ScottTopologyanditsRelationtothe
AlexandroffTopology


Presentedby

WaelMohammedAlͲHanafi


Supervisedby

AssistantProfessor:HishamB.Mahdi

SubmittedinPartialFulfillmentof

TheRequirementsforM.Sc.Degree

August,2008
Table of Contents

Table of Contents i

Table of Symbols ii

Acknowledgements iv

Abstract v

Introduction vi

1 Preliminaries 1
1.1 Partially Ordered Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . 1
1.2 Directed - Complete Posets (dcpo) . . . . . . . . . . . . . . . . . . . .. .. . . . . . . . . . 6
1.3 Topological Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 Alexandroff Space 19
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 A Glimpse of Alexandroff Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3 Identification of Basic Topological Concepts . . . . . . . . . . . . .. . . . . . . . . 27

3 Scott Topology 30
3.1 Scott Open Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2 Scott Topology and Alexandroff Topology . . . . . . . . . . . . . . . . . . . . . . . . 33

4 Scott Topology and Approximation Relation 41

4.1 The Approximation Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2 Continuous Posets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3 Scott Bases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

Conclusion 56

Bibliography 57

i
Table of symbols

WLOG: Without Loss of Generality, (57).

τ (≤): T0 -Alexandroff space, (23).

σ(P ): Scott topology on the poset P , (35).

M : the set of all maximal elements of a poset X, (4).

m : the set of all minimal elements of a poset X, (4).

M (A) : the set of all maximal elements of a subset A of X, (4).

m(A) : the set of all minimal elements of a subset A of X, (4).

: top element in a poset, (4).

⊥: bottom element in a poset, (4).

↑ x: the set of all elements greater than or equal to x, (5).

↓ x: the set of all elements less than or equal to x, (5).

x̂ :=↑ x ∩ M , (29).

KD : the set of all compact elements of a poset D, (8).

↓K x: the set of all compact elements less than or equal to x, (9).

sup D (or D): the supremum(or the join) of the set D, (4).

inf D (or D): the infimum (or the meet) of the set D, (4).

x  y: x is incomparable with y, (11).

ii
x
y: x approximates y (or x is way-below y), (44).

x 
y: x doesn’t approximate y (or x is not way-below y), (46).

⇓ x : {y : y
x}; i.e., the set of all elements that approximate x, (45).

⇑ x : {y : x
y}; i.e., the set of all elements that x approximates, (45).

iii
Acknowledgements

After God, I am grateful to my mother for her continuous prayers, support, and

love. I wish my father - God’s mercy be upon his soul - were alive, since he always

dreamed of this moment. Also I am grateful to my wife for her patience, subsidiza-

tions and kindness. All my thanks go to my family for encouraging me to continue

my study. My special thanks go to my nephew Thaer he who worked very hard with

me to complete editing this research.

I would like to express my deep thanks to my supervisor Assistant Professor

Hisham B. Mahdi for his suggestions, guidance,and constant support during this

research.

I would also like to thank all the faculty members in the Mathematics Depart-

ment who taught me, especially Prof. Mohammed S. Elatrash, Prof. Eissa D. Habil

and Dr. Asad Y. Asad for their special effects and worm relationship.

Finally, I would like to thank my colleagues at work and my friends for their

support and encouragement.

iv
Abstract

In this thesis, we survey the general topological concepts for the Scott topology, one

of the fundamental foundations of theoretical computer science. We shall concen-

trate on the definition of the T0 -Alexandroff space and some of its topological iden-

tifications so that the relation between the Scott topology and the T0 -Alexandroff

topology might be clearly discussed. We introduce here the property of being a

T0 -space for the Scott topology and due to this we establish the main result for this

research that the Scott topology and the T0 -Alexandroff topology coincide on finite

posets while in general- every Scott open subset is open in T0 -Alexandroff topology

and the converse need not be true.

The Main results of this research:

* In finite posets, each subset has a top element if and only if it is directed.

* In finite posets, every ideal is Scott-closed.

* On any poset X, the Scott topology is a T0 -space.

* A subspace of Scott topology is a Scott subspace.

* Every algebraic dcpo is continuous.

* In a continuous finite poset P , no proper subset is a basis for P .

v
Introduction

Topology is thought of as one of the branches of mathematics that has applications

in practical life. Nowadays, topology has proved to be an essential tool for certain

aspects of theoretical computer science [13]. Conversely, the problems that arise in

the computational setting have provided new and interesting stimuli for topology

[13]. These problems also have increased the interaction between topology and re-

lated areas of mathematics such as order theory and topological algebra [13] [Order

theory is a branch of mathematics that studies various kinds of binary relations that

capture the intuitive notion of ordering, providing a framework for saying when one

thing is “less than” another. Domain theory deals with partially ordered sets to

model a domain of computation. The goal is to interpret the elements of such an

order as pieces of information or (partial) results of a computation, where elements

that are higher in the order extend the information of the elements below them in

a consistent way[23]].

A concept that plays an important role in the theory is the one of a directed sub-

set of a domain, i.e. of a non-empty subset of the order in which each two elements

have some upper bound [23]]. In view of our intuition about domains, this means

that every two pieces of information within the directed subset are consistently ex-

vi
tended by some other element in the subset. Hence we can view directed sets as

consistent specifications, i.e. as sets of partial results in which no two elements are

contradictory [23].

In fact, the concept of the directed set paves to the fundamental concept in this

research namely, the directed-complete poset (in short, dcpo) which occupies a large

area in this research.

In this thesis, we survey the general topological concepts for the Scott topology

-which is of fundamental importance in domain theory since it lies at the heart of the

structure of domains([12])-. Also, this research compares between the Scott topol-

ogy and one of the Alexandroff topology types; namely the T0 -Alexandroff topology.

This thesis consists of four chapters: the Preliminaries, the Alexandroff space,

the Scott topology and finally the Scott topology and approximation relation.

In the first chapter - which is divided into three sections-, we begin the first

section with an essential definition; the definition of the partial order relation and

consequently the definition of a poset . Then, we pave the way to represent, diagram-

matically, the elements of the poset according to its partial order. The definitions of

some terminologies related to the whole set or to the position of the elements within

it are given, such as: maximal, maximum, minimal, infimum, up set (or upper set

as some authors prefer) or down set . . . etc.

In the second section, we give the definition of the directed-complete poset (briefly,

dcpo ) which is of special importance in the last two chapters. Also, we introduce

the definition of the algebraic dcpo. In the third section,we introduce some related

topological concepts; since this is a topological issue and the topological concepts

vii
will naturally be used.

The second chapter consists of three sections. The first paves the way to the

Alexandroff topology by a quick historical introduction including the definition of

the Alexandroff topology as a special class of topologies and then its specialization

order.

