -polytope and its vertex , we write for a polytope obtained from by truncating the vertex . If is simple, then
[TABLE]
Lemma 2.3**.**
For with , if then .
Proof.
Since ,
[TABLE]
By Lemma 2.2, for any -polytope with and , we can make a -polytope with
[TABLE]
by truncating a simple vertex from . Since , this implies
[TABLE]
The above inclusion and Lemma 2.1 prove the desired statement. ∎
Now, we prove the main result of this section. For a convex polytope , we write for its dual polytope. In the rest of the paper, if a face of a convex polytope is a simplex, then we call it a simplex face. A face which is not a simplex is called a non-simplex face.
Theorem 2.4**.**
.
Proof.
By Lemma 2.3, it is enough to show that
[TABLE]
Let . By Lemma 2.1, if and . Observe . Then, to prove (2), what we must prove is
[TABLE]
(See also Table 1.) Note that this observation says for .
Let be the cyclic -polytope with vertices. Then (see [Br, §18]). Hence and , and therefore . Let be the polytope obtained from by truncating its vertex. Note that every vertex of is simple. Since a truncation of a simple vertex creates a simplex facet, contains a simplex facet . Let be the polytope obtained from by adding a pyramid over . Then
[TABLE]
and
[TABLE]
Hence . We already see . Then, using truncations of simple vertices and Lemma 2.2, we have . Also, since , by the same argument we have . These complete the proof of the theorem. ∎
3. Necessity
In this section, we prove the necessity part of Theorem 1.2. We first show that any element in is not contained in . We introduce some lemmas which we need. The following fact appears in [Gr, §6.1] (see also [Zi, Problem 6.8]).
Lemma 3.1**.**
There are exactly four combinatorially different -polytopes with facets. They are
- ()
Pyramid over a square pyramid;
- ()
Pyramid over a triangular prism;
- ()
A polytope obtained from a -simplex by truncating its vertex;
- ()
Product of two triangles.
Here are Schlegel diagrams and a list of facets of and .
P_{A}$$P_{B}$$P_{C}$$P_{D}
[TABLE]
Recall that a convex polytope is said to be simplicial if all its proper faces are simplices. A simplicial -sphere is a simplicial complex whose geometric realization is homeomorphic to the -sphere. The boundary complex of a simplicial -polytope is a simplicial -sphere. The next statement easily follows from the Lower Bound Theorem (see [Ka]) and the Upper Bound Theorem (see [St, Corollary II.3.5]) for simplicial spheres.
Lemma 3.2**.**
Let be a simplicial -sphere.
- (i)
.
- (ii)
If , then is neighbourly, that is, every pair of vertices of are connected by an edge.
Proof.
By the Lower Bound Theorem and the Upper Bound Theorem, we have
- •
if , then ;
- •
if , then ;
- •
if , then .
These clearly imply (i). The statement (ii) follows from the fact that if the number of facets of a simplicial -sphere equals to the bound in the Upper Bound Theorem, then it must be neighbourly (see e.g. the proof of [Br, Theorem 18.1]). ∎
We now prove that any element in is not contained in .
Proposition 3.3**.**
If is a -polytope, then .
Proof.
Suppose to the contrary that . We first consider the case when is odd. Then, since , has one vertex having degree and all other vertices have degree . Then has one facet with and all other facets of are simplices. However this implies that the -polytope must be simplicial, which contradicts Lemma 3.2(i).
Next, we consider the case when is even. In this case, , so one of the following two cases occurs:
- (a)
has one facet with and all other facets of are simplices;
- (b)
has two facets and with and all other facets of are simplices.
Since there are no simplicial -polytope with facets by Lemma 2.2(i), the case (a) cannot occur. Also, if the case (b) occurs, then and can have at most one non-simplex facet. However, Lemma 3.1 says that any -polytope with facets have at least two non-simplex facets. ∎
Next, we show that any element of is not contained in . Let . The following result was proved by Pineda-Villavicencio, Ugon and Yost [PUY1, Theorems 6 and 19].
Theorem 3.4**.**
Let be a -polytope.
- (i)
If , then .
- (ii)
If , then
By considering the special case when of the above theorem, we obtain the following.
Corollary 3.5**.**
.
By Proposition 3.3 and Corollary 3.5, to prove Theorem 1.2, we only need to prove . We will prove this in the rest of this paper.
Let be a -polytope. For faces of , we write for the polyhedral complex generated by . Let be a subset of the set of facets of . Then any -face of is contained in at most two facets of . We write
[TABLE]
We often use the following trivial observation: If is the set of non-simplex facets of , then is a simplicial complex.
