Simplicial Complexes: Simplices

Meeting 6
February 4, 1999
Simplicial Complexes
Topics:
simplices, simplicial complexes, abstract simplicial complexes, geometric realizations, nerves.
Simplices. Points, edges, triangles, and tetrahedra
. If dim = ` then is called an `-face. = ;

and = are improper faces and all others are proper
faces of . The number of `-faces of is equal to the
number of ways we can choose ` + 1 from k + 1 points,
which is

(k + 1)! :
k+1 =
(` + 1)!(k `)!
`+1
The total number of faces is
k k + 1
X
= 2k+1 :
`
+
1
`= 1
are low-dimensional examples of simplices. We use combinations of points to dene simplices in general dimensions. Let S be a nite set in d-dimensional Euclidean
space denoted as Rd . An ane
P combination
P of the
points pi 2 S is a point x = i pi with i = 1.
The ane hull, a S , is the set of all ane combinations. Equivalently, it is the intersection of all hyperplanes that contain S . The points in S are anely
independent, of a. i., if none is the ane combination
of the other points in S . For example, the ane hull of
three a. i. points is a plane, that of two a. i. points is
a line, and the ane hull of a single point is the point
itself.
A convex combination is an ane combination with
non-negative coecients: i 0 for all pi 2 S . The
convex hull, conv S , is the set of all convex combinations. Equivalently, it is the intersection of all halfspaces that contain S . A simplex is the convex hull of
a set of a. i. points. If S Rd is a set of k + 1 a. i.
points then the dimension of the simplex = conv S
is dim = k and is called a k-simplex. The largest
number of a. i. points in Rd is d + 1, and we have simplices of dimension 1 through d. Figure 1 shows the
ve types of simplices in R3. The points in any sub-
Simplicial complexes. A simplicial complex is a nite collection K of simplices such that
(i) 2 K and implies 2 K , and

(ii) ; 2 K implies \ ; .
Note that ; is a face of every simplex and thus belongs
to K by condition (i). Condition (ii) therefore allows
for the possibility that and be disjoint. Figure 2
shows three sets of simplices that violate one of the
two conditions and therefore do not form complexes.
A subcomplex is a subset that is a simplicial complex
0
-1
Figure 1: The ( 1)-simplex is the empty set, a 0-simplex

is a point or vertex, a 1-simplex is an edge, a 2-simplex is
a triangle, and a 3-simplex is a tetrahderon.
Figure 2: To the left we are missing an edge and two

vertices. In the middle the triangles meet along a segment
that is not an edge of either triangle. To the right the
edge crosses the triangle at an interior point.
set T S are a. i., so the convex hull of T is again a

simplex. Specically, = conv T is the subset of points
x 2 with i = 0 whenever pi 2 S is not in T . The
simplex is a face of , and we denote this relation by
itself. Observe that every subset of a simplicial complex

satises condition (ii). To enforce condition (i) we add
18
faces and simplices to the subset. Formally, the closure

of a subset L K is the smallest subcomplex that
contains L:
Cl L = f 2 K j 2 Lg:
A particular subcomplex is the i-skeleton that consists
of all simplices 2 K whose dimension is i or less. The
vertex set is
Vert K = f 2 K j dim = 0g;
it is the same as the 0-skeleton except it does not contain the ( 1)-simplex. The dimension of K is the
largest dimension of any simplex: dim K = maxfdim j
2 K g. If k = dim K then K is a k-complex and it is
pure if every simplex is a face of a k-simplex. The underlying space is the set of points covered by simplices:
j K j = S K = S2K . A polyhedron is the underlying
space of a simplicial complex. Sometimes it is useful
concepts of face, subcomplex, closure, star, link extend

straightforwardly from geometric to abstract simplicial
complexes.
The set system together with the inclusion relation
forms a partially ordered set, or poset, denoted as
(A; ). Posets are commonly drawn using Hasse diagrams where sets are nodes, smaller sets are below
larger sets, and inclusions are edges, see Figure 4. Inclusions implied by others are usually not drawn. To
Figure 4: From left to right the poset of the empty set, a

vertex, an edge, a triangle, a tetrahedron.
draw the Hasse diagram of a k-simplex we draw the