The second section goes more closer and gives the definitions of the open sets and

closed sets in the T0 -Alexandroff space. The third section of this chapter gives the

concepts of the interior, closure, the boundary and the derived set of a subset of this

space.

The third chapter-which is the main one- is divided into two sections. The first,

gives the definition of the Scott-open set and then shows that the collection of all

Scott open sets forms a topology called the Scott topology. Also, it shows that the

Scott topology is sober over an algebraic dcpo. The base of the Scott topology is

given by means of the set of all compact elements. The second compares between

the Scott topology and the Alexandroff topology on finite sets and in general.

The last chapter with its three sections gives the definitions of an important

concepts. The first section introduces the concept of the approximation (or the

way-below) relation and some of its fundamental properties. By the use of this es-

sential concept, the definition of the continuous poset comes in the second section.

Finally, the base of the Scott topology on a poset P is defined on the collection of

the approximation relation of each element in the poset P . Also the base of the

Scott topology is defined by the use of filters.

At the end of this introduction, we would like to say that collecting the items

viii
of this subject wasn’t so easy since we have had no paper in our hands talking ex-

plicitly about the Scott topology. What we have found was some definitions and

propositions. Most of what has been found was without proof, and if there was any,

it was in short and need to be reproved again. So, most of the proofs in this research

has been written by us. Not only this but also the rearrangements of the research;

naming the chapters and the sections together with the comments in the beginning

of the chapters and the sections and between lines. It was not so easy to accomplish

all these things without the help of the supervisor.

ix
Chapter 1

Preliminaries

The interaction between topology and order theory has a plenty soil over computer

science. Domain theory, where this reaction happens, deals with a special relation

over a set P and this relation orders the elements of P . This relation is called a

partial order and this together with the set under consideration is called a poset.

These definitions and more will be given in the current chapter together with some

related topological concepts in order to be ready to study the next chapters.

1.1 Partially Ordered Sets

Definition 1.1.1. [18] A relation ≤ on a set P is called partial order (simply order )
on P if for every a, b, c ∈ P :
(i) a ≤ a ( reflexivity),
(ii) a ≤ b and b ≤ a implies a = b ( anti-symmetry),
(iii) a ≤ b and b ≤ c implies a ≤ c (transitivity).
The set P together with a partial order ≤ is called a partially ordered set ( briefly
a poset).

1
Example 1.1.1. The set N of all natural numbers forms a poset under the usual
order ≤ on R. Similarly, the set of integers Z, rationales Q and real numbers R
under the usual order ≤ form posets.

Example 1.1.2. Let X be a set. The set P(X) of all subsets of X under the relation
”contained in” (⊆) forms a poset.

Diagrammatical representation of a poset:

Each poset can be represented by the help of a diagram. To draw the diagram of a

poset, we represent each element by a small circle (a dot ) and any two comparable

elements are joined by lines in such a way that if a ≤ b then a lies below b in the

diagram. Non-comparable elements are not joined. Thus, there will not be any

horizontal lines in the diagram of a poset.

Example 1.1.3. The set {2, 3, 4, 6} under divisibility relation forms a poset with a
diagram as below:

Example 1.1.4. If X = {1, 2, 3}, then the poset P(X) = {φ, {1}, {2}, {3}, {1, 2}, {1, 3},
{2, 3}, {1, 2, 3}} under ⊆ relation is represented by the diagram below.

2
Example 1.1.5. Let X = {a, b, c, d, e}. Then, the diagram below defines a partial
order on X as follows :
x ≤ y if and only if x = y or one can go from x to y in the diagram in the indicated
direction; i.e., upward.

Definitions 1.1.2. [18]and [5] 1. Let a, b be two elements in an ordered set. We

say that a precedes b ( or a is smaller than b) if a ≤ b. In this case we say that b
follows a or larger than a. Furthermore, we write a < b if a ≤ b and a 6= b.
2. A subset C of a poset P is a chain if any two elements of C are comparable.
Alternative names for a chain are linearly ordered set and totally ordered set. Thus,
if C is a chain and x, y ∈ C then either x ≤ y or y ≤ x.
3. If a relation R on a set A (which is a subset R of A × A) defines a partial order,
then the inverse relation R−1 is also a partial order; it is called the inverse order (or
the dual order ).

Definition 1.1.3. [18] Let A be a subset of a poset X. Then, the order in X induces
an order in A in the following natural way: If a, b ∈ A, then a ≤ b as elements in A
if and only if a ≤ b as elements in X.
Equivalently, if R is a partial order in X, then the relation RA = R ∩ (A × A) -
called the restriction of R to A - is a partial order in A. The ordered set(A, RA ) is
called partially ordered subset of the ordered set (X, R).
It should be noted that a chain C as an ordered subset of X is totally ordered.
Clearly, if X is totally ordered, then every subset of X will be totally ordered.

3
Definition 1.1.4. [18] Let X be an ordered set. An element a ∈ X is called
maximal if whenever a ≤ x then x = a, that is; if no element in X follows a except
a. Similarly, an element a ∈ X is called minimal if whenever x ≤ a then x = a,
that is; if no element in X precedes a except a itself.

We denote the set of all maximal (resp. minimal) elements of an ordered set X

by M (resp. m). If A is any subset of X, we write M (A) (resp. m(A)) to denote

the set of maximal (resp. minimal) elements of A under the induced order.

If there is an element > ∈ X such that x ≤ > for all x ∈ X, then > is called

maximum (or top) element. On the other hand, if there is an element ⊥ ∈ X such

that ⊥ ≤ x for all x ∈ X, then ⊥ is called minimum (or bottom) element.

It should be noted that the set M of all maximal elements of a poset X may be an

empty set. In the case where |M | = 1 (M contains only one element), the set X has

a top element >. Dually, if |m| = 1, then X has a bottom element ⊥.

Example 1.1.6. Recall Example 1.1.5. The element “ a” is maximal while both d
and e are minimal elements.

Definitions 1.1.5. [18] Let A be a subset of a poset X. An element u ∈ X is an

upper bound of A if x ≤ u ∀ x ∈ A. The least upper bound (or the supremum)
W
of A - denoted by sup A (or A)- is an upper bound that precedes each upper
bound of A. An element ` ∈ X is a lower bound of A if ` ≤ x ∀ x ∈ A. The
V
greatest lower bound (or the infimum) of A - denoted by inf A (or A)- is a lower
bound of A that follows each lower bound of A. For a subset A, sup A and inf A
may not exist. A is said to be bounded above if it has an upper bound, and bounded
below if it has a lower bound. If A has both upper and lower bounds, then it is
bounded. For a subset A of a poset X, the set of all upper (resp. lower) bounds of
A is denoted by Au ( resp. A` ).