We say that a -polytope is almost simplicial if all facets of except for one facet are simplices. (We consider that simplicial polytopes are not almost simplicial.) The next lemma can be checked by using a complete list of -polytopes with at most vertices (see [FMM]), but we write its proof for completeness.
Lemma 3.6**.**
Let be a -polytope.
- (i)
Suppose that is almost simplicial and . Then is the pyramid over a triangular bipyramid.
- (ii)
Suppose that is almost simplicial and . Then does not contain a triangular bipyramid as a facet.
- (iii)
Suppose that and has exactly two non-simplex facets and . Then none of and are square pyramids.
Proof.
(i) Let be the unique non-simplex facet of . Clearly, is simplicial and since, for each -face of , there is a unique -face of that contains it other than . Since a -simplex and a triangular bipyramid are the only simplicial -polytopes having at most facets, is a triangular bipyramid. Then, since has facets, must be the pyramid over .
(ii) Let be the unique non-simplex facet of . If is a triangular bipyramid, then by subdividing into two tetrahedra without introducing edges, one obtains a simplicial -sphere with facets. Since is a triangular bipyramid, there are two vertices and of such that and are not connected by an edge in . These vertices are not connected by an edge in by the construction of , which contradicts Lemma 3.2(ii) saying that must be neighbourly.
(iii) Suppose to the contrary that is a square pyramid. We claim that is also a square pyramid. Indeed, since is a simplicial complex, contains exactly one non-simplex facet, and this facet must be a square and equals to . Also,
[TABLE]
Let be a simplicial -sphere obtained from by subdividing the square into two triangles. Then , but since the number of -faces of a simplicial -sphere is even, and therefore . This forces that is a square pyramid.
Let and , where denotes the convex hull of points . Note that is a square. We assume that and are non-edges of . By subdividing each and into two tetrahedra by adding an edge , we can make a simplicial -sphere with . By the construction of , is not an edge of , but this contradicts Lemma 3.2(ii). ∎
We also recall some known results on -vectors of simplicial balls and their boundaries. For a simplicial complex of dimension , its -vector is defined by
[TABLE]
where . A simplicial -ball is a simplicial complex whose geometric carrier is homeomorphic to a -dimensional ball. The following facts are known. See [St, Chapter II and Problem 12].
Lemma 3.7**.**
Let be a simplicial -ball and its -vector. Then
- (i)
* and .*
- (ii)
.
- (iii)
* for all .*
- (iv)
The -vector of the boundary complex of is
[TABLE]
We now complete the proof of Theorem 1.2.
Theorem 3.8**.**
.
Proof.
Suppose to the contrary that there is a -polytope such that and . Let be the vertices of with and let . Since
[TABLE]
and for each , must be one of the following.
- (1)
;
- (2)
;
- (3)
;
- (4)
;
- (5)
;
- (6)
;
- (7)
.
Below we will show a contradiction for each case.
(1) Suppose . Then has a -face with . This must be a simplicial -polytope since all other facets of are simplices, which contradicts Lemma 3.2(i) saying that there are no simplicial -polytopes with facets.
(2) Suppose . has only two non-simplex -faces and . These -faces can have at most one non-simplex -face. By the assumption on , or must have facets. This contradicts Lemma 3.1 saying that any -polytope with facets has at least two non-simplex facets.
(3) Suppose . Let and be the -faces of with and . Then is not simplicial by Lemma 3.2(i) and therefore is also not simplicial. Hence and are almost simplicial and is a -polytope which is not a simplex. Since and , this contradicts Lemma 3.6(i) and (ii).
(4) Suppose . Let and be -faces of with and . Since is a simplicial complex, the polytopes and have at most two non-simplex -faces. By Lemma 3.1, and must be the polytope , and must have square pyramids as its -faces. Thus, the -polytope has tetrahedra and square pyramids as its facets. This implies
[TABLE]
However, this contradicts Theorem 1.1 saying that .
(5) Suppose . Let and be -faces of with and . Since all other -faces of are simplices, and have at most two non-simplex -faces. Then, by Lemma 3.1, must be the polytope , and and have a square pyramid as its facets. By Lemma 3.6(i), and are not almost simplicial. Hence and have exactly two non-simplex facets, but this contradicts Lemma 3.6(iii).
