MATH422 Cheat Sheet

O§ul Can Yurdakul

Last Updated: 27/06/2021

Contents
1 Topology of Subsets of Euclidean Spaces 1
1.1
n
Open and Closed Subsets of Sets in R . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Continuous Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Connectedness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Compactness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5 Homeomorphisms & Quotient Maps . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Topological Surfaces 5
2.1 Arcs, Disks, and 1-Spheres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Surfaces in Rn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Surfaces via Gluing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4 Properties of Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.5 Connected Sum and Classication of Compact Connected Surfaces . . . . . . . . 8
3 Simplicial Complexes 9
3.1 Ane Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2 Convex Sets & Simplices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.3 Simplicial Complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.4 Simplicial Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.5 The Euler Characteristic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.6 Simplicial Curvature and the Simplicial Gauss-Bonnet Theorem . . . . . . . . . . 15

1 Topology of Subsets of Euclidean Spaces

1.1 Open and Closed Subsets of Sets in Rn

Balls:

Open: Let p ∈ Rn and r > 0. The open ball centered at p with radius r is

Or (p; Rn ) = {x ∈ Rn : kx − pk < r}

Relatively Open: Let p ∈ A ⊂ Rn and r > 0. The open ball in A of radius r is

Or (p; A) = {x ∈ A : kx − pk < r} = Or (p; Rn ) ∩ A

Closed: Let p ∈ Rn and r > 0. The closed ball centered at p with radius r is

Cr (p; Rn ) = {x ∈ Rn : kx − pk ≤ r}

1
1 TOPOLOGY OF SUBSETS OF EUCLIDEAN SPACES

Relatively Closed: Let p ∈ A ⊂ Rn and r > 0. The closed ball in A of radius r is

Cr (p; A) = {x ∈ A : kx − pk ≤ r} = Or (p; A)

Open/Closed Set: A set A ⊂ B is said to be open in B if for every p ∈ A, there exists some
εp > 0 real such that Op (εp ; B) ⊂ A. A set A ⊂ B is said the be closed in B if its
complement in B is
n
open in B . Setting B = R , we obtain the regular open and closed set
denitions.
For closed and open sets in Rn , we have

- ∅ and Rn are both open and closed.

- Finite intersection of open sets is open.
- Finite union of closed sets is closed.
- Arbitrary union of open sets is open.
- Arbitrary intersection of closed sets is closed.

Relative Opennes/Closedness as Intersection: Let A ⊂ Rn . A subset S ⊂ A is

open/closed in A if and only if there exists some open/closed subset U ⊂ Rn such
that S = U ∩ A.
If in particular A is open, then U ∩ A is open, and so relative openness is the same as
general openness.

Open Neighbourhood: If p ∈ A, then an open neighbourhood of p in A is an open

subset of A containing p.
Producing Open Rectangles: For A ⊂ Rn and B ⊂ Rm ,
i) for U ⊂ A and V ⊂ B , then U × V is an open subset of A × B.
open open

ii) If W ⊂ A × B is an open subset of A × B , then for every (p1 , p2 ) ∈ W there exist

εA , εB > 0 such that OεA (p1 ; A) × OεB (p2 ; B) ⊂ W .
reals

Closure: There exists a smallest closed subset containing a given subset. Let D ⊂ A ⊂ Rn .
The closure of D in A is dened to be the intersection of all closed subsets of A containing
D, denoted D or sometimes ClA (D).

Open/Closed Maps: Let A ⊂ Rn , B ⊂ Rm and f : A → B . Then f is said to be an open

map if for every U ⊂ A, f (U ) is also open in B . Similarly, f is said to be a closed map
open

if for every C ⊂ A, f (C) is also closed in B.

clsd

1.2 Continuous Maps

Continuity: Let A ⊂ Rn and B ⊂ Rm , and f :A→B be a map. The map f is continuous if

with Open Preimage: for every U ⊂ B, the inverse image f −1 (U ) is open in A:

open

U ⊂ B ⇒ f −1 (U ) ⊂ A
open open

with ε − δ : for every p ∈ A and every ε > 0, there exists some δ > 0 such that if x∈A
with kx − pk < δ then kf (x) − f (p)k < ε. This means

f (Oδ (p; A)) ⊂ Oε (f (p); B)

with Open Neighbourhoods: for every p ∈ A and every subset U ⊂ B containing

open

f (p), there exists V ⊂ A containing p such that f (V ) ⊂ U .

open

2 Contents
1 TOPOLOGY OF SUBSETS OF EUCLIDEAN SPACES

Continuity of Restrictions: Let A ⊂ Rn , B ⊂ Rm and f : A → B . Suppose that A = A1 ∪A2

and f |A1 and f |A2 are both continuous. If A1 and A2 are both open or both closed in A,
then f is continuous.