4
Example 1.1.7. [18] Let X = {a, b, c, d, e, f, g} be a set ordered by the following
diagram:

Let B = {c, d, e}. The elements a, b and c are upper bounds of B and f is the only
lower bound of B. The element g is not a lower bound of B, since g doesn’t precede
d. Moreover, inf B = f ∈
/ B, while sup B = c ∈ B.

Definition 1.1.6. [5] We say y covers x or x is covered by y and write y  x or

x  y if x < y and when x ≤ z < y then z = x.

Definition 1.1.7. [5] A subset O of a poset P is a down set (or lower set) if,
whenever x ∈ O and y ≤ x, then we have y ∈ O. On the other hand, a subset
U of a poset P is an up set (or upper set) if, whenever x ∈ U and x ≤ y, we
have y ∈ U . For x ∈ P , we define the down set ↓ x = {y ∈ P : y ≤ x}; and
the up set ↑ x = {y ∈ P : x ≤ y}. For a set B ⊆ P , we define the down set
↓ B = {y ∈ P : (∃x ∈ B) y ≤ x} and the up set ↑ B = {y ∈ P : (∃ x ∈ B) x ≤ y}.
In this case, ↑ x =↑ {x} and ↓ x =↓ {x}.

If A is a down set of P , then the complement Ac is an up set, since if a ∈ Ac and

a ≤ b , then b ∈ Ac by a contrapositive argument.

5
Example 1.1.8. Let R be the set of all real numbers with its usual order. Let
A, B ⊆ R be such that A = [3, ∞) and B = (−∞, 0], then A is an up set and B is
a down set.

Definition 1.1.8. [5] A poset P satisfies the ascending chain condition (ACC) , if
for any increasing sequence x1 ≤ x2 ≤ . . . ≤ xn ≤ . . . in P , there exists k ∈ N such
that xk = xk+1 = . . .. The dual of (ACC) is the descending chain condition (DCC).
If a poset satisfies both ACC and DCC, we say P is of finite chain condition (F CC).

Example 1.1.9. A collection of subsets of a finite set X when ordered by inclusion

satisfies the ACC and if ordered by reverse inclusion it satisfies the DCC.

Example 1.1.10. Each finite poset is of F CC.

1.2 Directed - complete posets (dcpo)

A special kind of posets is considered in this section, and its importance will appear

later. Here is the definition of this kind:

Definition 1.2.1. [4] Let (D, ≤) be a partially ordered set. A subset U of D is

called directed if U is inhabited (i.e., U 6= φ) and ∀ u, v ∈ U, ∃ w ∈ U such that
u ≤ w and v ≤ w.

Example 1.2.1. Recall Example 1.1.2 for a non-empty set X, we have that (P(X), ⊆)
is a directed set, since for any non-empty subsets A, B ∈ P(X), take C = A ∪ B for
the second condition.

Lemma 1.2.2. [10] Let P be a poset. A non-empty chain in P is directed.

Proof. Let P be a poset and let U be a non-empty chain in P . Let u, v ∈ U . Since

in the chain, each two elements are comparable, then u ≤ v or v ≤ u. If u ≤ v,
then u ≤ v and v ≤ v. Similarly, if v ≤ u, we have v ≤ u and u ≤ u. Thus, U is
directed.

6
Example 1.2.2. The set of natural numbers N, the set of integers Z, the set of
rationales Q and the set of real numbers R are directed sets under the usual order.

Lemma 1.2.3. Let P be a poset. Then, for any x ∈ P , the set

↓ x = {y ∈ P : y ≤ x}

is directed with x as its join.

Proof. For any y, z ∈↓ x, we have y ≤ x and z ≤ x. Thus ↓ x is directed. Since

W
y ≤ x for each y ∈↓ x, then x is an upper bound in ↓ x, and hence x = ↓ x.

Proposition 1.2.4. In a finite poset P , a subset has a top element “ > ” if and
only if it is directed.

Proof. Let P be a finite poset.

(⇒) Let U ⊆ P be a non-empty subset with a top element >U . Then for any u ∈ U ,
u ≤ >U . Consequently, ∀ u, v ∈ U , take w = >U ∈ U so that u ≤ w and v ≤ w.
Thus, U is directed.
(⇐) Let U ⊆ P be a directed subset. Then, U 6= φ. Since P is finite, then so
is U . Let U = {u1 , u2 , . . . , un }. Now, for any ui , uj ∈ U, ∃ uk ∈ U such that
ui ≤ uk and uj ≤ uk . Also, for any um ∈ U, ∃ uw ∈ U such that uk ≤ uw and
um ≤ uw . Thus, by the transitivity of “ ≤ ” and the directedness of U we have
uw = max{ui , uj , uk , um , uw }. Continuing in this fashion our process must come
to an end since U is finite. That is; there must be an element u ∈ U such that
u = max{u1 , u2 , . . . , un }. Hence, U has a top element.

Definitions 1.2.5. [4] (1) A partially ordered set D is called directed-complete

(briefly dcpo) if every directed subset has supremum.
(2) A complete partially order set (briefly cpo) is a dcpo with a least element.

Example 1.2.3. [21] Every finite poset is a dcpo.

Proof. By Proposition 1.2.4, in finite posets, each directed subset has a top element
and hence, has a supremum. Thus, we are done.

7
Example 1.2.4. The set of real numbers R, the set of rationales Q, the set of natural
numbers N and the set of integers Z fail to be dcpos under the usual order, since all
these sets are directed subsets of themselves and no one has a supremum. Moreover,
each finite subset of any of them is a dcpo under the usual order. Furthermore, each
finite subset is a cpo.

The following definition plays a big rule in the later chapters.

Definition 1.2.6. [4] An element a of a dcpo D is called compact if, for any directed
W
subset U of D, a ≤ U implies that ∃u ∈ U such that a ≤ u (i.e.,↑ a ∩ U 6= φ).
The set of all compact elements of D will be denoted by KD

Lemma 1.2.7. [4] Whenever it exists, the supremum of any finite set of compact
elements is compact.

Proof. Let D be a dcpo and let A = {ai }ni=1 be a finite set of compact elements in D.
Suppose that the supremum b ∈ D of A exists. By the definition of the supremum,
W
we have ai ≤ b ∀ ai ∈ A. Now, let U be any directed subset of D such that b ≤ U .
W
So, we have ai ≤ b ≤ U ∀ ai ∈ A. Since ai is compact ∀ i, there exists ui ∈ U
such that ai ≤ ui for all i = 1, 2, 3, . . . , n. Let u = max{u1 , u2 , . . . , un }. Then, u
exists in U by the argument of Proposition 1.2.4. Thus, ai ≤ u ∀ ai in A. Hence u
is an upper bound of A. Since b is the least upper bound of A, then b ≤ u and so b
is compact.

