Topological Methods in Combinatorics

László Lovász
Fall 2013

Contents
1 Preliminaries 1
1.1 Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Simplicial complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Homotopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Combinatorial homotopy theory 3

2.1 Contractibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 The Nerve Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Topological connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3 Brouwer’s fixed point theorem 7

3.1 Evasive graph properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

4 The Borsuk–Ulam Theorem 9

4.1 Simplicial complexes associated with graphs . . . . . . . . . . . . . . . . . . . 10
4.2 Topological connectivity and chromatic number . . . . . . . . . . . . . . . . . 12
4.3 The Ham Sandwich Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.4 The Necklace problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1 Preliminaries
1.1 Notation
The r-dimensional unit ball is denoted by B r . Specifically, B 0 is a single point. Its boundary,
the (r − 1)-dimensional unit sphere is denoted by S r−1 . If two topological spaces T1 , T2 are
homeomorphic, then we write T1 ∼ = T2 .

1.2 Simplicial complexes

A simplicial complex K is a finite collection of nonempty finite sets such that X ∈ K, ∅ = 6
Y ⊆ X implies Y ∈ K. The union of all members of K is denoted by V (K). The elements of
V (K) are called the vertices of K, the elements of K are called the simplices of K.
The dimension of a simplex S ∈ K is dim(S) = |S| − 1. The dimension of K is the
maximum dimension of any simplex in K. The k-dimensional skeleton of a simplicial complex
K is the simplicial complex K|k = {S ∈ K : |X| ≤ k + 1}.
Let n = |V (K)|. Embed V (K) in Rn−1 so that the n points representing the vertices
are not contained in one hyperplane. For each simplex S, take the convex hull conv(S) of
the elements in S, and take the union of these simplices. This set G(K) ⊂ Rn−1 is called a
geometric realization of K. The sets conv(S) (S ∈ K) are the faces of G(K). By the “general
position” assumption on the vertices, it follows that the faces intersect each other in their

1
common subface only: conv(S1 )∩conv(S2 ) = conv(S1 ∩S2 ). It is also clear that all geometric
realizations of the same simplicial complex are homeomorphic.
Let K1 and K2 be two simplicial complexes and let f : V (K1 ) → V (K2 ) be a mapping.
We say that f is simplicial if f (S) ∈ K2 for every simplex S ∈ K1 . Every simplicial mapping
can be extended linearly to get a continuous mapping fb : G(K1 ) → G(K2 ).
The baricentric subdivision B(K) of a simplicial complex K is obtained as follows. We
create a new vertex vS for every non-empty simplex S (the “baricenter” of S), and define
vS1 , . . . , vSk to be a simplex for every S1 ⊂ · · · ⊂ Sk . It is easy to see that G(B(K)) is
homeomorphic with G(K). In fact, there is a “canonical” homeomorphism β : G(B(K)) →
G(K), where β(vS ) is the baricenter of conv(S) (which is a face of G(K)), and β is extended
linearly over each face in G(B(K)).

1.3 Homotopy
Let T1 and T2 be two topological spaces. Two continuous maps f0 , f1 : T1 → T2 are
called homotopic (denoted f0 ' f1 ), if they can be deformed into each other. More exactly,
there exists a continuous mapping F : T1 × [0, 1] → T2 such that f0 (x) = F (x, 0) and
f1 (x) = F (x, 1).
We say that T1 and T2 are homotopy equivalent (denoted by K1 ' K2 ), if there exist
continuous maps f : T1 → T2 and g : T2 → T1 such that f ◦ g ' idT2 and g ◦ f ' idT1 .
Let T1 ⊆ T2 be two topological spaces. We say that T1 is a retract of T2 if there is a
continuous map ϕ : T2 → T1 such that ϕ|T1 = idT1 . If in addition ϕ ∼ idT2 (as maps
T2 → T2 ), then we call T1 a deformation retract of T2 .
It is easy to see that every deformation retract of a space is homotopy equivalent to it.
This does not remain valid for retracts in general, but every retract of a contractible space
is contractible.
A topological space T is contractible if it is homotopy equivalent to the space with a single
point. This is equivalent to saying that the map T → {p} (where p ∈ T ) is homotopic with
the identity map idT . This property does not depend on the choice of p.
A topological space is k-connected (k ≥ 0) if for every 0 ≤ r ≤ k, every continuous map
f : S r → T extends to a continuous map f : B r+1 → T . Equivalently, f is homotopic to a
constant map. Sometimes it is convenient to define (−1)-connected as “non-empty”.
We say that a simplicial complex is contractible [k-connected] if its geometric realization is.
We define homotopy equivalence of two simplicial complexes and homotopy of two simplicial
maps between simplicial complexes in a similar way.

Example 1.1 Any two continuous maps from a topological space into a contractible space
are homotopic. Every convex body is contractible. A simple graph (as a 1-dimensional
simplicial complex) is contractible if and only if it is a tree.
Any face of a convex polytope is a retract of the polytope. The 2-point space consisting
of the endpoints of a segment is not a retract of this segment. An arc of a circle is a retract,
but not a deformation retract, of the full circle.
If a simplicial complex has a vertex u that is contained in every maximal simplex, then
it is contractible (just contract its geometric realization homothetically from center u).

Example 1.2 The full simplex on a finite set V is the simplicial complex Σ(V ) = 2V \ {∅}.
The boundary of the full simplex Σ(V ) is the simplicial complex ∂(V ) = 2V \ {∅, V }. The
usual
P geometric representation of∼Σ(V|V)| is the standard simplex ∆V = {x ∈ RV : xi ≥
∼
and G(∂(V )) = S |V | .
0, i xi ≤ 1}. Clearly, G(Σ(V )) = B

Example 1.3 The following example will be useful later on. Let A1 , . . . , Am be disjoint
nonempty sets, and let M(A1 , . . . , Am ) consist of those subsets X ⊆ A1 ∪ . . . Am for which
|X ∩ Ai | ≤ 1 for every 1 ≤ i ≤ m. Then M(A1 , . . . , Am ) is a simplicial complex.

2
 An interesting special case is obtained when Ai = {ui , vi } has two elements. In this case
the simplices of M(A1 , . . . , Am ) can
P be identified with the proper faces of the m-dimensional
cross-polytope Xn = {x ∈ Rn : |x i | ≤ 1}. So M(A 1 , . . . , A m ) ∼
= S m−1
.
i
Let A0i ⊆ Ai , then M(A01 , . . . , A0m ) ⊆ M(A1 , . . . , Am ). In fact, M(A01 , . . . , A0m ) is a
retract of M(A1 , . . . , Am ): we can take any surjective map ϕi : Ai → A0i for i = 1, . . . , m,
then ϕ = ϕ1 ∪ · · · ∪ ϕm is a simplicial map M(A1 , . . . , Am ) → M(A01 , . . . , A0m ) that is a
retraction.