Pasting Lemma: A corollary to this is that we can dene an overall function f :A→B
in terms of two functions , say f1 : A1 → B and f2 : A2 → B , dened on both open
or both closed subsets A = A1 ∪ A2 ; given that f1 and f2 agree on the intersection
A1 ∩ A2 .

Continuity of Component Functions: Let A, B1 , . . . , Bk be subsets of a Euclidean space and

f : A → B1 × · · · × Bk a function. Denote with fi : A → Bi the component functions of f ,
which can also be written as πi ◦ f where πi is the projection function onto Bi . Then f is
continuous if and only if all its fi component functions are continuous.

1.3 Connectedness

(Dis)Connected Sets: Let A ⊂ Rn . A is said to be disconnected if we can write A = A1 ∪ A2

where A1 ∩ A2 = ∅ with A1 and A2 both open in A. If A is not disconnected, then we say
it is connected.
Equivalently,

- A is connected if A cannot be expressed as a disjoint union of two of its closed subsets.

- A is connected if the only subsets of A that are both open and closed are A and ∅.

Component: Let A ⊂ Rn . Then C⊂A is a component of A if it is non-empty, connected and

not a proper subset of a connected subset of A.

Intervals: Let A ⊂ R. Then A is connected if and only if it is an interval (in the most general
sense).

Continuous Image: Let A ⊂ Rn and B ⊂ Rm , and f : A → B a continuous map. If A is

connected, so is f (A).
Continuous image of a connected set is connected.
This fact can be used to prove the Intermediate Value Theorem.

Path Connectedness: Let A ⊂ Rn , x, y ∈ A. A path from x to y is a continuous map

c : [0, 1] → A such that c(0) = x and c(1) = y . A is said to be path connected if there
exists a path between any x, y ∈ A.
A path connected set is connected, but the converse statement need not hold.

1.4 Compactness

Cover, Subcover, Finite Cover, Open Cover: Let A ⊂ Rn . A cover U of A is a collection

of subsets of A whose union gives all of A.
If U = {Ui }i∈I is a cover of A indexed by some index set I, then a subcover of U is a
subcollection of U that is itself a cover of A, i.e. it is of the form {Uj }j∈J where J ⊂ I.
A nite cover is one that is constituted of nitely many sets.
An open cover of A constitutes of open subsets of A.

(Un)Bounded Sets: Let A ⊂ Rn . Then A is bounded if there exists some non-negative real r
such that A ⊂ Or (0n ; Rn ).

3 Contents
1 TOPOLOGY OF SUBSETS OF EUCLIDEAN SPACES

Compact Sets: Let A ⊂ Rn . We say that A is compact if every open cover of A has a nite
subcover.
The union of nitely many compact sets is compact.
A compact subset of a compact set is closed. A closed subset of a compact set is compact.
Compact sets are bounded.

Heine-Borel Theorem: A set A ⊂ Rn is compact if and only if it is closed in Rn and bounded.

The proof of this theorem uses

- A closed interval in R is compact.

- The product of nitely many compact subsets of a Euclidean space is compact.

Continuous Image: Let A ⊂ Rn and B ⊂ Rm , and f : A → B a continuous map. If A is

compact, so is f (A).
Continuous image of a compact set is compact.
This fact can be used to prove the Extreme Value Theorem.

1.5 Homeomorphisms & Quotient Maps

Homeomorphism: Let A ⊂ Rn , B ⊂ Rm and f : A → B. f is a homeomorphism (or a

topological equivalence) if it is a continuous bijection with continuous inverse.
Note that f −1 is continuous if and only if f is an open map.

Partition: Let
S A ⊂ Rn . A collection P = {Pi }i∈I of subsets of A is called a partition if
Pi = A and Pi ∩ Pj = ∅ whenever i 6= j .
i∈I

Quotient Map: Let A, B ⊂ Rn . A map q : A → B is called a quotient map if it is surjective

and U ⊂ B if and only if q
−1 (U ) ⊂ A.
open open

Note that by the second condition, a quotient map is by denition, automatically contin-
uous.