Proposition 1.2.8. In a dcpo,D, if each directed subset of D contains its supremum,

then KD = D.

Proof. Clearly KD ⊆ D. Now, let a ∈ D and let U be a directed subset of D such

W W W
that a ≤ U . Since U ∈ U (by hypothesis), then take u = U ∈ U and so
a ≤ u. Therefor, a is compact.

8
Corollary 1.2.9. For each finite subset B of N (or Z), KB = B.

Proof. By Example 1.2.3, B is a dcpo. Since any finite subset of N (or Z) contains
its supremum, then from Proposition 1.2.8, B = KB .

Proposition 1.2.10. Let P be a poset that satisfies ACC and let A ⊆ P . Then, A
is directed if and only if | M (A) | =1. That is; A is directed if and only if A has a
supremum.

Proof. Let A be a non-empty subset of a poset P where P satisfies the ACC.

(⇒) Let x1 , x2 ∈ M (A) ⊆ A. So, there exists u ∈ A such that x1 ≤ u and x2 ≤ u.
But x1 , x2 are maximal in A. So x1 = u and x2 = u. Therefore, x1 = x1 .
W
(⇐) Suppose that | M (A) | = 1. So, A exists in A. Therefore, ∀ x, y ∈ A, x ≤ z
W
and y ≤ z, where z = A. Hence, A is directed.

Definition 1.2.11. [4] A dcpo D is called algebraic if, for every x ∈ D, the set
W
↓K x = {a ∈ KD : a ≤ x} is directed and x = ↓K x. Alternately, algebraic dcpo’s
are referred to as domains(see [6]).

Lemma 1.2.12. In an algebraic dcpo D, the set ↓K x is non-empty for each x ∈ D.

Proof. Since D is an algebraic dcpo, then for each x ∈ D, the set ↓K x is directed
and hence ↓K x 6= φ.

Example 1.2.5. Let P = {1, 2, 3, . . . , n}, where n ∈ N, with natural order. Then,
P is an algebraic dcpo.

Proof. Clearly, P is a finite poset and hence is a dcpo. Also, P = KP (see Corollary
1.2.9). Therefore, for any x ∈ P , ↓K x = ↓ x = {m ∈ N : m ≤ x} which is
directed and has x as its join (see Lemma 1.2.3). Hence, P is an algebraic dcpo.

Example 1.2.6. Let D = (−∞, 1] with the usual order. It is clear that D is a dcpo.
Moreover, if y ∈ D and for x ≤ y, let U be the interval of real numbers (x, y), which
W W
is a directed subset of D with y = U . So, y ≤ U = y and there is no u ∈ U

9
such that y ≤ u. Therefore, y is not compact and consequently, KD = φ. Hence,
Dis not algebraic.

The following example shows a non-algebraic dcpo with non-empty set of com-

pact elements.

Example 1.2.7. Let P = [0, 1] with usual order. Then, P is a dcpo. Since 0 is the
W
bottom element in P , then for any directed subset U of P with 0 ≤ U , there is
u ∈ U such that 0 ≤ u (U 6= φ since it is directed). Thus, 0 ∈ KP .
W
Now, for any x ∈ P with x 6= 0, we have U = (z, x), z ∈ P is directed with U =x
but U contains no element u such that x ≤ u. Thus, x ∈
/ KP . Hence, KP = {0}
and so ∀ x ∈ P, ↓K x = {0} which is directed with 0 as its join ; i.e., x is not its
join except for x = 0 . Thus, P is not algebraic.

Corollary 1.2.13. If D is a finite algebraic dcpo, then each element in D is compact.

That is; D = KD .

Proof. Let D be a finite algebraic dcpo and let x ∈ D. Then, ↓K x is a finite

directed subset of compact elements with x as its join. Thus, by Lemma 1.2.7, x is
compact.

As a generalization of Example 1.2.5, we have the following lemma:

Lemma 1.2.14. Any finite linearly ordered set is an algebraic dcpo.

Proof. Let P be a finite linearly ordered set. Then P is a dcpo and P = KP . Thus,
for any x ∈ P, ↓K x =↓ x which is a directed set with x as its join. Hence, P is an
algebraic dcpo.

In any algebraic dcpo D, the ordering relation between its elements can be

recovered from the structure of KD .

10
Proposition 1.2.15. [4] Let D be an algebraic dcpo. For x, y ∈ D, x ≤ y if and
only if ↓K x ⊆↓K y, that is; x ≤ y if and only if ∀ a ∈ KD , a ≤ x implies a ≤ y.

Proof. ( ⇒ ) Suppose that x ≤ y and let a ∈↓K x. Then, a ≤ x. Since x ≤ y, then

a ≤ y, that is; a ∈↓K y. Thus, ↓K x ⊆↓K y.
W W
(⇐) Suppose that ↓K x ⊆↓K y. Then ↓K x ≤ ↓K y. Since D is algebraic, so
W W
↓K x = x and ↓K y = y and hence, x ≤ y.

Before we close this section let us pass on to the dual of the notion “ directed

set”, namely the “ filtered set”.

Definition 1.2.16. [7] A non-empty subset S of a poset P is said to be filtered if

given x, y ∈ S, there exists z ∈ S such that z ≤ x and z ≤ y.

Example 1.2.8. Any subset of N (Z, Q and R) is a filtered set under the usual
order. In general, each chain is a filtered set.

Lemma 1.2.17. If P is any poset with bottom element ⊥, then any subset of P
containing ⊥ is a filtered set.

Proof. Let P be a poset with bottom element ⊥ and let S be any subset of P such
that ⊥ ∈ S. Then, for any x, y ∈ S, ⊥ ≤ x and ⊥ ≤ y. Hence, S is a filtered
set.

Corollary 1.2.18. Every cpo is a filtered set.

Proof. Straightforward (See Definition 1.2.5 part (2)).

Definition 1.2.19. [7] Let P be a poset. A non-empty subset F of P is called a

filter if F is a filtered up set.

Example 1.2.9. Let the poset D = [2, 3] ∪ {1} and suppose that the elements of
D are ordered as follows: the elements of the closed interval [2, 3] are ordered by
the usual order “ ≤”. For 1 and any x ∈ [2, 3) we have xk1 (i.e. x and 1 are

11
incomparable). Finally, if x = 3 then 1 ≤ 3.
We will use the broken line to represent the usual order for all the entries of the
interval, while the line segment to represent the order between two elements only.

Now, let S = [a, 3] ⊆ [2, 3], where a < 3. Clearly S is a filter. The set U =
[2, 2.5] ∪ {3} is a filtered set which isn’t a filter (since it isn’t an up set). Consider
the set A ⊆ D − {3}. Then 1 ∈ A and A contains more than one point. Moreover,
A is not a filtered set, since for any x ∈ A − {1}, there is no element z in A such
that z ≤ x and z ≤ 1.