2 Combinatorial homotopy theory

2.1 Contractibility
The bad news: given a simplicial complex, it is algorithmically undecidable whether it is con-
tractible. This implies that it is algorithmically undecidable whether two simplicial complexes
are homotopy equivalent, or two simplicial maps are homotopic. It is somewhat surprising
that in spite of these facts, there is a useful combinatorial theory of homotopy equivalence
and contractibility.
Lemma 2.1 Let T be a topological space and let K be a simplicial complex. Let f, g : T →
G(K) be two continuous maps such that f (x) and g(x) are contained in the same face of
G(K) for all x ∈ T . Then f ∼ g.

Proof. Φ(t, x) = (1 − t)f (x) + tg(x) ∈ G(K2 ) defines a homotopy of f and g. 

Lemma 2.2 For a simplicial complex K with V (K) = V , the following are equivalent:
(i) K is contractible.
(ii) K is k-connected for every k ≥ 0.
(iii) K is a retract of Σ(V ).

Proof. (i)⇒(ii): Trivial.

(ii)⇒(iii): We prove by induction on n that for every simplicial complex K0 ⊇ K with
V (K0 ) = V and |K0 | = n, G(K0 ) has a retraction to G(K). For n = 2|V | we get the assertion.
Let S be a maximal simplex in K0 such that S ∈ / K, and let L = K0 \ {S}. By induction,
we have a retract ϕ : G(L) → G(K). Then ϕ gives a continuous map ϕ0 : ∂(G(Σ(S))) →
G(K). By k-connectivity, this extends to a continuous map ϕ1 : G(Σ(S)) → G(K). Then
ϕ ∪ ϕ1 : G(K0 ) → G(K) is a retraction.
(iii)⇒(i): Σ(V ) is contractible, and so every retract of it is contractible as well. 

Lemma 2.3 Let K be a simplicial complex and U ⊆ V (K). Suppose that K ∩ Σ(U ) is
contractible. Then K ∪ Σ(U ) ∼ K.

Proof. By Lemma 2.2, there is a retraction ϕ0 : G(Σ(U )) → G(K ∩ Σ(U )). Define
ϕ : G(K ∪ Σ(U )) → G(K ∪ Σ(U )) by
(
ϕ0 (x) if x ∈ G(Σ(U )),
ϕ(x) =
x, otherwise.

Clearly ϕ(x) ∈ G(K) for every x, and ϕ, as a map G(K∪Σ(U )) → G(K), is a retraction. Since
x and ϕ(x) are contained in the same simplex of K ∪ Σ(U ), it follows that ϕ ∼ idG(K∪Σ(U )) .
Hence ϕ is a deformation retraction, showing that K ∪ Σ(U ) ∼ K. 
This lemma generalizes to adding more than one simplex.

3
Lemma 2.4 Let K be a simplicial complex and let U1 , . . . , Um ⊆ V (K). Suppose that for
every 1 ≤ i1 < · · · < ir ≤ m, the restriction K ∩ Σ(Ui1 ∩ · · · ∩ Uir ) is either empty or
contractible. Then K ∼ K ∪ Σ(U1 ) ∪ · · · ∪ Σ(Um ).

Proof. By induction on m. If m = 0, then we have nothing to prove. Let m ≥ 1. By

Lemma 2.3, we have K ∼ K0 = K ∪ Σ(U1 ). Furthermore,

K0 ∩ Σ(Ui1 ∩ · · · ∩ Uir ) = (K ∩ Σ(Ui1 ∩ · · · ∩ Uir )) ∪ Σ(U1 ∩ Ui1 ∩ · · · ∩ Uir )

is either empty or contractible for every 2 ≤ i1 < · · · < ir ≤ m, again by Lemma 2.3. By the
induction hypothesis, this implies that

K0 ∼ K0 ∪ Σ(U2 ) ∪ · · · ∪ Σ(Um ) = K ∪ Σ(U1 ) ∪ · · · ∪ Σ(Um ).

This proves the Lemma. 

Lemma 2.5 Let K1 and K2 be contractible simplicial complexes. If K1 ∩ K2 is contractible,

then so is K1 ∪ K2 .

Proof. Let Vi = V (Ki ), V = V1 ∪ V2 and S = V1 ∩ V2 . First, assume that K1 ∩ K2 is a

single simplex. The complex Σ(V1 ) ∪ Σ(V2 ) is contractible, hence by Lemma 2.2 there is a
retraction ϕ : G(Σ(V )) → G(Σ(V1 )) ∪ G(Σ(V2 )). Similarly, Ki is contractible, and so there is
a retraction ϕi : G(Σ(Vi )) → G(Ki ). Since Σ(S) ⊆ Ki , we have ϕi |G(Σ(S)) = idG(Σ(S)) , and so
ϕ1 ∪ϕ2 is a well-defined continuous map G(Σ(V1 ))∪G(Σ(V2 )) → G(K1 )∪G(K2 ) = G(K1 ∪K2 ),
which is clearly a retraction. So (ϕ1 ∪ ϕ2 ) ◦ ϕ : G(Σ(V )) → G(K1 ∪ K2 ) is a retraction, which
proves that K1 ∪ K2 is contractible by Lemma 2.2.
Next, we treat the more general (but still not completely general) case when K1 ∪ Σ(S) =
K2 ∪ Σ(S). Then K1 ∪ Σ(S), K1 ∪ Σ(S) and (K1 ∩ K2 ) ∪ Σ(S) are contractible by Lemma 2.5,
and hence by the special case proved above, so is (K1 ∪ Σ(S)) ∪ (K2 ∪ Σ(S)) = K1 ∪ K2 ∪ Σ(S).
By Lemma 2.5, this implies that K1 ∪ K2 is contractible.
Finally, in the general case, consider the baricentric subdivisions B(Ki ). These are
contractible, and so is B(K1 ) ∩ B(K2 ) = B(K1 ∩ K2 ). Furthermore, we have V (B(K1 )) ∩
V (B(K2 )) = K1 ∩ K2 , and the restriction of B(Ki ) to this is B(K1 ∩ K2 ) for i = 1, 2. Hence
the previous special case applies and we get that B(K1 ) ∪ B(K2 ) = B(K1 ∪ K2 ) is contractible.
This implies that K1 ∪ K2 is contractible. 
Let K be a simplicial complex, and let f : U → V (K) be any mapping. Define the
simplicial complex

f −1 (K) = {X ⊆ U : f (X) ∈ K}.