(6) Suppose . Let and be -facets of with and . Since all other -faces of are simplices, each of and can have at most three non-simplex -faces. By Lemma 3.1, and must be the polytope and have exactly two non-simplex -faces. Since is a simplicial complex, one of the following situations must occur:
- (a)
is a simplicial polytope;
- (b)
has exactly two square pyramids as its facets and all other facets are simplices.
However, (a) cannot occur by Lemma 3.2(i) and (b) cannot occur by Lemma 3.6(iii).
(7) Suppose . Let be the -faces of with . Observing that each of must be one of and in Lemma 3.1. It is not hard to see that one of the following situations must occur;
- (a)
All the are ;
- (b)
All the are ;
- (c)
and are . and are .
We first show that (a) cannot occur. We may assume that is a square pyramid for , where . Since has only one square as its -faces, must be this square, and the polyhedral complex generated by the facets of that do not contain is a shellable simplicial -ball by a line shelling [Zi, §8.2]. Let be this ball. Clearly,
[TABLE]
Hence has facets. Also, since each is a pyramid over the square , the faces can be written as
[TABLE]
[TABLE]
[TABLE]
[TABLE]
with . Then it follows that is the join of the -cycle and the -cycle, and its -vector is
[TABLE]
since its entries coincide with the coefficients of the polynomial . Let . Lemmas 3.7 and 3.8 say
[TABLE]
Then it is easy to see . Let be a shelling of . Let
[TABLE]
and , where is the vertex set of . Then
[TABLE]
for all (see [Zi, §8.3]). Since , we have for all . By the definition of a shelling, is a missing face of , that is, is not a face of but any its proper face is a face of . Thus has a missing face of dimension . However, the join of two cycles of length does not have any missing face of dimension , and is the join of the -cycle and the -cycle, a contradiction.
We next prove that (b) cannot occur. Observe that has non-simplex facets. Since is a simplicial complex, each must be a triangular prism. Then it is easy to see that we can write
[TABLE]
where each is a triangular prism with triangles and (we assume that each is an edge of ). Using this formula, one conclude that
[TABLE]
is the disjoint union of two copies of the boundary of a -simplex.
If is not a face of , then, for each with , there is a unique -face of that contains . This implies that, since has only 8 facets other than , either or must be a face of (otherwise has at least 10 facets other than ). We assume that is a face of . Let be the simplex -faces of and assume . Then is a pseudomanifold with
[TABLE]
Let be the number of interior vertices of . By the Lower Bound Theorem for pseudomanifolds with boundary [Fo] (see also [Ta, Theorem 1.2]), must have at least facets. Since only has facets, we have . However, this implies that is either the -simplex or the cone over the boundary of the -simplex, contradicting the fact that has facets.
We finally prove that (C) cannot occur. Since is a simplicial complex, must be a triangular prism and is a square pyramid for all and . It is not hard to see that can be written as
[TABLE]
where is a triangular pyramid with triangles and (we assume that and are edges). A routine computation shows
[TABLE]
is the union of two copies of the boundary of a -simplex intersections in the edge . Then the exactly same argument as in the case (b) works, namely, one case show that either or must be a face of and conclude a contradiction by the Lower Bound Theorem for pseudomanifolds with boundary. ∎
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[Ba] D.W. Barnette, The projection of the f 𝑓 f -vectors of 4 4 4 -polytopes onto the ( E , S ) 𝐸 𝑆 (E,S) -plane, Discrete Math. 10 (1974), 201–209.
- 2[BR] D.W. Barnette and J.R. Reay, Projection of f 𝑓 f -vectors of four-polytopes, J. Combinatorial Theory, Ser. A 15 (1973), 200–209.
- 3[BL] M. Bayer and C.W. Lee, Combinatorial aspects of convex polytopes, in Handbook of Convex Geometry, P. Gruber and J. Wills, eds., North-Holland, Amsterdam, 1993, pp. 485–534.
- 4[Br] A. Brøndsted, An introduction to convex polytopes, Graduate Texts in Mathematics, vol. 90, Springer, New York, 1983.
- 5[Fo] A. Fogelsanger, The generic rigidity of minimal cycles, Ph.D. Thesis, Cornell University, 1988.
- 6[FMM] K. Fukuda, H. Miyata and S. Moriyama, Classification of Oriented Matroids, abirable at http://www-imai.is.s.u-tokyo.ac.jp/ ∼ similar-to \sim hmiyata/oriented_matroids/
- 7[Gr] B. Grünbaum, Convex Polytopes, Wiley Interscience, New York, 1967.
- 8[Ka] G. Kalai, Rigidity and the lower bound theorem. I, Invent. Math. 88 (1987), 125–151.