Example 1.2.10. Let P = N, the set of all natural numbers, be ordered as follows:
1 ≤ 2 ≤ 4 ≤ 6 ≤ . . . and 1 ≤ 3 ≤ 5 ≤ 7 ≤ . . ., and x k y ∀ x ∈ {2, 4, 6, . . .} and
y ∈ {3, 5, 7, . . .}. Clearly P is a filter.

Definition 1.2.20. [7] A non-empty subset I of a partially ordered set (P, ≤) is an

ideal, if the following conditions hold:
(i) I is a down set. That is; for every x ∈ I, y ≤ x implies that y is in I.
(ii) I is a directed set. That is; for every x, y ∈ I, there is some element z ∈ I, such
that x ≤ z and y ≤ z.

12
Example 1.2.11. Let P = N, the set of all natural numbers, be ordered by the
reverse order of that in Example 1.2.10. Clearly P is an ideal.

1.3 Topological concepts

Many topological concepts will be used in the next chapters, so we give here their

definitions to keep the next chapters for the discussion of the new and related con-

cepts.

We are going to introduce first a structure on a set X by use of a collection of subsets

of X. For this, we shall soon set forth a set of axioms which a collection of subsets

must obey in order to fall within the circle of our studies. Any collection of subsets

of X satisfying these axioms will be called a topology of X[15].

Definition 1.3.1. [15] Let X 6= φ be a set. Then a topology on X is a subset τ of

P(X)(the power set of X) obeying the following axioms:
(a) X and φ belong to τ .
(b) If U1 and U2 belongs to τ , then U1 ∩ U2 belongs to τ .
(c) If {Uα : α ∈ ∆} is an indexed family of sets, each of which belong to τ , then
[
Uα belongs to τ .
α∈∆

We shall call the elements of a topology on any set X, open subsets of X.

13
Definition 1.3.2. [15] A topological space is a set X together with a topology τ on
X. The notion (X, τ ) will often be used for a topological space, but the shortened
notation the space X will also be used when no confusion arises concerning the
topology on X.

When R is the set under consideration, the standard topology will always be

assumed, unless otherwise explicitly stated.

Definition 1.3.3. [19] Given two topologies τ1 and τ2 on a set X. We say that τ1
is weaker (smaller,coarser) than τ2 , or τ2 is stronger(larger,finer) than τ1 if τ1 ⊆ τ2 .

Example 1.3.1. The left ray topology on R is coarser than the usual topology, since
each set of the form (−∞, a) which is open in the left ray topology, is also open in
the usual topology while the set A = (1, 9) belongs to the usual topology but not to
the left ray topology.

Definition 1.3.4. [19] If X is a topological space and E ⊆ X, we say E is closed if

X − E is open.

Theorem 1.3.5. [19] If F is the collection of closed sets in a topological space X,

then
(a) X and φ both belong to F,
(b) any intersection of members of F belongs to F,
(c) any finite union of members of F belongs to F.

Definition 1.3.6. [19] If X is a topological space and E ⊆ X, the closure of E in

X is the set
\
E = Cl(E) = {K : K is a closed set containing E}.

When confusion is possible as to what space the closure is to be taken in, we will

write ClX (E).

14
Remark 1.3.7. : By part (b) of Theorem 1.3.5, it is clear that E is closed and it is
the smallest closed set containing E. The intersection of all closed sets containing E
is closed, where the precise meaning of “smallest” is that if the sets containing E is
ordered by K1 ≤ K2 if f K1 ⊆ K2 .

Theorem 1.3.8. [15] A subset A of a space X is closed if and only if A = A.

Some properties of the closure are now considered:

Theorem 1.3.9. [15] Let A and B be subsets of the space X. Then

(a) φ = φ.
(b) A ∪ B = A ∪ B.
(c) A = A.
(d) If A ⊆ B, then A ⊆ B.

Definition 1.3.10. [15] Let (X, τ ) be a topological space and A ⊆ X. A point

x ∈ X is an interior point of A if there exists an open set U containing x such that
U ⊆ A. The set of interior points of A is called the interior of A and is denoted by
Int(A) (or Ao ). A point x ∈ X is an exterior point of A if there exists an open set
U containing x such that U ∩ A = φ. The set of exterior points of A is called the
exterior of A and is denoted by Ext(A) (or(X − A)o ).A point x ∈ X is a boundary
point of A if every open set in X containing x contains at least one point of A, and
at least one point of X − A. The set of boundary points of A is called the boundary
of A and is denoted by Bd(A) (or F rX (A) ).

Definition 1.3.11. [15] Let X be a topological space, x ∈ X, and A ⊆ X. Then x is

a cluster point (or accumulation point, limit point) of A if every open set containing
x contains at least one point of A different from x. For any set A in the space X,
the set of all cluster points of A is called the derived set of A. The derived set of A
0
is denoted by A .

15
 0
We can write x ∈ A if and only if ∀ U ∈ τ such that x ∈ U, U ∩ (A − {x}) 6= φ.

The relation between the closure and the derived sets of a set A is introduced in the

following theorem:

0
Theorem 1.3.12. [19] A = A ∪ A .

Definition 1.3.13. [15] Let X be a topological space. Then, A ⊆ X is dense in X

if A = X.

Definition 1.3.14. [18] A subset A of a topological space X is said to be nowhere

o
dense in X if the interior of the closure of A is empty, i.e., (A) = φ.

Definition 1.3.15. [15] Let (X, τ ) be a topological space. A base for τ is a collection
B of subsets of X such that:
(a) each member of B is also a member of τ and
(b) if U ∈ τ and U 6= φ, then U is the union of sets belonging to B.

Since B ⊆ τ , and using part (b) of the definition provided that U 6= φ, we have

that, if U 6= φ, then U ∈ τ if and only if U is the union of members of B. Therefore,

a base for τ completely determines τ by arbitrary unions of members of B. Also we

see that any topology is a base for itself. So, any topology has at least one base.

Theorem 1.3.16. [19] B is a base for a topology on X if and only if:

[
(a) X = B and
B∈B
(b) whenever B1 , B2 ∈ B with p ∈ B1 ∩ B2 , there is some B3 ∈ B with

p ∈ B3 ⊆ B1 ∩ B2 .

Definition 1.3.17. [19] If X is a topological space and x ∈ X, a neighborhood

(abbreviated nhood ) of x is a set U which contains an open set V containing x.
Thus, evidently, U is a nhood of x if and only if x ∈ U o . The collection Ux of all
nhoods of x is called the nhood system of x.