Lemma 2.6 Let K1 and K2 be simplicial complexes and let f : V (K1 ) → V (K2 ) be a
surjective simplicial map. Suppose that for every S ∈ K2 , the complex K1 ∩ Σ(f −1 (S)) is
contractible. Then K1 ∼ K2 .

Proof. We start with proving the special case when K1 = f −1 (K2 ). In this case f −1 (S) ∈
K1 for every S ∈ K2 , so the contractibility condition is trivially satisfied. By the definition
of f −1 (K2 ), the mapping f : V (K1 ) → V (K2 ) is simplicial, and so it extends to a continuous
mapping fb : G(K1 ) → G(K2 ).
For every u ∈ V (K2 ), let g(u) ∈ G(K1 ) be the center of gravity of the simplex f −1 (u) ∈ K1 .
If S = {u1 , . . . , ud } ∈ K2 , then f −1 (S) ∈ K1 , and hence g(u1 ), . . . , g(ud ) are contained in the
face conv(f −1 (S)) of G(K1 ). Hence we can extend the map g linearly to a continuous map
gb : G(K2 ) → G(K1 ).

4
 It is clear that gb ◦ fb = idG(K2 ) . On the other hand, let x ∈ G(K1 ), say x ∈ conv(S),
where S ∈ K1 . Then both x and fb(b g (x)) are contained in the face conv(f −1 (f (S)), and
hence f ◦ gb ∼ idG(K1 ) by Lemma 2.1. This proves that K1 ∼ K2 .
b
Now in the general case, we already know that f −1 (K2 ) ∼ K2 . Let S1 , . . . , Sm be the
maximal simplices in K2 , and let Ui = f −1 (Si ). Then

K1 ∩ Σ(Ui1 ∩ · · · ∩ Uir ) = K1 ∩ Σ(f −1 (Si1 ∩ · · · ∩ Sir ))

is contractible for every 1 ≤ i1 < · · · < ir , and hence by Lemma 2.4,

K1 ∼ K1 ∪ Σ(U1 ) ∪ · · · ∪ Σ(Um ) = Σ(U1 ) ∪ · · · ∪ Σ(Um ) = f −1 (K2 )

Thus K1 ∼ f −1 (K2 ) ∼ K2 . 

2.2 The Nerve Theorem

Let G = (V, E) be a bipartite graph with bipartition V = U ∪ W . The neighborhood complex
on U NU = NU (G) of G is the simplicial complex consisting of all subsets A of U such that
the elements of A have a common neighbor. NW = NW (G) is defined analogously.
Lemma 2.7 The two neighborhood complexes of a bipartite graph are homotopy equivalent.

Proof. For every simplex S ∈ NU , let f (S) ⊆ W denote the set common neighbors of
nodes in S. Clearly f (S) ∈ NW , so f : NU → NW . We define g : NW → NU analogously.
If S1 ⊂ · · · ⊂ Sk , then f (S1 ) ⊃ · · · ⊃ f (Sk ), and hence f is simplicial as a map
V (B(NU )) = NU → V (B(NW )) = NW . Thus f defines a continuous map f : G(B(NU )) →
G(B(NW )). We get a continuous map g : G(B(NW )) → G(B(NU )) analogously.
Let α : G(B(NU )) → G(NU ) and β : G(B(NW )) → G(NW ) be the canonical homeomor-
phisms. We claim that the maps fb = β ◦ f ◦ α−1 and gb = α ◦ g ◦ β −1 establish the homotopy
equivalence of G(NU ) and G(NW ).
Consider the map h = gb◦ fb = α◦g ◦f ◦α−1 : G(NU ) → G(NU ). We claim that x and h(x)
belong to the same simplex for every x ∈ G(K1 ). Indeed, let S ∈ NU be a smallest simplex
such that x ∈ conv(S). Then α−1 (x) ∈ conv{S1 , . . . , Sk } for a simplex in B(NU ) such that
S1 ⊂ · · · ⊂ Sk = S. It follows that conv(g(f (S1 )), . . . , g(f (S1 ))) contains g(f (α−1 (x))), and
so α(conv(g(f (S1 )), . . . , g(f (Sk )))) contains h(x). This (geometric) simplex is contained in
conv(g(f (S))). Also, trivially g(f (S)) ⊇ S, and hence x ∈ conv(S) ⊆ conv(f (g(S))). So
Lemma 2.1 implies that h and idG(NU ) are homotopic.
We get similarly that fb ◦ gb ∼ idG(NW ) , and so G(NU ) ∼ G(NW ) as claimed. 
Let H be a hypergraph consisting of nonempty sets. The nerve of H is defined as the
simplicial complex nerve(H) whose vertices are the sets in H, and {X1 , . . . , Xr } ∈ nerve(H)
if and only if X1 , . . . , Xr ∈ H and X1 ∩ · · · ∩ Xr 6= ∅.
Corollary 2.8 Let K be a simplicial complex, and assume that H ⊆ K contains all maximal
simplices in K. Then K ∼ nerve(H) ∼ nerve(K).

More generally:
Theorem 2.9 (Nerve Theorem) Let K be a simplicial complex and let K1 , . . . , Km be sub-
complexes such that K = K1 ∪ · · · ∪ Km . Assume that Ki1 ∩ · · · ∩ Kir is either empty or
contractible for every 1 ≤ i1 < · · · < ir ≤ m, r ≥ 1. Then K ∼ nerve{V (K1 ), . . . , V (Km )}.

Proof. Let Vi = V (Ki ). First, we prove the case when Ki = K ∩ Σ(Vi ). In this case
Ki1 ∩ · · · ∩ Kir = K ∩ Σ(Vi1 ∩ · · · ∩ Vir ). Lemma 2.4 implies that K ∼ K ∪ Σ(V1 ) ∪ · · · ∪ Σ(Vm ) =

5
Σ(V1 ) ∪ · · · ∪ Σ(Vm ). By Corollary 2.8, we have Σ(V1 ) ∪ · · · ∪ Σ(Vm ) ∼ nerve{V1 , . . . , Vm },
which proves the theorem in this case.
For the general case (when Ki may be smaller than K ∩ Σ(Vi )), we replace each Ki by
B(Ki ) as in the proof of Lemma 2.5. 

2.3 Topological connectivity

The lemmas from the previous sections can be generalized, with some care, from contractibil-
ity to k-connectivity.
Lemma 2.10 Let K be a simplicial complex and k ≥ 0. Then the following are equivalent:
(a) K is k-connected.
(b) The (k + 1)-dimensional skeleton K|k+1 is k-connected.
(c) K|k+1 is a retract of Σ(V (K))|k+1 .

Lemma 2.11 Let K be a simplicial complex and U ⊆ V (K).

(a) If K is k-connected and K ∩ Σ(U ) is (k − 1)-connected, then K ∪ Σ(U ) is k-connected.
(b) If K ∪ Σ(U ) and K ∩ Σ(U ) are k-connected, then K is k-connected.

Lemma 2.12 Let K1 and K2 be two k-connected simplicial complexes and assume that K1 ∩
K2 is (k − 1)-connected. Then K1 ∪ K2 is k-connected.

The Nerve Theorem 2.9 has the following version for k-connectivity.
Theorem 2.13 (Connectivity Nerve Theorem) Let K be a simplicial complex and let
K1 , . . . , Km be subcomplexes such that K = K1 ∪ · · · ∪ Km . Assume that Ki1 ∩ · · · ∩ Kir is
either empty or (k − r + 1)-connected for every 1 ≤ i1 < · · · < ir ≤ m, 1 ≤ r ≤ k + 1. Then
K is k-connected if and only if the nerve of {V (K1 ), . . . , V (Km )} is k-connected.

Example 2.14 As an application of this theorem, we analyze the connectivity of the simpli-
cial complex M(A1 , . . . , Am ) from Example 1.3. If |Ai | = 1 for some i, then every maximal
simplex contains the single element of Ai , and hence M(A1 , . . . , Am ) is contractible. If
|Ai | ≥ 2 for every i, then M(A1 , . . . , Am ) is (m − 2)-connected. This is trivial if m ≤ 1,
so we may assume that m ≥ 2. Let Mx be the set of all simplices in M(A1 , . . . , Am ) that
contain x or can be extended to contain x. Then Mx is contractible by the argument above.
Furthermore,
[
M(A1 , . . . , Am ) = Mx .
x∈A1

Clearly

Mx1 ∩ · · · ∩ Mxr = M(A2 , . . . , Am )

if x1 , . . . , xr are distinct elements of A1 (r ≥ 2). The complex Mx1 ∩ · · · ∩ Mxr is (m −

3)-connected by induction. The Connectivity Nerve Theorem 2.13 applies, and gives that
M(A1 , . . . , Am ) is (m − 2)-connected.

Example 2.15 Let V = {1,
. . . , n}, let 0 ≤

a ≤ b ≤ n, and consider the bipartite graph G
between color classes U = Va and W = Vb , where X ∈ U is adjacent to Y ∈ W if and only
if X ⊆ Y . We claim that the neighborhood complexes NU = NU (G) and NW = NW (G) are
(b − a − 1)-connected.
We know that NU ∼ NW (Lemma 2.7), so it suffices to prove that NU is (b − a −
1)-connected. This is trivial in some cases: if a = b (recall that (−1)-connected means

6
non-empty); if a = 0 (a one-point space is contractible); and if b = n (a full simplex is
contractible). So we may assume that 0 < a < b < n.
For every i ∈ V , let Ui = V \{i}


a and let Ki = NU ∩ Σ(Ui ). Since b < n, we have
NU = ∪i Ki . By induction on n, we know that Ki is (b − a − 1)-connected.
More generally, let 1 ≤ i1 < · · · < ir ≤ n. If r ≤ n − b, then Ki1 ∩ · · · ∩ Kir is
0
0

just the neighborhood complex of the bipartite graph between levels Va and Vb , where
V 0 = V \ {i1 , . . . , ir }. Hence by induction, Ki1 ∩ · · · ∩ Kir is (b − a − 1)-connected. If
n − b ≤ r ≤ n − a, then Ki1 ∩ · · · ∩ Kir is a full simplex, so it is contractible. If r > n − a,
then Ki1 ∩ · · · ∩ Kir = ∅. So the Connectivity Nerve Theorem 2.13 applies, and we get that
NU is (b − a)-connected.

An important use of connectivity is the following extension property of maps.

Lemma 2.16 Let K is a (k + 1)-dimensional simplicial complex, let K0 be a subcomplex of
K, and T a k-connected space. Then every continuous map f : G(K0 ) → T extends to a
continuous map G(K) → T .

Proof. Let S be a simplex in K \ K0 with minimal dimension. Then ∂(S) ⊆ K0 and so

the map f is already defined on G(∂(S)). Hence by the k-connectivity of T , f extends to a
continuous map G(K0 ∪ {S}) → T . Repeating this argument, we extend f to a continuous
map G(K) → T . 

3 Brouwer’s fixed point theorem

A simplicial complex K such that G(K) ∼
= S r is called a triangulation of the sphere S r .
Lemma 3.1 (Sperner’s lemma) Let K be a triangulation of the sphere S r . Let
f : V (K) → {0, 1, . . . , r} be any labeling of the vertices of S r . Then the number of sim-
plices whose vertices are labeled by different numbers is even.

Theorem 3.2 (Brouwer’s fixed point theorem) Every continuous map B r → B r has a
fixed point. 

There are a number of equivalent facts.

Corollary 3.3 (a) Let K be a contractible simplicial complex. Then every continuous map
G(K) → G(K) has a fixed point.
(b) S r−1 is not a retract of B r .
(c) Let f : Σr → Σr be a continuous map such that for every face F of Σr , f (F ) ⊆ F .
Then f is surjective.

We need the following group-theoretic corollary of Brouwer’s Fixed Point Theorem.

Corollary 3.4 Let T be a compact contractible topological space and let G be a finite cyclic
group acting on T . Then the elements of G have a common fixed point.

Proof. Let ϕ be the action of a generator of G. Then ϕ has a fixed point by Corollary 3.3,
and this is a fixed point of every other element of G as well. 
This last corollary does not remain valid for all finite groups, but it does remain valid for
certain non-cyclic finite groups. One such class of finite groups is the following.

7
Theorem 3.5 [41] Let T be a compact contractible topological space and let Γ be a finite
group acting on T . Assume that Γ has a normal p-subgroup Γ1 such that Γ/Γ1 is cyclic.
Then the elements of G have a common fixed point. 

Example 3.6 To illustrate how to apply these results, let us return to Example 1.3. We
show that if |Ai | ≥ 2 for all 1 ≤ i ≤ m, then M(A1 , . . . , Am ) is not (m − 1)-connected. Let
A0i = {ui , vi } ⊆ Ai . The embedding M(A01 , . . . , A0m ) → M(A1 , . . . , Am ) gives continuous
map of f : ∂Xn → G(M(A1 , . . . , Am )). Suppose that f extends to a map fb : Xn →
G(M(A1 , . . . , Am )). Since M(A01 , . . . , A0m ) is a retract of M(A1 , . . . , Am ), we may assume
that fb maps into G(M(A01 , . . . , A0m )) = ∂Xn . But this means that ∂Xn is a retract of Xn , a
contradiction.

3.1 Evasive graph properties

Consider any property Pn of graphs with n nodes (we assume that if a graph has this property
then every graph isomorphic with it also has it). We say that Pn trivial, if either every graph
has it or no one has it. A graph property is monotone if whenever a graph has it each of its
subgraphs has it. For most graph properties that we investigate (connectivity, the existence
of a Hamiltonian circuit, the existence of perfect matching, colorability, etc.) either the
property itself or its negation is monotone.
We say that a property Pn of n-node graphs is evasive, if it cannot be computed by a
decision tree of depth less than n2 . The Aanderaa–Rosenberg–Karp conjecture says that

every nontrivial monotone graph property is evasive. This conjecture is unsolved, but it has
been proved for a number of graph properties. Here we prove the following
Theorem 3.7 (Kahn, Saks, Sturtevant) Every nontrivial monotone property of graphs
on n nodes, where n is a prime power, is evasive.

The analogous theorem is also proved for bipartite graphs (Yao). See also http://www.
cs.elte.hu/~lovasz/complexity.pdf and http://arxiv.org/abs/cs/0205031 for other
special cases, provable by topology and also by different tools.
Before proving this theorem, it will be useful to generalize it. We call a Boolean function
f : {0, 1}n → {0, 1} evasive, if it cannot be computed by a decision tree of depth less than n.
We define the automorphism group of a Boolean function as the set of permutations of the
variables that does not change the value of the function. Recall that this permutation group
is called transitive, if for any two variables x and y there is an automorphism that moves x to
y. We say that the function f is monotone decreasing (or just “monotone” for this section),
if f (x1 , . . . ,n ) ≤ f (y1 , . . . , yn ) whenever xi ≥ yi for all 1 ≤ i ≤ n.
To see how graph properties fit in, let G be graph with V = V (G) = {1, . . . , n}. For every
pair i, j ⊆ V , let us introduce a Boolean variable xij with value 1 if i and j are adjacent and
0 if they are not.
In this way, any property of n-point graphs can be considered as a Boolean
function with n2 variables. This Boolean function has a transitive automorphism group: for
any two pairs of nodes there is a permutation of the nodes taking one into the other, which
leads to an isomorphic graph and hence does not change the value of the Boolean function.
The following Generalized Aanderaa–Rosenberg–Karp Conjecture is open: If the auto-
morphism group of a non-constant monotone Boolean function is transitive, then the function
evasive. We prove it in a special case.
Theorem 3.8 If the automorphism group of a non-constant monotone Boolean function f
contains a transitive subgroup Γ, and Γ has a normal subgroup ∆ of prime power order such
that Γ/∆ is cyclic, then f is evasive.

Corollary 3.9 If the automorphism group of a non-constant monotone Boolean function is

transitive, and it is a cyclic group, then it is evasive.

8
Corollary 3.10 If the automorphism group of a non-constant monotone Boolean function
is transitive, and the number of its variables is a prime power, then it is evasive.

Next, we describe how this problem relates to topology. Let f : {0, 1}n → {0, 1} be
a monotone decreasing Boolean function that is not identically 0. For x ∈ {0, 1}n , let
supp(x) = {i ∈ {1, . . . , n} : xi = 1}. We associate with f the following simplicial complex:

Kf = supp(x) : x ∈ {0, 1}n , x 6= 0, f (x) = 1 .



Lemma 3.11 If f is non-evasive, then Kf is contractible.

Proof. If f is non-evasive, then there is decision tree computing f with depth at most
n − 1. This decision tree starts with checking a variable, say x1 . The two branches of the
tree compute the Boolean functions f0 (x2 , . . . , xn ) = f (0, x2 , . . . , xn ) and f1 (x2 , . . . , xn ) =
f (1, x2 , . . . , xn ). Since the depth of these branches is at most n − 2, the functions f0 and f1
are non-evasive. By induction, we may assume that Kf0 and Kf1 are contractible.
Let Kf0 1 = Kf1 ∪ {X ∪ {1} : X ∈ Kf1 }, then all maximal simplices of Kf0 1 contain 1, and
so Kf0 1 is contractible. We have Kf0 ∪ Kf0 1 = K and Kf0 ∩ Kf0 1 = Kf1 , and so Lemma 2.5
implies that K is contractible. 
Next, we observe the following.
Lemma 3.12 If f is a nonconstant monotone Boolean function with a transitive automor-
phism group Γ, then Γ acts on G(Kf ) and has no fixed point.

Proof. It is clear that every automorphism γ of f is a simplicial map γ : Kf → Kf , and

this extends to a continuous map γ b : G(Kf ) → G(Kf ).
We claim that these maps have no common fixed point. Suppose that x ∈ G(Kf ) satisfies
b(x) = x for every γ ∈ Γ. There is a unique smallest simplex S such that x ∈ conv(S). Then
γ
x=γ b(x) =∈ conv(γ(S)), so by the minimality of S, we must have γ(S) = S. Since the group
γ is transitive, this can only happen for every γ ∈ Γ if S = V (Kf ). But then the function is
identically true. 

Proof of Theorem 3.8. If f is non-evasive, then G(Kf ) is contractible by Lemma 3.11.

If f is nontrivial, then Γ acts on G(Kf ) without fixed points. But this contradicts the
following theorem of Oliver from topology: If Γ is a finite group that has a normal subgroup
∆ of prime power order such that Γ/∆ is cyclic, then every continuous action of Γ on a
contractible simplicial complex has a fixed point. 

Proof of Theorem 3.7. We consider V (G) as the set of elements of a finite field Fq (where q
is any prime poser), and define Γ as the group of linear transformations of the form x 7→ ax+b,
where a, b ∈ F bbq and a 6= 0. Then Γ acts on pairs {i, j} ⊆ Fq transitively. Furthermore,
if ∆ denotes the subgroup of Γ of transformations of the form x 7→ x + b, then |∆| = q is a
prime power, and Γ/∆ ∼ = Fq∗ (the multiplicative group of nonzero elements of Fq ), which is
cyclic by basic results in algebra. So Theorem 3.8 applies and proves the theorem. 

4 The Borsuk–Ulam Theorem

Let T be a topological space and let ι be an antipodality on T , i.e., a homeomorphism
ι : T → T such that ι is an involution (ι ◦ ι = idT ) and ι has no fixed points. Sometimes we
denote ι(x) by −x. T , together with the antipodality ι, is called an antipodality space.
An antipodal map between antipodality spaces T1 and T2 is a continuous map f : T1 to
T2 that satisfies f (−x) = −f (x).

9
Lemma 4.1 If T is a k-connected topological space with an antipodality, then there exists
an antipodal map of S k+1 into T .

The three assertions in the following theorem are all essentially equivalent to each other,
and are called the Borsuk–Ulam Theorem.
Theorem 4.2
(a) There is no antipodal map from S r into S r−1 (r ≥ 1).
(b) For every continuous map f : S r → Rr there exists an x ∈ S r such that f (x) = f (−x).
(c) If S r is covered by r + 1 sets, and either each of these is closed or each of these is open,
then one of these sets contains an antipodal pair.

The Borsuk–Ulam theorem has the following following discrete version. Let P be a full-
dimensional convex polytope in Rd . We say that two faces A and B of P are opposite if
there exists a linear function ` : Rd → R that is maximized by the set of points in A and
minimized by the set of points in B.
Theorem 4.3 [6] Let P be a full-dimensional convex polytope in Rd and f : P → Rd−1 .
Then P has two opposite faces A and B such that f (A) ∩ f (B) 6= ∅.

The following slight extension of this theorem will be needed later. We say that the map
f : P → Rd−1 is generic if for every pair of faces A and B with dim(A) + dim(B) < d − 1,
we have f (A) ∩ f (B) = ∅, and for every pair of faces A and B with dim(A) + dim(B) = d − 1,
the set f (A) ∩ f (B) is finite.
Theorem 4.4 [36] Let P be a full-dimensional convex polytope in Rd and let f : P → Rd−1
be a generic map. Then P has two opposite faces A and B such that dim(A)+dim(B) = d−1
and |f (A) ∩ f (B)| is odd.

We say that the polytope P in Rd is generic, if for every two opposite faces A and B,
dim(A) + dim(B) = d − 1.
Theorem 4.5 Let P be a generic full-dimensional convex polytope in Rd and let f : P →
Rd−1 be a generic map. Then
X
|f (A) ∩ f (B)| ≡ 1 (mod2),

where the summation extends over all pairs of opposite faces A and B of P .

4.1 Simplicial complexes associated with graphs

Let G = (V, E) be any graph. The neighborhood complex N (G) of G is the simplicial complex
consisting of all subsets A of V such that the elements of A have a common neighbor.
For a bipartite graph, the neighborhood complex consists of two components.
The set Hom(F, G) of homomorphisms F → G can also be equipped with a topological
structure. We say that a set of homomorphisms ϕ1 , . . . , ϕk : F → G is a cluster if for every
edge uv ∈ E(F ) and any 1 ≤ i < j ≤ k, we have ϕi (u)ϕj (v) ∈ E(G). It is clear that these
clusters form a simplicial complex Hom(F, G) (i.e., they are closed under taking subsets).
This construction is “functorial”, which means that every homomorphism ψ : G1 →
G2 induces a simplicial (and hence continuous) map ψb : Hom(F, G1 ) → Hom(F, G2 ) in
a canonical way: For every homomorphism
ϕ : F → G1 , we define ψ(ϕ) b

= ϕψ. It is
trivial that this map from V Hom(F, G1 ) = Hom(F, G1 ) to V (Hom F, G2 ) = Hom(F, G2 )
maps clusters onto clusters. Similarly, Hom(., .) is a “contravariant functor” in its first
variable, which means that every homomorphism ξ : F1 → F2 induces a simplicial map
ϕ̌ : Hom(F2 , G) → Hom(F1 , G).

10
 The complex Hom(K2 , G) plays a special role. We will denote it by H(G). The points of
H(G) can be thought of as ordered pairs (u, v), where u and v are adjacent nodes of G. This
space has a natural antipodality: ι(u, v) = (v, u) defines a simplicial map that extends to a
ι : G(H(G)) → G(H(G)). If G1 → G2 is a graph
bijective and fixed-point-free involution b
homomorphism, then the map Hom(K2 , G1 ) → Hom(K2 , G2 ) it induces is antipodal.
Example 4.6 The complex N (Kn ) is homeomorphic to the (n − 2)-dimensional sphere:
N (Kn ) = Σ(V ) \ {V } (here V = V (Kn ) = {1, . . . , n}).
The complex H(Kn ) is homotopy equivalent to the (n − 2)-dimensional sphere S n−2 .
Define a map f : G(H(Kn )) → Rn as follows. Let e1 , . . . , en be the standard basis in Rn .
For (u, v) ∈ V (H(Kn )), let f (u, v) = ei − ej . Extend f linearly over G(H(Kn )).
We claim that the origin is not in the range of f . Indeed, consider a point x ∈
conv((u1 , v1 ), . . . , (um , vm )), where {(u1 , v1 ), . . . , (um , vm )} ∈ H(Kn ). Note that this implies
that ui 6= vj for 1 ≤ i, j ≤ m. If f (x) = 0, then
m
X
ai (eui − evj ) = 0
i=1
P
for some ai ≥ 0, i ai = 1. We may assume that a1 > 0, then the inner product of the left
side with eu1 is positive, which is a contradiction.
It follows that x 7→ f (x)/kf (x)k is well defined, and gives an antipodal map G(H(Kn )) →
S n−1 . Since the range is contained in the hyperplane H = {x :
P
x
i i = 0}, this map goes
into S n−1 ∩ H, which is a copy of S n−2 .
The construction of the inverse map is not given here (we will need this map only).

It is not a coincidence that H(Kn ) and N (Kn ) turned out to be homotopically equivalent,
as this is shown by the following lemma.
Lemma 4.7 For every graph G, the complexes H(G) and N (G) are homotopy equivalent.

Proof. We define a simplicial map f : V (H(G)) → V (N (G)) as follows: For every

(u, v) ∈ V (H(G)) we let f (u, v) = u. It suffices to show that this map satisfies the conditions
of Lemma 2.6: for every simplex S ∈ N (G), the complex H(G) ∩ Σ(f −1 (S)) is contractible.
For every T ∈ N (G), let Q(T ) denote the set of nodes adjacent to every node of T . Then
Q(T ) 6= ∅ by definition, and T × Q(T ) ∈ H(G). For a given S ∈ N (G), we have
[
H(G) ∩ Σ(f −1 (S)) = Σ(T × Q(T )), (1)
T ∈Σ(S)

and hence by Corollary 2.8,

H(G) ∩ Σ(S × Q(S)) ∼ nerve{T × Q(T ) : T ∈ Σ(S)} (2)

Now it is easy to see that for T1 , . . . , Tr ⊆ S, we have

T1 × Q(T1 ) ∩ · · · ∩ Tr × Q(Tr ) 6= ∅ ⇔ T1 ∩ · · · ∩ Tr 6= ∅. (3)

Indeed, the implication ⇒ is trivial, and if T1 ∩ · · · ∩ Tr 6= ∅ than (u, v) ∈ T1 × Q(T1 ) ∩ · · · ∩

Tr × Q(Tr ) for any u ∈ T1 ∩ · · · ∩ Tr and v ∈ Q(S). Thus

nerve{T × Q(T ) : T ∈ Σ(S)} = nerve(Σ(S)) ∼ Σ(S).

Since Σ(S) is contractible, this proves the Lemma. 

11
Remark 4.8 It would be tempting to try a shorter argument and use the formula
[
H(G) = Σ(T × Q(T )),
T ∈N (G)

which implies
H(G) ∼ nerve{T × Q(T ) : T ∈ N (G)}.
But the nerve on the right is difficult to handle, since (3) does not remain valid if T1 , . . . , Tr
are not faces of the same simplex.

4.2 Topological connectivity and chromatic number

Theorem 4.9 [32] If the neighborhood complex N (G) of a graph G is k-connected (equiva-
lently, the homomorphism complex Hom(K2 , G) is k-connected), then χ(G) ≥ k + 3.

Proof. By Lemma 4.1 there is an antipodal map S k+1 → Hom(K2 , G). Suppose that
χ(G) ≤ k + 2, then there is a homomorphism G → Kk+2 , which induces an antipodal map
Hom(K2 , G) → Hom(K2 , Kk+2 ). As we have seen in Example 4.6, there is an antipodal
map Hom(K2 , Kk+2 ) → S k . Composing these maps, we get an antipodal map S k+1 → S k ,
contradicting the Borsuk-Ulam Theorem. 
The following related theorem shows why using the homomorphism complex leads to
generalizations.
Theorem 4.10 [4]. If Hom(C2r+1 , G) is k-connected as a topological space for some r ≥ 1,
then the chromatic number of G is at least k + 4.
The Kneser graph Kkn is defined as the graph whose nodes are all k-subsets of an n-set,
and two are adjacent if and only if they are disjoint (n ≥ k).
Theorem 4.11 [32] For n ≥ 2k, the chromatic number of the Kneser graph Kkn is n−2k +2.
Corollary 4.12 If all k-element subsets of a (2k + r − 1)-element set are divided into r
classes, then one of the classes contains two disjoint k-sets.
By Theorem 4.9, it suffices to prove the following:
Lemma 4.13 The N (Kkn ) is (n − 2k − 1)-connected.

Proof. The neighborhood complex N (Kkn ) can be described as follows: its vertices are
all k-subsets of V = {1, . . . , n}, and vertices S1 , . . . , Sm } form a simplex if any only if
|S1 ∪ · · · ∪ Sm | ≤ n − k. It was shown in Example 2.15 that this complex is (n − 2k − 1)-
connected. 

4.3 The Ham Sandwich Theorem

Theorem 4.14 Let A1 , . . . , Ad be measurable sets in Rd with finite measure. Then there
exists a closed halfspace H such that µ(H ∩ Ai ) = 21 µ(Ai ) for all i.

Proof. For every vector v = (v0 , v1 , . . . , vd ) ∈ S d , we define

fi (v) = λ{x ∈ Ai : v0 + v1 x1 + · · · + vd xd ≥ 0} − λ{x ∈ Ai : v0 + v1 x1 + · · · + vd xd ≤ 0}
(where λ is the Lebesgue measure). The map v 7→ (f1 (v), . . . , fd (v)) is antipodal, so by the
Borsuk-Ulam Theorem there is a v ∈ S d such that f1 (v) = · · · = fd (v) = 0. But then the
halfspace H defined by the inequality v0 + v1 x1 + · · · + vd xd ≥ 0 satisfies the conditions of
the Theorem. (We need that the hyperplane v0 + v1 x1 + · · · + vd xd = 0 has measure 0.) 

12
4.4 The Necklace problem
Theorem 4.15 Consider an (open) necklace consisting of pearls of k colors. Suppose that
there is an even number of pearls of each color. Then we can split the necklace at k points,
and divide the arising pieces between two robbers so that each robber gets exactly half of the
pearls of each color.

Theorem 4.16 (Continuous version) Let A1 ∪ · · · ∪ Ak be a partition of [0, 1] into mea-

surable parts. Then there is a partition J1 ∪ · · · ∪ Jk+1 of [0, 1] into intervals, and a set
I ⊆ {1, . . . , k + 1} such that
X 1
λ(Jr ∩ Ai ) = λ(Ai )
2
r∈I

for every i = 1, . . . , k.

Proof. For z ∈ S k , let J1 (z) ∪ · · · ∪ Jk+1 (z) be the partition of [0, 1] into intervals with
λ(Ji (z)) = zi2 . Define
k+1
X
fi (z) = sgn(zr )λ(Ai ∩ Jr (z)) (i = 1, . . . , k).
r=1

Then fi is continuous (this needs a little argument), and fi (−z) = −fi (z). Hence by the
Borsuk-Ulam Theorem, there is a z ∈ S k such that f1 (z) = · · · = fk (z) = 0. Then I =
{j : sgn(zr ) = 1} satisfies the requirements of the theorem. 

References
[1] N. Alon, Splitting necklaces, Adv. in Math. 63 (1987), 247–253.
[2] N. Alon, I. Bárány, Z. Füredi, D. Kleitman, Point selections and weak -nets for convex
hulls
[3] N. Alon, P. Frankl, L. Lovász, The chromatic number of Kneser hypergraphs, Trans.
Amer. Math. Soc. 298 (1986), 359–370.
[4] E. Babson and D.N. Kozlov: Proof of the Lovász conjecture, Annals of Math. 165
(2007), 965–1007.
[5] R. Bacher, Y. Colin de Verdière, Multiplicités des valeurs propres et transformations
étoile-triangle des graphes, preprint, 1994.
[6] E.G. Bajmóczy, I. Bárány, On a common generalization of Borsuk’s and Radon’s theo-
rem, Acta Mathematica Academiae Scientiarum Hungaricae 34 (1979) 347–350.
[7] I. Bárány, Z. Füredi, L. Lovász, On the number of halving planes Combinatorica 10,
175–183.
[8] I. Bárány, S.B. Shlosman and A. Szűcs, On a topological generalization of a theorem of
Tverberg, J. London Math. Soc. 23, 158–164.
[9] M. Ben-Or: Lower bounds for algebraic computation trees, 15th ACM STOC (1983)
80–86.
[10] A. Björner, Topological Methods, in: Handbook of Combinatorics (ed. R. Graham, M.
Grötschel, L. Lovász), Chapter 34, 1819–1872. North-Holland, Amsterdam (1995).

13
[11] A. Björner, Some Cohen-Macaulay Complexes arising in Group Theory, in Commutative
Algebra and Combinatorics (ed. M. Nagata and H. Matsumura), Advanced Studies in
Pure Mathematics 11, North-Holland, Amsterdam, 1987, pp. 13–19.
[12] A. Björner, B. Korte, L. Lovász, Homotopy properties of greedoids, Advances in Appl.
Math. 6 (1985), 447–494.

[13] A. Björner, L. Lovász, A. Yao, Linear decision trees, hyperplane arrangements, and
Möbius functions, in: Proc. 24th ACM SIGACT Symp. on Theory of Computing, 170–
177.
[14] A. Björner, L. Lovász, Linear decision trees, subspace arrangements, and Möbius func-
tions Journal of the Amer. Math. Soc. 7, 677–706.

[15] The connectivity of chessboard complexes (A. Björner, S. Vrec̀ica, R. Živaljevic̀) Journal
of the London Math Soc. 49 (1994) 25–39.
[16] T. Böhme, On spatial representations of graphs, in: Contemporary Methods in Graph
Theory (R. Bodendieck, ed.), BI-Wiss.-Verl. Mannheim, Wien/Zurich, 1990, pp. 151–
167.

[17] Y. Colin de Verdière, Sur un nouvel invariant des graphes et un critère de planarité,
Journal of Combinatorial Theory, Series B 50 (1990) 11–21.
[18] Y. Colin de Verdière, On a new graph invariant and a criterion for planarity, in: Graph
Structure Theory (N. Robertson, P. Seymour, eds.), Contemporary Mathematics, Amer-
ican Mathematical Society, Providence, Rhode Island, 1993, pp. 137–147.
[19] A. Dold: Lectures on Algebraic Topology, Springer-Verlag, 1972.
[20] A. Dold: Parametrized Borsuk–Ulam theorems, Comment. Math. Helv. 63 (1988) 275–
285.

[21] J. Folkman, The homology groups of a lattice, J. Math. Mech. 15 (1966) 631–636.
[22] M. Goresky and R. MacPherson: Stratified Morse Theory, Ergebnisse, Band 14,
Springer-Verlag, Berlin (1988)
[23] E. Győri, On division of graphs to connected subgraphs, in: Combinatorics I (eds.
A. Hajnal and V.T. Sós), Coll. Math. Soc. J. Bolyai 18 (1987) 485–494.

[24] H. van der Holst, A short proof of the planarity characterization of Colin de Verdière,
Journal of Combinatorial Theory, Series B 65 (1995) 269–272.
[25] H. van der Holst, M. Laurent, A. Schrijver, On a minor-monotone graph invariant,
Journal of Combinatorial Theory, Series B 65 (1995) 291–304.

[26] H. van der Holst, L. Lovász, A. Schrijver, On the invariance of Colin de Verdière’s graph
parameter under clique sums, Linear Algebra and Its Applications 226 (1995) 509–517.
[27] J. Kahn, M. Saks and D. Sturtevant, A topological approach to evasiveness, it Combi-
natorica 4 (1984) 297–306.

[28] D. Kozlov: Combinatorial Algebraic Topology, Springer, Berlin, 2008.

[29] M.A. Krasnoselsky and P.P. Zabrejko, Geometricheskie Zadachi Nelinejnogo Analiza,
Nauka, Moscow, 1975.
[30] L. Lovász, On the number of halving lines, Annales Univ. Eötvös 14 (1971), 107-108.

14
[31] L. Lovász, A homology theory for spanning trees of a graph, Acta Math. Hung. 30,
241-251.
[32] L. Lovász, Kneser’s conjecture, chromatic number, and homotopy, J. Comb. Theory A
25 (1978), 319-324.
[33] L. Lovász, Topological and algebraic methods in graph theory, in: Graph Theory and
Related Topics, Academic Press, 1979, 1-14.
[34] L. Lovász, Matroids and Sperner’s Lemma, Europ. J. Combin. 1, 65-66.
[35] L. Lovász, Self-dual polytopes and the chromatic number of distance graphs on the
sphere, Acta Sci. Math. Szeged 45, 317-323.

[36] L. Lovász, A. Schrijver, A Borsuk theorem for antipodal links and a spectral character-
ization of linklessly embeddable graphs (preprint)
[37] J. Mather: Invariance of the homology of a lattice, Proc. Amer. Math. Soc. 17 (1966)
1120–1124.

[38] J. Matoušek: Using the Borsuk-Ulam Theorem: Lectures on Topological Methods in

Combinatorics and Geometry, Springer (2003).
[39] J. Milnor: On the Betti numbers of real algebraic varieties, Proc. Amer. Math. Soc. 15
(1964) 275–280.
[40] J.R. Munkres: Elements of Algebraic Topology, Addison-Wesley, Menlo Park (1984).

[41] R. Oliver: Fixed-point sets of group actions on finite acyclic complexes, Comment. Math.
Helvet. 50 (1975), 155–177.
[42] N. Robertson, P.D. Seymour, R. Thomas, A survey of linkless embeddings, in: Graph
Structure Theory (N. Robertson, P. Seymour, eds.), Contemporary Mathematics, Amer-
ican Mathematical Society, Providence, Rhode Island, 1993, pp. 125–136.
[43] N. Robertson, P. Seymour, R. Thomas, Sachs’ linkless embedding conjecture, Journal
of Combinatorial Theory, Series B 64 (1995) 185–227.
[44] R.P. Stanley: Enumerative Combinatorics, Vol. 1, Wadsworth & Brooks/Cole, Monterey
(1986).

[45] M. Steele and A. Yao, Lower bounds for algebraic decision trees, J. Algorithms 3 (1982),
1–8.
[46] R. Thom: Sur l’homologie des variétés algébraiques réelles, in: Differential and Algebraic
Topology (ed. S. S. Cairns), Princeton Univ. Press, Princeton (1965).

[47] K. Wagner, Über eine Eigenschaft der ebene Komplexe, Mathematische Annalen 114
(1937) 570–590.
[48] R. Živaljević: User’s guide to equivariant methods in combinatorics, Publ Inst. Math.
Belgrade 59(73) (1996), 114–130.

[49] R.T. Živaljević, S.T. Vrećica: The colored Tverberg’s problem and complexes of injective
functions, J. Combinatorial Theory Ser. A 61 (1991) 309–318.

15