Hasse diagrams for two (k 1)-simplices. One is the
diagram of a (k 1)-face of and the other is the
diagram for the star of the vertex u 2 . Finally,
we connect every simplex in the star of u with the
simplex fug in the closure of . An abstract simplicial complex A is a subsystem of the power set of
Vert A. We can therefore think of it as a subcomplex of
an n-simplex, where n + 1 = card Vert A. This view is
expressed in the picture of an abstract simplicial complex shown as Figure 5.
Figure 3: Star and link of a vertex. To the left the solid
edges, and shaded triangles belong to the star of the solid

vertex. To the right the solid edges and vertices belong to
the link of the hollow vertex.
to consider substructures of a simplicial complex. The

star of a simplex consists of all simplices that contain
, and the link consists of all faces of simplices in the
star that do not intersect :
St = f 2 K j g;
Lk = f 2 Cl St j \ = ;g;
see Figure 3 for examples. The star is generally not
closed, but the link is always a simplicial complex.
Vert A
A
0
Figure 5: The onion is the power set of Vert A. The area

below the waterline is an abstract simplicial complex.
Abstract simplicial complex. If we replace each

simplex in a simplicial complex by the set of its vertices, we get a system of subsets of the vertex set. In
doing so we throw away the geometry of the simplices,
which allows us to focus on the combinatorial structure. Formally, a nite system A of nite sets is an
abstract simplicial complex if 2 A and implies
2 A. This requirement is similar to condition (i) for
geometric simplicial complexes. A set 2 A is an (abstract) simplex and its dimension S
is dim S= card 1.
The vertex set of A is Vert A = A = 2A . The
Geometric realization. We can think of an abstract
simplicial complex as an abstract version of a geometric

simplicial complex. To formalize this idea we dene a
geometric realization of an abstract simplicial complex
A as a simplicial complex K together with a bijection
' : Vert A ! Vert K such that 2 A i conv '() 2 K .
Conversely, A is called an abstraction of K .
Given A, we can ask for the lowest dimension that
19
region to its generator: '(Va ) = a. This function denes a geometric realization of Nrv C :
allows a geometric realization. For example, graphs are

1-dimensional abstract simplicial complexes and can always be realized in R3. Two dimensions are sometimes
but not always sucient. This result generalized to
k-dimensional abstract simplicial complexes: they can
always be realized in R2k+1 and sometimes R2k does not
suce. To prove the suciency of the claim we show
that the k-skeleton of every n-simplex can be realized in
R2k+1. Map the n + 1 vertices to points in general position in R2k+1. Specically, we require that any 2k + 2
of the points are a. i. Two simplices and of the kskeleton have a total of at most 2(k +1) vertices, which
are therefore a. i. In other words, and are faces of a
common simplex of dimension at most 2k +1. It follows
that \ is a common face of both.
D = fconv '() j 2 Nrv C g:

This is the Delaunay triangulation of S . What happens if the points in S are not in general position? If
k + 1 4 Voronoi regions have a non-empty common
intersection then Nrv C contains the corresponding abstract k-simplex. So instead of making a choice among
the possible triangulations of the (k +1)-gon, the nerve
takes all possible triangulations together and interprets
them as subcomplexes of a k-simplex. The disadvantage of this method is of course the fact that Nrv C can
no longer be realized in R2.
Theorem 1. Every k-dimensional abstract simplicial

complex has a geometric realization in R2k+1.
Bibliographic notes. During the rst half of the
twentieth century, combinatorial topology was a ourishing eld of Mathematics. We refer to Paul Alexandrov [1] as a comprehensive text originally published as
a series of three books. This text roughly coincides with
a fundamental reorganization of the eld triggered by a
variety of technical results in topology. One of the successors of combinatorial topology is modern algebraic
topology where the emphasis shifts from combinatorial
to algebraic structures. We refer to James Munkres [5]
for an introductory text in that area.
We proved that every k-complex can be geometrically realized in R2k+1. Examples of k-complexes that
require 2k +1 dimensions are provided by Flores [2] and
independently by van Kampen [3]. One such example
is the k-skeleton of the (2k + 2)-simplex. For k = 1
this is the complete graph of ve vertices, which is one
of the two obstructions of graph planarity identied by
Kuratowski [4].
Nerves. A convenient way to construct abstract sim-
plicial complexes uses arbitrary nite sets. The nerve