16
Definition 1.3.18. [19] A nhood base at x in the topological space X is a subcol-
lection Bx taken from the nhood system Ux having the property that each U ∈ Ux
contains some V ∈ Bx .

Theorem 1.3.19. [19] Let X be a topological space and suppose a nhood base has
been fixed at each x ∈ X. Then,

A = {x ∈ X : each basic nhood of x meets A}

Definition 1.3.20. [19] If (X, τ ) is a topological space and A ⊆ X, the collection

τA = {G ∩ A : G ∈ τ } is a topology on A, called the relative topology of A. This
topological space is denoted by (A, τA ).
The fact that a subset of X is being given this topology is signified by referring to
it as a subspace of X.
When a topology is used on a subset of a topological space without explicitly being
described, it is assumed to be the relative topology.

Definition 1.3.21. [15] Let(X, τ ) be a topological space. Then,

(i) the space (X, τ ) is a To -space if for each pair of distinct points x, y ∈ X, there is
either an open set containing x but not y or an open set containing y but not x,
(ii) the space (X, τ ) is a T1 -space if for each pair of distinct points x, y ∈ X, there
is an open set in X containing x but not y and an open set in X containing y but
not x ,
(iii) the space (X, τ ) is called a T2 -space ( or a Hausdorff space) if for each pair of
distinct points x, y ∈ X, there are disjoint open sets U and V in X with x ∈ U and
y ∈V.

In fact, (i),(ii) and (iii) in the above definition are always given within a package

known as the Separation Axioms.

Example 1.3.2. The indiscrete topology on a set is not a To -space, while the discrete
topology on a set is not only To but also T1 and T2 (singletons are open in the discrete
topology).

17
Remark 1.3.22. Every T1 -space is a T0 -space and every T2 -space is a T1 -space.

Definition 1.3.23. [15] Let {Aα : α ∈ ∆} be a family of subsets of the space X

[
and B ⊆ X. The family {Aα : α ∈ ∆} covers B if B ⊆ Aα . If ∆ is finite and
α∈∆
{Aα : α ∈ ∆} covers B, then {Aα : α ∈ ∆} is called a finite cover of B. If each
Aα , α ∈ ∆, is open (closed) in X and {Aα : α ∈ ∆} covers B, then {Aα : α ∈ ∆}
is called an open (closed) cover of B.

Definition 1.3.24. [15] Let {Aα : α ∈ ∆} be a cover of B ⊆ X. Then the family

{Aβ : β ∈ Ω ⊆ ∆} is a subcover of {Aα : α ∈ ∆} of B if {Aβ : β ∈ Ω ⊆ ∆} is a
cover of B.

Definition 1.3.25. [15] A space X is called compact if each open cover of X has
a finite subcover. A subset A of the space (X, τ ) is compact if the space (A, τA ) is
compact.
Indeed, to prove a space is not compact, we need only exhibit one open cover which
has no finite subcover.

Definition 1.3.26. [18] A space X is locally compact if and only if every point in
X has a compact nhood.

Proposition 1.3.27. Every compact space is locally compact.

Theorem 1.3.28. [22] Let B be a base for a topological space X. Then, X is

compact if and only if each cover {Bα ∈ B : α ∈ ∆} of X can be reduced to a finite
subcover.

18
Chapter 2

Alexandroff Space

2.1 Introduction

In this section we study a class of topological spaces called To -Alexandroff spaces.

An Alexandroff space is a topological space (X, τ ) that satisfies the property that

a finite intersection of open sets is open. What about an arbitrary intersection of

open sets?. This property doesn’t hold in all topological spaces. For example, in

the standard topology on R,

\ 1 1
(− , ) = {0}
n∈N
n n

which is not open in R.

In fact, this property holds in a special class of topological spaces called Alexandroff

spaces. This subject was first studied in 1937 by P. Alexandroff [14] under the name

of Diskrete Räume (discrete space). The name is not valid now, since a discrete space

is a space where the singletons are open. He gave an example of a To -Alexandroff

space on a poset(P, ≤) taking B = {↑ x : x ∈ P } to be the unique minimal base.

The induced topology on P -denoted by τ (≤)- is a To -Alexandroff space. If (X, τ )

19
is an Alexandroff space, he defined its (Alexandroff ) specialization order ≤τ on X

as follows:

∀ a, b ∈ X, a ≤τ b if a ∈ {b}. (2.1.1)

The specialization order is reflexive and transitive. It is antisymmetric - and hence

a partial order - if and only if X is To . Moreover, if (X, ≤) is a poset and if τ (≤)

is its induced To -Alexandroff topology, then the specialization order of τ (≤) is the

order ≤ itself, i.e.; ≤τ (≤) =≤ .

On the other hand, if (X, τ ) is a To -Alexandroff space with the specialization order

≤τ , then the induced topology by the specialization order is the original topology,i.e.;

τ (≤τ ) = τ [2]. Therefore, To -Alexandroff spaces are completely determined by their

specialization orders.

Note 2.1.1. See for example: Example 2.2.5 and Example 2.2.6.

Here is a general definition for the specialization order of a topological space:

Definition 2.1.2. [10] For the topological space (X, τ ) we define the specialization
order ≤τ on X, for any x, y ∈ X, by

x ≤τ y if and only if ∀ O ∈ τ, x ∈ O implies y ∈ O.

Here, we are interested in Alexandroff spaces that satisfy the separation axiom

To . We use their specialization orders in proofs to illustrate the results and the con-

cepts. The importance of this study comes from the fact that we can characterize

topological properties just by looking at its specializing order (poset).

For example, if we define a topological space X to be submaximal if each dense sub-

set is open, then for the To -Alexandroff space X, X is submaximal if each element in

20
the corresponding poset -the space X after being ordered by the specialization order

≤τ - is either maximal or minimal, i.e., the graph of its corresponding poset contains

two rows; the row of the maximal elements and the row of the minimal elements([5]).

2.2 A glimpse of Alexandroff space

Now, let us go forward to study this class of topological spaces. First of all, let’s

recall the definitions of the neighborhood and nhood base of a point x in a space X,

namely Definition 1.3.17 and Definition 1.3.18.

Remark 2.2.1. If the intersection of all the nhoods of x ∈ X exists, then the set
\
V (x) = U is the smallest basic nhood of x (since, if W is any open set
U ∈τ ,x∈U
\
such that x ∈ W , then V (x) = U ⊆ W ). So, the collection Bx with only one
U ∈τ ,x∈U
element, Bx = {V (x)}, will be called the minimal nhood base of x.

Definition 2.2.2. [5] Let X be a T0 -Alexandroff space. Then, for each x ∈ X, ↑ x

or V (x) will denote the minimal basic nhood of x.

Definition 2.2.3. [14] An Alexandroff space X is a space in which any arbitrary

intersection of open sets is open.

Remark 2.2.4. An equivalent statement of that in the above definition is that: an

Alexandroff space X is a space in which each singleton has a minimal nhood base.
T
Proof. (⇒)Suppose that arbitrary intersection of open sets is open. That is; U ∈τ
for all U ∈ τ . Let Bx = {U ∈ τ : x ∈ U } be a nhood base of x ∈ X. Then,
\
V (x) = U is open. So, V (x) ⊆ U for all U ∈ Bx . Thus, B́x = {V (x)} is the
U ∈τ ,x∈U
minimal nhood base of x.

21
(⇐) Suppose that for each x ∈ X, x has a minimal nhood base. That is; ∀ x ∈
X, ∃ Bx = {V (x)}, a minimal nhood base of x. Let {Uα : α ∈ ∆} be the collection of
\
all open sets in X. Let y ∈ Uα . Then, y ∈ Uα for all α ∈ ∆. So, y ∈ V (y) ⊆ Uα
α∈∆ \ \
for all α ∈ ∆. Therefore, V (y) ⊆ Uα . Thus, y ∈ V (y) ⊆ Uα . So, this
α∈∆ α∈∆
intersection is open. Hence, arbitrary intersection of open sets is open.

Example 2.2.1. It is easy to check that the discrete topology on any non-empty set
is Alexandroff.

Lemma 2.2.5. [5] Any finite space is Alexandroff.

Proof. It is obvious that any finite space is Alexandroff since any finite space has
finite number of subsets and consequentially finite number of open sets. So, arbitrary
intersections of these finite number of open sets is open.

Definition 2.2.6. [5] A topological space (X, τ ) is called locally finite if each element
x of X is contained in a finite open set and a finite closed set.

Proposition 2.2.7. [5] Any finite space is locally finite.

Proof. Let (X, τ ) be a finite space. Since X is both open and closed set, then each
element in X is contained in a finite open set and a finite closed set. Hence, X is a
locally finite space.

Corollary 2.2.8. [5] Each T0 -locally finite space is a T0 -Alexandroff space.

Proof. It is enough to show that every locally-finite space is Alexandroff space. So,
let X be a locally finite space. Then, each x ∈ X is contained in an open set and
hence has a basic nhood. That is; x has a minimal nhood base.

The converse of Corollary 2.2.8 need not be true. Here is a counterexample:

Example 2.2.2. Let P = N ∪ {⊥} and let P be ordered as indicated in any of

the two diagrams. Clearly, P is a poset. Then, the T0 -topology defined on P with

22
minimal base B = {↑ x : x ∈ P } is a T0 -Alexandroff (see Note 2.1.1) which is not
a T0 -locally finite, since (see diagram (a)) for each a ∈ P such that 3 ≤ a, ↑ a is
not finite. Also (as in diagram (b)), ↑ ⊥ = P which is not finite.

It is worth noting that, there is a definition of similar name - locally finite- on a

poset, but before introducing it, we have the following definition:

Definition 2.2.9. [22] Let P be a poset. A subset I of P is called a poset interval,

or simply an interval if there exist elements a, b ∈ P such that

I = {t ∈ P : a ≤ t ≤ b} = [a, b].

The elements a, b are called the endpoints of I. Clearly a, b ∈ I. Also, the endpoints
of a poset interval are unique. That is; if [a, b] = [c, d], then a = c and b = d (To see
this: Let I = {t ∈ P : a ≤ t ≤ b} = [a, b] = [c, d].) Since c, d ∈ I, then a ≤ c and
d ≤ b. Similarly, we have c ≤ a and b ≤ d. Thus, a = c and b = d).

Remark 2.2.10. [22] It is easy to see that the name is derived from that of an interval
on a number line. From this analogy, one can easily define poset intervals without
one or both endpoints. Whereas an interval on a number line is linearly ordered, a
poset interval in general is not.

Definition 2.2.11. [22] A poset P is called locally finite if every interval [x, y] in
P is finite.

23
Corollary 2.2.12. Every finite poset is locally finite.

The converse of this corollary need not be true. The following example is a

counterexample:

Example 2.2.3. The set Z of integers with the usual order is a locally finite poset
but not finite, while Q is neither.

Example 2.2.4. Recall Example 2.2.2. It is clear that the poset P is locally finite
as a poset but not locally finite as a space.

Definition 2.2.13. [5] A To -Alexandroff space whose corresponding poset satisfy

the ACC is called Artinian To -Alexandroff spaces.

Throughout this section, the symbol (X, τ (≤)) denotes the To -Alexandroff space

where ≤ is its (Alexandroff) specialization order.

Now, it is time to be more closer to the To -Alexandroff spaces.

Theorem 2.2.14. [5] If (X, τ (≤)) is a To -Alexandroff space then a subset A of X

is open if and only if it is an up set with respect to the specialization order; that is,
A is open if and only if A =↑ A. And A is closed if and only if it is a down set;
that is, A is closed if and only if A =↓ A.

Proof. (1) For the first statement:

(⇒) Suppose that A ∈ τ (≤) and let x ∈ A. Let y ∈ X such that x ≤ y. Since
B = {↑ x : x ∈ X} is the minimal nhood base of x, then x ∈↑ x ⊆ A. But y ∈↑ x,
so y ∈ A. Thus, A is an up set.
(⇐) Suppose that A is an up set. Let x ∈ A. Since A is up set, then we have
x ∈↑ x ⊆↑ A = A. Since ↑ x ∈ B = {↑ x : x ∈ X}, then ↑ x is open and so do A.
(2) For the second statement:
Claim: For an open set B, ↑ B = (↓ B c )c ,

24
Proof of the claim: Let x ∈ B c and let y ∈ X such that y ≤ x. By the specialization
order on X, y ∈ {x} ⊆ B c (since B c is closed). Thus, y ∈ B c . Therefore, B c is a
down set and hence B c =↓ B c . Thus, (↓ B c )c = (B c )c = B =↑ B and hence the
claim.
Now, A is closed if and only if Ac is open if and only if Ac =↑ Ac = (↓ A)c (by the
claim) if and only if A =↓ A.

Proposition 2.2.15. [14] An Alexandroff topology (X, τ (≤)) is a T0 -space if and

only if x 6= y in X implies V (x) 6= V (y).

Proof. (⇒) If x 6= y, there exists an open set U that contains one and not the other.
If x ∈ U , then x ∈ V (x) ⊆ U and y ∈
/ V (x). Thus, V (x) 6= V (y). Similarly if y ∈ U .
(⇐) Let x 6= y. So, V (x) 6= V (y). Suppose that x, y ∈ V (x), then x ∈
/ V (y). To
see this, suppose to contrary that x ∈ V (y). Then, x, y ∈ V (x) ∩ V (y). Hence,
V (x) ∩ V (y) is a nhood of y which is a proper subset of V (y) which contradicts the
fact that V (y) is the minimal basic nhood containing y.