Identication Space: If X ⊂ Rn and P = {Pi }i∈I is a partition of X , then Y ⊂ Rm is

an identication space of X and P if there is a quotient map q : X → Y such that
{q −1 (y) : y ∈ Y } = P .
The interpretation of this denition is that the quotient map given collapses each part of
X in P to a single point, and each point in Y is mapped on by some part of X, almost as
if
bij
Pi 7−→ y (but not under q !)

Continuity & Quotient Maps: Let X , Y and Z be subsets of Euclidean spaces, f :X→Y
a quotient map and g : Y → Z. Then g is continuous if and only if g◦f is continuous.

Homeomorphic Identication Spaces: Let X ⊂ Rn and P be a partition of X . If Y ⊂ Rm

and Z⊂ Rk are both identication spaces of X and P , then we have Y ≈ Z , meaning Y
and Z are the same spaces up to homeomorphism.

Sewing Sets Together: Let X, Y ⊂ Rn X 0 ⊂ X , Y 0 ⊂ Y . If h : X 0 → Y 0

be disjoint sets and
is a homeomorphism, then we can dene a partition P(h) on X ∪ Y , which consists of pairs
{x, h(x)} for all x ∈ X 0 and {z} for all z ∈ (X \ X 0 ) ∪ (Y \ Y 0 ). A set W ⊂ Rm is said to
be the result of attaching X and Y through h, denoted X ∪h Y , if W is the identication
space of X ∪ Y and P(h).

Continuous Maps and Compactness: Let A ⊂ Rn , B ⊂ Rm and f :A→B a continuous

map. Then if A is compact

4 Contents
2 TOPOLOGICAL SURFACES

i) then f is a closed map.

ii) and f is a surjection, then f is a quotient map.
iii) and f is a bijection, then f is a homeomorphism.

2 Topological Surfaces

2.1 Arcs, Disks, and 1-Spheres

Basic Denitions: We will give some specic names to the following sets as follows:

D2 = {x ∈ R2 : kxk ≤ 1} (Standard Unit Disk in R2 )

Int D2 = {x ∈ R2 : kxk < 1} (Standard Open Disk in R2 )
S 1 = {x ∈ R2 : kxk = 1} = ∂D2 (Unit Circle in R2 )

Then

Arc: A subset of Rn that is homeomorphic to the interval [−1, 1] is called an arc.

Disk: A subset of Rn that is homeomorphic to D2 is called a disk.

1-Sphere: A subset of Rn that is homeomorphic to S1 is called a 1-sphere.

Invariance of Domain: Let U ⊂ Rn be homeomorphic to Rn . Then U must be open in Rn .

Non-homeomorphic Dimensions: For n and m distinct positive integers, Rn 6≈ Rm .

Boundaries and Interiors of Disks: Let B ⊂ Rn be a disk, and h1 , h2 : D2 → B be homeo-

morphisms. Then

h1 (S 1 ) = h2 (S 1 ) and h1 (Int D2 ) = h2 (Int D2 )

and so, independent from the homeomorphism generating the disk, we can make the fol-
lowing denitions: For any homeomorphism h : D2 → B

Interior of a Disk: Int B = h(Int D2 )

Boundary of a Disk: ∂B = h(∂D2 ) = h(S 1 )

Boundaries and Interiors of Arcs: A similar fact can be proven about arcs and their bound-
aries as well: Their interior and boundaries are independent from the homeomorphism
generating them. So, for any homeomorphism h : [−1, 1] → A

Interior of an Arc: Int A = h(Int[−1, 1]) = h((−1, 1))

Boundary of an Arc: ∂A = h(∂[−1, 1]) = h({−1, 1})

Identity Outside of a Disk: Let h : R2 → R2 be a homeomorphism. h is called as the identity

outside a disk if there is
2
some disk B ⊂ R such that h|R2 \A = id.

Schönies Theorem: Let C ⊂ R2 be a 1-sphere. Then there is a homeomorphism H : R2 → R2

1
such that H(S ) = C and H is the identity outside a disk.
Corollary:

i) Jordan Curve Theorem: The set R2 \ C has precisely 2 components, one bounded
and another unbounded.

5 Contents
2 TOPOLOGICAL SURFACES

ii) The union of C and the bounded component of R2 \ C is a disk, of which C is the
boundary.
This second fact does not hold in dimensions higher than 2, as seen by the counter
example of the trefoil knot.
iii) Let B1 , B2 ⊂ R2 be two disks. Then there is a homeomorphism H : R2 → R2 such
that H(B1 ) = B2 and H is the identity outside a disk.