of such a set C is the system of subsets with non-empty
intersection:
\
Nrv C = fX C j X 6= ;g:
T
T
If Y X then X Y . Hence X 2 Nrv C implies Y 2 Nrv C , which shows that the nerve is an
abstract simplicial complex. Consider for example the
case where C is a covering of some geometric space, such
as in Figure 6. Every set in the covering corresponds
[1] P. S. Alexandrov. Combinatorial Topology. Dover,

New York 1956.
[2] A. Flores. U ber n-dimensionale Komplexe die in
R2n+1 selbstverschlungen sind. Ergeb. Math. Koll. 6
(1933/34), 4-7.
[3] E. R. van Kampen. Komplexe in euklidischen Raumen.
Abh. Math. Sem. Univ. Hamburg 9 (1933), 72{78.
[4] K. Kuratowski. Sur le probleme des courbes en topologie. Fund. Math. 15 (1930), 271{283.
[5] J. R. Munkres. Elements of Algebraic Topology. Addison-Wesley, Redwood City, California, 1984.
Figure 6: A covering with eight sets to the left and its

nerve to the right. The sets meet in triplets but not in
quadruplets, which implies that the nerve is 2-dimensional.
to a vertex, and k + 1 sets with non-empty intersection
dene a k-simplex.
We have seen an example of such a construction earlier. The Voronoi regions of a nite set S R2 dene
a covering C = fVa j a 2 S g of the plane. Assuming
general position the Voronoi regions meet in pairs and
in triplets, but not in quadruplets. The nerve therefore
consists only of abstract vertices, edges, and triangles.
Consider the function ' : C ! R2 that maps a Voronoi
20

Simplicial Complexes: Simplices

Uploaded by

Copyright:

Available Formats

Simplicial Complexes: Simplices

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Simplicial Complexes: Simplices

Uploaded by

Copyright:

Available Formats

Meeting 6

simplices, simplicial complexes, abstract simplicial complexes, geometric realizations, nerves.

Simplices. Points, edges, triangles, and tetrahedra

  . If dim  = ` then  is called an `-face.  = ;

Simplicial complexes. A simplicial complex is a nite collection K of simplices such that

(i)  2 K and    implies  2 K , and

Figure 1: The ( 1)-simplex is the empty set, a 0-simplex

Figure 2: To the left we are missing an edge and two

set T  S are a. i., so the convex hull of T is again a

itself. Observe that every subset of a simplicial complex

faces and simplices to the subset. Formally, the closure

concepts of face, subcomplex, closure, star, link extend

Figure 4: From left to right the poset of the empty set, a

draw the Hasse diagram of a k-simplex we draw the

Figure 3: Star and link of a vertex. To the left the solid

edges, and shaded triangles belong to the star of the solid

to consider substructures of a simplicial complex. The

Figure 5: The onion is the power set of Vert A. The area

Abstract simplicial complex. If we replace each

Geometric realization. We can think of an abstract

simplicial complex as an abstract version of a geometric

allows a geometric realization. For example, graphs are

D = fconv '( ) j 2 Nrv C g:

Theorem 1. Every k-dimensional abstract simplicial

Bibliographic notes. During the rst half of the

Nerves. A convenient way to construct abstract sim-

plicial complexes uses arbitrary nite sets. The nerve

[1] P. S. Alexandrov. Combinatorial Topology. Dover,

Figure 6: A covering with eight sets to the left and its

You might also like

. If dim = ` then is called an `-face. = ;

(i) 2 K and implies 2 K , and

set T S are a. i., so the convex hull of T is again a

D = fconv '() j 2 Nrv C g: