Meeting 8
February 11, 1999
Euler Characteristic
Topics:
alternating sums, shelling, shellability, cell complexes, 2-manifolds.
Alternating sums. The Euler characteristic of a
simplicial complex K is the alternating sum of the number of simplices:
(K ) = s0 s1 + s2 : : : + ( 1)d sd ;
where d = dim K and si is the number of i-simplices
in K . It is common to omit the ( 1)-simplex from the
sum. A simplex of even dimension contributes 1 and a
simplex of odd dimension contributes 1, therefore
X
( 1)dim :
(K ) =
A shelling of K is an ordering of the triangles such
that each pre x de nes a triangulation of B2 . K is
shellable if it has a shelling. A shelling constructs
K without every creating pinch points, disconnected
pieces, or holes along the way, see Figure 12. Suppose
27
26
25
28
29
30
;6=2K
As an example consider the complex B = Cl , which
consists of and all its faces. Assuming d = dim we
have
d + 1
d
X
i
( 1) i + 1
(B ) =
i=0
d + 1
d
+1
1
= (1 1)
( 1)
0
= 1:
The boundary complex of B is S = B fg, and its
Euler characteristic is
(S ) = (B ) ( 1)dim
= 1 ( 1)dim :
This is 0 if the dimension of is even and 2 if the
dimension is odd.
24
31 22
32 21
23
19 18
16
20
17
10
15
14
13
4
2
1
7
12 11 8
Figure 12: The numbers specify a shelling of the triangulation.
1 ; 2; : : : ; n is a shelling. When we add i , for i 2, its
intersection with the complex of its predecessors either
consists of a single edge with both vertices, or of two
edges with three vertices, see Cases (b) and (c) in Figure
13. Case (a) occurs only for 1. Cases (d) through (m)
cannot occur in a shelling because the union of either
the rst i 1 or the rst i triangles in not homeomorphic
to B2.
Let Ki be the closure of the set of rst i triangles
in the shelling. At the beginning we have (K1 ) =
1. If i shares one edge with its predecessors then we
e ectively add one triangle, two edges, and one vertex.
If i shares two edges we add one triangle and one edge.
In either case we have (Ki ) = (Ki 1). In words,
the Euler characteristic remains the same during the
entire construction, and therefore (K ) = (Kn ) = 1.
Modulo the existence of a shelling we have thus proved
that every triangulation of the closed disk has Euler
characteristic equal to 1. Although we cannot prove it,
we state a vast generalization of this result.
Theorem 2. If K and L are triangulations of the same
topological space then (K ) = (L).
Shelling. Consider the case where K is a triangulation of the closed disk, B2. We draw K in the plane and
get a triangulated polygon like the one shown in Figure
12. Our goal is to prove (K ) = 1. Since K can be any
triangulation of B2, this amounts to proving that every
triangulation of the closed disk has Euler characteristic
equal to 1. We use the concept of shelling to prove this
claim.
24
13, or c is also a boundary vertex, as in Case (i) of
Figure 13. We may assume the latter. Let ax and
zc be the rst and last edge on the boundary path
from a to c that does not pass through b. We are
done if one of the two triangles next to ax and zc
satis es the requirements for n . Otherwise, both
triangles meet the rest as in Case (i), and we continue the search by substituting xz for ac. The
search cannot continue inde nitely because K has
only nite size. Eventually, we nd a triangle n
that meets the rest in two edges as shown in Case
(c) of Figure 13.
This result is the reason the Euler characteristic can be
considered a property of the topological space, rather
than of a triangulation of that space.
Shellability. To complete the proof that (K ) = 1
for every triangulation K of the closed disk we still need
to nd a shelling.
Lemma 3. Every triangulation of B2 is shellable.
Proof. We construct a shelling backwards. Speci cally, we prove that every triangulation K of n 2
triangles has a triangle n that meets the rest in one
of the two allowable con gurations shown as Cases (b)
and (c) in Figure 13.
(a)
(b)
(c)
(d)
(e)
(f)
z
(g)
By induction we may assume that the complex de ned
by the remaining n 1 triangles has a shelling. We
append n and have a shelling of K .
Cell complexes. The Euler characteristic in not restricted to simplicial complexes and can be computed
as the alternating sum of the number of cells in more
general complexes. If two complexes have homeomorphic underlying spaces we require they have the same
Euler characteristic. For a large class of complexes this
is indeed the case. Without exploring the most general
setting that satis es this requirement we consider nite
cell complexes. Its cells are pairwise disjoint open balls,
the boundary of a cell is the closure minus the cell, and
the boundary of each cell is the union of other cells in
the complex. An example is the dunce cap constructed
from a triangular piece of paper by gluing the three
edges as indicated in Figure 14.
i
x a
(h)
(i)
(j)
(k)
(l)
(m)
a
c
Figure 13: The 13 ways a triangle can intersect with the
complex of its predecessors. Only cases (a), (b), (c) occur
in a shelling.
Figure 14: The dunce cap to the left consists of one 2cell, one edge, and one vertex. Its triangulation to the
right consists of 27 triangles, 39 edges, and 13 vertices.
contains no interior vertex. We consider
the dual graph, whose nodes are the triangles in K
and whose arcs connect two nodes if they share a
common edge. The dual graph is a tree and therefore contains at least two leaves. A leaf triangle is
connected to only one other node, which it meets
in one edge as shown in Case (b) of Figure 13.
Case 2. K contains no leaf. Then K contains at least
one interior vertex. Let ab be an edge on the
boundary and abc the triangle next to ab. Either
abc intersects the rest of K as in Case (c) of Figure
Case 1. K
Let Z be a nite cell complex. For technical reasons
we assume there is a simplicial complex K and a homeomorphism f : j Z j ! j K j such that the image of the
closure of each cell is the underlying space of a subcomplex of K . K triangulates j Z j as well as the closure of
every cell in z 2 Z . Let S B K be subcomplexes
that triangulate the boundary and the closure of z . We
think of B S as a triangulation of the interior, which is
25
the cell itself. Since gluing along the boundary does not
a ect the interior we can pretend that S triangulates a
sphere. The Euler characteristic of BS is therefore
(B
S)
a4
a2
a3
= (B ) (S )
= 1 (1 ( 1)d )
= ( 1)d ;
a1
a4
a3
a2
Figure 16: The polygonal schema of the double torus.
where d is the dimension of z . We see that the contribution of B S to (K ) is the same as the contribution of
z to (Z ). It follows that the Euler characteristics are
the same: (Z ) = (K ). This observation greatly simpli es the hand-computation of the Euler characteristic
for many spaces.
torus is constructed by gluing edges in pairs as indicated by the labels. After gluing we are left with 2g
edges and one vertex. The Euler characteristic is therefore
(Tg ) = 2 2g:
Given a triangulated orientable 2-manifold we can
therefore use the Euler characteristic to compute the
genus and decide the topological type of the 2-manifold.
2-manifolds. A 2-dimensional manifold can be constructed from a piece of paper by gluing edges along its
boundary. As an example consider the torus, T, which
can be constructed from a square by gluing edges in opposite pairs as shown in Figure 15. The square together
Bibliographic notes. The history of the Euler char-
a
b
a1
acteristic is described in an entertaining book by Imre
Lakatos, who studies progress in mathematics from a
philosophical perspective [3]. The nal word on the subject is contained in a seminal paper by Henri Poincare,
who proves that the Euler characteristic is equal to the
alternating sum of Betti numbers [4]. Betti numbers are
ranks of homology groups and will be discussed later in
this course. The Euler characteristic has been developed and applied in a number of directions, see e.g.
[5]. An algorithm for constructing a shelling of a triangulation of the closed disk is described in a paper by
Danaraj and Klee [1]. A good treatment of the classi cation of 2-manifolds and their polygonal schemas can
be found in the text by Peter Giblin [2].
a
b
Figure 15: Edges with the same label are glued so their
arrows agree. After gluing we have two edges and one
vertex.
with its two edges and one vertex forms a cell complex
for the torus, with Euler characteristic
(T)
= 1 2+1
= 0:
[1] G. Danaraj and V. Klee. A representation of 2dimensional pseudomanifolds and its use in the design
of a linear time shelling algorithm. Ann. Discrete Math.
2 (1978), 53{63.
[2] P. J. Giblin. Graphs, Surfaces and Homology. Second
edition, Chapman and Hall, London, 1981.
[3] I. Lakatos. Proofs and Refutations. Cambridge Univ.
Press, Cambridge, England, 1976.
[4] H. Poincare. Complement a l'analysis situs. Rendiconti del Circolo Matematico di Palermo 13 (1899), 285{
343.
[5] Yu. A. Shashkin. The Euler Characteristic. Little
Math. Library, Mir Publ., Moscow, 1989.
The straightforward treatment of the torus can be extended to general 2-manifolds using the complete characterization of 2-manifolds, which was one of the major
achievements in nineteenth century mathematics. The
list of orientable 2-manifolds consists of the 2-sphere,
the torus with one hole, the torus with two holes, etc.
The number of holes is the genus of the 2-manifold. The
torus with g holes can be constructed from its polygonal
schema, which is a regular 4g-gon with edges
a1a2 a1 a2 a3a4 a3 a4 : : : a2g 1a2g a2g 1 a2g ;
where an edge without minus sign is directed in counterclockwise and one with minus is directed in clockwise
order around the 4g-gon, see Figure 16. The g-hole
26