2.2 Surfaces in Rn

(Topological) Surface: A subset Q ⊂ Rn is called a (topological) surface if each point p∈Q

has an open neighbourhood that is homeomorphic to the open unit dist int D2 .

Sameness of Surfaces: Two surfaces Q1 and Q2 are considered to be the same if they are
homeomorphic.

2.3 Surfaces via Gluing

Polygonal Disk: A polygonal disk is a disk that sits in some plane in Rn whose boundary is a
polygon.

Gluing Scheme: If D is a polygonal disk, a gluing scheme S for the edges of D is a labelling
of each edge of F with an arrow and a letter, where each used letter appears on precisely
two edges.

Ane Linear Map: An ane linear map L : [a, b] → [c, d] is a map that satises

L ((1 − t)a + tb) = (1 − t)L(a) + tL(b) for t ∈ [0, 1]

Edge Sets: Let D be a polygonal disk and S be a gluing scheme on the edges of D. Then, S
partitions the edges of D into 2-element sets where the edges matched under S are paired,
say E1 , . . . , Ek . These are called as the edge sets.

Vertex Sets: For each edge set, we have two ane linear maps that send one edge to the other
such that the end points are matched according to the directions set by S. Of course, one
is the inverse of the other, so we can represent the mapping of each edge set with a single
Li by choosing one of these mappings, and the other one is L−1
i . Then dene the following
partition:
[x] = {y ∈ D : y = L±1 ±1
i ◦ Lj (x) for some i, j ∈ {0, 1, . . . , k}}
where we write L0 for the identity map. These sets, which form a partition of the vertices
of D, are called the vertex sets, denoted P(S) or P(L1 , . . . , Lk ).

Constructing Surfaces: Let D be a polygonal disk and S a gluing scheme on its edges. A
subset X ⊂ Rn is said to be obtained from D and S if X is an identication space of D
surj.
and P(S), that is, there is a quotient map q : D −→ X such that if x, y ∈ D are points,
then q(x) = q(y) if and only if x and y are in the same set in P(S).
n
For each polygonal disk D and gluing scheme S on D , there is a surface Q ⊂ R that is
obtained from D and S .

Compact Connected Surfaces via Gluing: Let Q ⊂ Rn be a compact connected sur-

face. Then there is a polygonal disk D and a gluing scheme S for the edges of D such
that Q is obtained from D and S.

6 Contents
2 TOPOLOGICAL SURFACES

This fact is proved based on another statement asserting any surface can be triangu-
lated. See Section 3 for more.

Various Important Surfaces and Their Gluing Schemes:

Möbius Strip: M 2. M 2 is not really a surface as it has a boundary, but it appears within
many important surfaces, so it is mentioned here.

E. Bloch

Sphere: S2, no need to elaborate.

Torus: T 2

E. Bloch

Klein Bottle: K2

K2 can be obtained from two M 2 's via a

homeomorphism of their boundaries.:

K 2 = (M 2 )1 ∪h (M 2 )2

E. Bloch

7 Contents
2 TOPOLOGICAL SURFACES

Projective Plane: P2

The removal of an open disk from P 2 yields

M 2:
P 2 \ int B ≈ M 2
This means that we can form a P2 using an
M 2 and a disk D2 via a homeomorphism
of their boundaries.

E. Bloch

2.4 Properties of Surfaces

Compactness: Any surface Q obtained from a polygonal disk by gluing is compact, as they are
the continuous images
2
(via the quotient map) of the compact D . So, any compact and
connected surface is compact.

Oritentability: A surface is orientable if it does not contain a Möbius strip. It is non-orientable

if it does.

Connectedness:

Around Points: Let Q be a surface and q ∈ Q. Then q has an open neighbourhood in Q

which is path connected.

Overall: A surface in Rn is connected if and only if it is path connected.

2.5 Connected Sum and Classication of Compact Connected Surfaces

Connedted Sum: Let Q1 and Q2 be compact and connected surfaces. For each, choose a disk
B1 and B2 , respectively. Let h : ∂B1 → ∂B2 be a homeomorphism. The attaching space

(Q1 \ int B1 ) ∪h (Q2 \ int B2 )

is called the connected sum of Q1 and Q2 , denoted Q1 #Q2 .

Exitence and Uniqueness upto Homeomorphism: This attaching space indeed ex-
ists and is a surface in some Rn . Any two surfaces obtained in such a way are
homeomorphic.

Basic Properties: Let A, B and C be compact and connected surfaces. Then

i) A#B ≈ B #A
ii) (A#B)#C ≈ A#(B #C)
iii) A#S 2 ≈ A
Notice that these properties yield a commutative group-like structure, where S2 acts
like the identity. We are only missing the inverses of each element. This algebraic
structure is called a commutative semigroup. The next property illustrates that we
indeed cannot have inverses, other than the trivial case.

8 Contents
3 SIMPLICIAL COMPLEXES

Trivial Inverses: Let A and B be compact connected surfaces. If we have A#B ≈ S 2 ,

then A≈B≈ S2.

Connected Sum of Important Surfaces: The following homeomorphisms hold:

P 2 #P 2 ≈ K 2 : This one is easy to prove. Remember that we are removing the interior of
a disk while constructing the connected sum. Then, the two P2 after the removal of
2
the said disk are homeomorphic to M , and then we are taking a homeomorphism of
2
their boundary, which we had seen results in K .

P 2 #T 2 ≈ P 2 #P 2 #P 2

Orientability and Connected Sum: Let Q1 and Q2 be compact connected surfaces. Then
Q1 #Q2 is orientable if and only if both Q1 and Q2 are orientable.

Clasication of Compact Connected Surfaces: Any compact connected surface is homeo-

morphic to a sphere, a connected sum of tori or a connected sum of projective planes, that
is, it belongs to the following list:

- S2,
- T 2 , T 2 #T 2 , T 2 #T 2 #T 2 , . . .
- P 2 , P 2 #P 2 , P 2 #P 2 #P 2 , . . .
The surfaces in this list are all distinct (i.e. not homeomorphic to one another).

3 Simplicial Complexes

3.1 Ane Preliminaries

Ane Combination, Span, Subspace & Independence: Let x0 , . . . , x k ∈ Rn . An ane

combination of these points, with coecients t0 , . . . , tk ∈ R is the linear combination

k
X k
X
t i xi with ti = 1
i=0 i=0

A set X ⊂ Rn is said to be an ane subspace if it is closed under ane combinations.

The ane span of these points is the set of all ane combinations of these points:

k k
( )
X X
n
aspan (x0 , . . . , xk ) = y∈R :y= ti xi with ti = 1
i=0 i=0

The set {x0 , . . . , xk } is said to be anely independent if

k
X k
X
ti xi = 0n with ti = 0 ⇒ ti = 0 ∀i = 0, . . . , k
i=0 i=0

Relation to Linear Independence: The set {x0 , . . . , xk } is anely independent if and

only if the set {x1 − x0 , . . . , xk − x0 } is linearly independent.

Uniqueness: The set {x0 , . . . , xk } is anely independent if and only if every y in the ane
span of{x0 , . . . , xk } is uniquely expressible as an ane combination of x0 , . . . , xk . The
coecients ti in this ane combination are called the barycentric coordinates of y .

9 Contents
3 SIMPLICIAL COMPLEXES

Ane Linear Map: Let X ∈ Rn be an ane subspace. A map F : X → Rm is an ane linear

map if it preserves ane combinations, that is if x0 , . . . , xk ∈ X and t0 , . . . , tk ∈ R with
k
P
ti = 1, then
i=0
k k
!
X X
F t i xi = ti F (xi )
i=0 i=0

3.2 Convex Sets & Simplices

Convex Sets: For v, w ∈ Rn , let vw denote the line segment between v and w, given by

vw = {x ∈ Rn : x = tv + (1 − t)w for 0 ≤ t ≤ 1}
A subset X ⊂ Rn is said to be convex if for every pair of points v, w ∈ X , the line segment
vw is entirely contained in X.
Image Under Ane Linear Maps: For a convex set C ⊂ Rn and an ane linear map
F : Rn → Rm , the set F (C) is also convex. This means that convexity is preserved
under ane linear maps.

Convex Hull: It is intuitive (and true) that every set X is contained in some convex set, so
we might ask for the smallest such set. Let X ⊂ Rn be any set. The convex hull of X,
denoted conv X , is dened by
\
conv X = {C ⊂ Rn : X ⊂ C and C is convex}

This set is indeed convex, and it a subset of every convex set containing X.
Simplex: Let a0 , . . . , a k ∈ Rn be anely independent points where k is a non-negative integer.

Via Convex Hull: The simplex spanned by the points a0 , . . . , ak is the convex hull of
these points, denoted
ha0 , . . . , ak i = conv{a0 , . . . , ak }
The points a0 , . . . , ak are called the vertices of the simplex, and ha0 , . . . , ak i is said
to be k -dimensional, and called a k -simplex. This makes sense, given the linear
independence relation above. Our k + 1 many points give us a k -dimensional object.

Source

Via Ane Combinations: A simplex is equivalently expressed in terms of its vertices'

ane combinations with an extra condition. Let a0 , . . . , a k ∈ Rn be anely indepen-
dent points, where k∈ N+ . Then
k k
( )
X X
ha0 , . . . , ak i = x ∈ Rn : x = ti ai with ti = 1 AND ti ≥ 0 ∀i
i=0 i=0

again, these ti coordinates are unique, and are called barycentric coordinates.

10 Contents
3 SIMPLICIAL COMPLEXES

Unique Vertices: Let {a0 , . . . , ak } and {b0 , . . . , bp } be two sets of anely independent
points in Rn . Then we have

ha0 , . . . , ak i = hb0 , . . . , bp i ⇒ {a0 , . . . , ak } = {b0 , . . . , bp }

This means that the set of vertices of a simplex uniquely determines the simplex.

Unique Ane Subspace: Let µ = ha0 , . . . , ak i be a k-simplex in Rn . Then the ane

subspace aspan (a0 , . . . , ak ) is the unique ane subspace (k -plane) containing µ.
Face of a Simplex: Let σ = ha0 , . . . , ak i be a k -simplex in Rn . A face of σ is the simplex
generated by a non-empty subset of {a0 , . . . , ak }. If this subset is proper subset, then the
corresponding face is called a proper face. A face of σ that is an r -simplex is called an
r-face of σ .
Combinatorial Boundary: The combinatorial boundary of a k -simplex σ = ha0 , . . . , ak i
is the union of all the proper faces of σ.
[
Bd σ = hτ i Mind the abuse of notation.

τ ({a0 ,...,ak }

The combinatorial boundary, by this logic, is given as

k k
(
X X
Bdha0 , . . . , ak i = x ∈ Rn : x = ti ai with ti = 1,
i=0 i=0
)
ti ≥ 0 for some i and there exists j such that tj = 0

Notice how imposing a zero in the barycentric coordinates forces the elements to be
on a proper face. If there are s zeros with 0 < s ≤ k in the barycentric coordinates of
some x, then x lies on a (k − s)-face.
Combinatorial Interior: The combinatorial interior is the set given by

Int σ = σ − Bd σ
The combinatorial interior, by this logic, is given as

k k
( )
X X
Intha0 , . . . , ak i = x ∈ Rn : x = ti ai with ti = 1, ti > 0 ∀i
i=0 i=0

Notice how the strict positivity constraint keeps the points strictly inside the simplex
and not on the boundary.

General Disks and Spheres: We will use of the following notation from now on:

Dk = {x ∈ Rn : kxk ≤ 1} Closed unit k -disk

k−1 n
S = {x ∈ R : kxk = 1} Unit (k − 1)-sphere
We had already introduced D2 , S 2 and S1 already.

Simplices & Boundaries as Disks & Spheres: Let σ = ha0 , . . . , ak i be a k -simplex in Rn .

Then, there is a homeomorphism
h : Dk → σ
such that
h(S k−1 ) = Bd σ
If σ is a 1-simplex, then it is an arc. If it is a 2-simplex, then it is a disk. In both cases,
we have
Bd σ = ∂σ & Int σ = int σ

11 Contents
3 SIMPLICIAL COMPLEXES

Compactness and Connectedness: Every k -simplex σ = ha0 , . . . , ak i and Bd σ are compact

and path connected.

3.3 Simplicial Complexes

Simplicial Complex: A simplicial complex K in Rn is a nite collection of simplices in Rn

such that

i) if σ ∈ K , then all faces of σ are in K .

ii) if σ, τ ∈ K with non-empty intersection, then σ∩τ must be a face of both σ and τ.
For a simplicial complex K , K ((i)) is dened as the set of all i-simplices in K.

Subcomplex: A subcollection L of K is a subcomplex of K if L itself is a simplicial

complex.

Star & Link: Let K be a simplicial complex in Rn , and σ ∈ K. The star of σ is dened as

star (σ, K) = {η ∈ K : η is a face of a simplex in K, which has σ as a face}

Based on this denition, the link of σ is dened as

link (σ, K) = {η ∈ star (σ, K) : η ∩ σ = ∅}

Simplicial Maps: Let K be a simplicial complex in Rn and let L be a simplicial complex in

Rm . A map
f : K ((0)) → L((0))
is called a simplicial map if

ha0 , . . . , ai i ∈ K ⇒ hf (a0 ), . . . , f (ai )i ∈ L

Simplicial Isomorphism: Such a map is a simplicial isomorphism if it is bijective, and

its inverse is also a simplicial map. If there is a simplicial isomorphism between K
and L, we say that they are simplicially isomorphic.

Underlying Space: Let K be a simplicial complex in Rn . The underlying space of K , denoted

|K|, is the subset
n
of R given by the union of all simplices in K .
[
|K| = σ
σ∈K

Notice of |K| is dierent from K as a set: K contains a nite amount of nite sets, while |K|
contains points from Rn and is most likely an uncountable set (it is countable if K = K ((0)) )

Unique Simplex per Point: For each x ∈ |K|, there exists a unique simplex η ∈ K
such that x ∈ Int η . Otherwise, two simplices intersect on a set that is not a face of a
simplex in K .

Subdivision: Let K andK 0 be simplicial complexes in Rn . The simplicial complex K 0 is said

to subdivide K if |K| = |K 0 | and every simplex in K 0 is a (not necessarily proper) subset
of a simplex of K.

Induced Map of a Simplicial Map: Let K be a simplicial complex in Rn and let L be a

m
simplicial complex in R , with f : K ((0)) → L((0)) a simplicial map. The induced map
of the underlying spaces |K| and |L| is the map |f | : |K| → |L| dened by extending f
anely over each simplex.

12 Contents
3 SIMPLICIAL COMPLEXES

Continuity: The induced map |f | of a simplicial map f is continuous.

Isomorphic Spaces: If f is a simplicial isomorphism between K and L, then |K| ≈ |L|.

Isomorphic Subdivision Spaces: If K and L have simplicially isomorphic subdivisions,
then |K| ≈ |L|.

Connectedness: Let K be a simplicial complex in Rn . Then, the following are equivalent:

i) |K| is path connected.

ii) |K| is connected.
iii) For any two simplices σ and τ in K, there is a collection of simplices ηi of K

τ = η1 , η2 , . . . , ηp = σ

such that ηi ∩ ηi+1 6= ∅ for all i = 1, . . . , p − 1

Simplicial Quotient Maps: Let K be a simplicial complex in Rn and L a simplicial complex

m
in R . A simplicial map f : K ((0)) → L((0)) is a simplicial quotient map if
i) For every simplex hb0 , . . . , bp i ∈ L, there exists a simplex ha0 , . . . , ap i ∈ K such that
f (ai ) = bi .
ii) If a, b ∈ K
((0)) are two vertices of a common simplex, then f (a) 6= f (b).
Notice how this denition is in parallel with the usage of quotient maps in the identication
space denition.

Induced Map: The induced map |f | : |K| → |L| is a quotient map.

Inverse Image under the Induced Map: Let y ∈ |l| be a point, and η = hb0 , . . . , bk i ∈
L be the unique simplex with x ∈ Int η with strictly positive barycentric coordinates
k k
|f |−1 (y)
P P
ti , i.e. y= ti bi , ti = 1 and t i > 0. Then, contains all points x ∈ |K|
i=0 i=0
with the same barycentric coordinates contained in some (possibly multiple) simplex
ha0 , . . . , ak i ∈ K such that f (ai ) = bi .

Admissible Partition: Let K be a simplicial complex. An admissible partition of K ((0)) is a

collection V= {Ai }i∈I of disjoint subsets of K ((0)) such that
Ai = K ((0))
S
i) i∈I
ii) No two vertices of the same simplex are in the same set Ai .

Existence of an Image Complex: For any admissible partition V of K ((0)) of a simpli-

0
cial complex, there exists some simplicial complex K in some Rm and a simplicial
quotient map f : K
((0)) →K 0((0)) such that

n o
f −1 (v) : v ∈ K 0((0)) = V

Induced Partition: An admissible partition of V of K ((0)) induces a partition of |K|, denoted

P(V), as follows: Two points x, y ∈ |K| are in the same partition if and only if for x∈
Intha0 , . . . , ak i and y ∈ Inthb0 , . . . , bk i of K such that
i) the 0-simplices ai and bi are in the same set in the partition V , and
ii) F (x) = y under the unique ane linear map F : ha0 , . . . , ak i → hb0 , . . . , bk i with
F (ai ) = bi (meaning they have the same barycentric coordinates in their own k -
simplices).

13 Contents
3 SIMPLICIAL COMPLEXES

Induced Preimage of the Image Subspace: We know that an image simplicial com-
plex K 0 exists for any admissible partition V of K ((0)) and a simplicial quotient map
f : K ((0)) → K 0((0)) that abides by V . The induced map of this quotient map also
abides by the induced partition P(V), meaning

 −1
|f | (x) : x ∈ |K 0 | = P(V)

Homeomorphic Identication Space: The identication space of |K| and P(V) is home-
omorphic to |K 0 |.

3.4 Simplicial Surfaces

When is the underlying space of a simplicial complex a surface?

Simplicial Surface: Let K be a simplicial complex in Rn . Then |K| is a surface if and only if
K is a 2-complex such that

i) each 1-simplex is the face of precisely two 2-simplices, and

ii) the underlying space of the link of every 0-simplex of K is a 1-sphere (homeomorphic
to S 1 ).
If K satises this condition, then K is called a simplicial surface.

Discarding the First Condition: The second condition in the statement above is actu-
ally enough by itself. So, if the underlying space of the link of every 0-simplex of K
is a 1-sphere (homeomorphic to S 1 ), then K is a simplicial surface.

Triangulation: Let Q ∈ Rn
be a topological surface. A simplicial complex K is said to triangu-
late Q if there is a homeomorphism h : |K| → Q. In this case, we say that Q is triangulated
by K and K together with the homeomorphism h is called a triangulation of K .

Existence: Any compact topological surface in Rn can be triangulated.

Common Triangulations: If a topological surface is triangulated by two simplicial sur-

faces K1 and K2 , K1 and K2 have simplicially isomorphic subdivisions.

3.5 The Euler Characteristic

Euler Characteristic of a Simplicial Complex: Let K be a 2-complex, and denote

V = K ((0)) E = K ((1)) F = K ((2))

Then the Euler characteristic of K, denoted χ(K) is dened as

χ(K) = V − E + F

Invariance under Triangulations: Let K1 and K2 be two 2-complexes that triangulate the
same compact surface Q ⊂ Rn . Then

χ(K1 ) = χ(K2 )

The proof of this statement uses the fact that if a complex L is a subdivision of a complex
K, then χ(K) = χ(L), but the reason for this will be apparent later on.

14 Contents
3 SIMPLICIAL COMPLEXES

Euler Characteristic of a Surface: The Euler characteristic of a compact topological surface

is that of a 2-complex that triangulates it. This denition is well-dened due to the above
fact.

Connected Sum: Let Q1 and Q2 be compact connected surfaces in Rn . Then

χ(Q1 #Q2 ) = χ(Q1 ) + χ(Q2 ) − 2

3.6 Simplicial Curvature and the Simplicial Gauss-Bonnet Theorem

Simplicial Curvature: Let K be a simplicial surface. The curvature of K at a vertex (0-

simplex) v is dened as
X
d(v) = 2π − ∠(v, η)
η: v∈η

The simplicial curvature d(v), also called the angle defect of v, is a measure of how much
the complex deviates from being at at v.

Simplicial Gauss-Bonnet Theorem: Let K be a simplicial surface. Then

X
d(v) = 2π χ(K)
v∈K ((0))

This simple statement is used in the proof of several important statements made earlier
without creating any circular argumentation.

Orientability and Euler Characteristic Characterization: The Classication Theo-

rem gives us an explicit list of all the compact connected surfaces, where the ones in
the T2 group are all orientable and those in the P2 group are all non-orientable. If
one examines the list, those in the same group all have dierent Euler characteristics.
Therefore we conclude that two compact and connected topological surfaces Q1 and
Q2 are homeomorphic if and only if they have the same Euler characteristic AND
they are either both orientable or both non-orientable.

Invariance of χ() under Subdivisions: If the simplicial surface L is (or is homeomor-

phic to) a subdivision of a simplicial surface K, then χ(K) = χ(L).

15 Contents