Suppose that A is a subset of a To -Alexandroff space (X, τ (≤)). Then two types

of topologies are induced on A. One is the To -Alexandroff space on A with respect

to the induced order ≤, and the other one is the induced topology τ (≤) |A which

makes A as a subspace. It is not difficult to see that the two types coincide (see

Theorem 2.2 of [2]).

Example 2.2.5. Let X = {a, b, c, d} with the partial order a ≤ b, a ≤ c and d ≤ c

as shown in the figure. Then, the To -Alexandroff topology is

τ = {φ, X, {a, b, c}, {b}, {c}, {d, c}, {b, c, d}, {b, c}}

with minimal base B = {{a, b, c}, {b}, {c}, {d, c}}. The set A = {a, b, d} is a down

25
set which is not up set, so it is closed and not open. (Note that Ac = {c} ∈ τ ).

In the above example, from the given partial order we could graph the diagram

of the poset and hence we could determine the base of the required topology and

therefore we can write down all the members of the topology.

Example 2.2.6. Let X = {a, b, c, d}, with the To -Alexandroff topology

τ = {φ, X, {a, b, c}, {b}, {c}, {b, d, c}, {b, c}}

We can find the specialization order as follows: the closed sets are
X, φ, {d}, {a, c, d}, {a, b, d}, {a}, {a, d}. Now, {a} = {a}, {d} = {d}, {b} = {b, a, d}
and {c} = {c, a, d}, so (see the relation 2.1.1), a ≤ b, d ≤ b, a ≤ c and d ≤ c and
hence the figure of the poset X is as shown in the following diagram:

Definition 2.2.16. [5] If (X, τ (≤)) is an Artinian To -Alexandroff space, we define

M to be the set of all maximal elements of X. For a point x ∈ X, we define
b =↑ x ∩ M . The point x is isolated in X if {x} is an open set, and hence maximal
x
in X. So, M is the set of all isolated points in X. If A is a subset of an Artinian
To -Alexandroff space, then we define M (A) to be the set of all maximal elements of
A under the induced order.

26
Example 2.2.7. In the last two examples, Example 2.2.5 and 2.2.6, M = {b, c}.
In Example 2.2.5, M (A) = {b, d}.

b ⊆ A ∀ x ∈ A, and if M (A) * M then A is

Lemma 2.2.17. [5] If A is open then x
not open.

Proof. Suppose that A is open and let x ∈ A. Since A is an up set and x ∈ A, then
↑ x ⊆ A. Therefor, x
b =↑ x ∩ M ⊆↑ x ⊆ A.
For the second statement, suppose that M (A) * M and A is open. Then, ∃ x ∈
M (A) such that x ∈
/ M . Since x ∈ M (A), and A is open then there isn’t any y ∈ X
such that x ≤ y. That is, x ∈ M which is a contradiction.

2.3 Identification of Basic Topological Concepts

To be more closer to the Alexandroff topology we must have a close look at the

topological concepts applied on posets that occupied by Alexandroff topology.

Definition 2.3.1. [5] Let A be a subset of a To -Alexandroff space X. A point x

is a cluster point of A if any V (x) intersects A − {x}. The point x ∈ A is isolated
/ A0 , where A0 is the set of all limit (or
if {x} is open in the subspace A; i.e., if x ∈
cluster) points of A.

Proposition 2.3.2. [5] Let (X, τ (≤)) be a To -Alexandroff space and let A ⊆ X.
(1) For x ∈ X, {x} =↓ x.
(2) Ao = {x ∈ A :↑ x ⊆ A}.
[
(3) A = ↓ x.
x∈A
0
(4) A = A\{x : x is maximal in A}.

27
Proof. (1) ↓ x is a down set and hence a closed set containing x. So, {x} ⊆↓ x.
c
Conversely, let y ∈↓ x, so y ≤ x. If y ∈ {x} , which is open set then for any w ∈ X
c
such that y ≤ w, w ∈ {x} . Therefore, ↑ y ∩ {x} = φ. Since x ∈↑ y, we get that
x∈
/ {x}, which is a contradiction.
(2) Let x ∈ {x ∈ A :↑ x ⊆ A}. So, ↑ x ⊆ A. But ↑ x is open, therefore ↑ x ⊆ Ao .
Thus, x ∈ Ao . Conversely, let y ∈ Ao ⊆ A. Since Ao =↑ Ao , then, Ao ⊆↑ A. So,
y ∈ {x ∈ A :↑ x ⊆ A}.


(3) If x ∈ A, then {x} =↓ x ⊆ A. So ↓ x ⊆ A. On the other hand, if x ∈ A then


x∈A

x ∈↓ x ⊆ ↓ x. So A ⊆ ↓ x, which is a closed set. Therefor A ⊆ ↓ x.
x∈A x∈A x∈A
(4) If x ∈ A then x ∈ A (since A = A ∪ A ) and ↑ x ∩ A \ {x} =
 φ (since ↑ x is an
open set containing x and x is a cluster point of A), so x is not maximal in A.
Thus, x ∈ A\{x : x is maximal in A} and hence A ⊆ A\{x : x is maximal in A}.
For the other inclusion, suppose that y ∈ A, and y is not maximal in A. Then, we
have that ↑ y ∩ A = φ. If ↑ y ∩ A = {y}, then y is an isolated point and hence y is
maximal in A, and this is not true (contradicts the assumption). So, we must have
y ∈ A .

Theorem 2.3.3. [5] Let (X, τ (≤)) be an Artinian To -Alexandroff space and let
A ⊆ X. Then
(1) Ao = φ if and only if A ∩ M = φ.


(2) A = ↓ x =↓ M (A).
x∈M (A)


(3) A = ↓ x \{x} =↓ M (A)\M (A).
x∈M (A)

(4) The subset A is dense if and only if M ⊆ A.

(5) The subset A is nowhere dense if and only if M ∩ A = φ.
(6) If | M | = 1, then any subset is either dense or nowhere dense.

Proof. (1) (⇒) Suppose that Ao = φ and let x ∈ A ∩ M . So, x is maximal of X in A

and hence ↑ x = {x} ⊆ A. Thus, the open set ↑ x = {x} ⊆ Ao , which contradicting

28
the assumption that Ao = φ.
(⇐) Suppose that A ∩ M = φ, and y ∈ Ao . So ↑ y ⊆ A. Since X satisfies the ACC,
we get a maximal element z in X such that y ≤ z and so z ∈↑ y ⊆ A. Therefor,
z ∈ A ∩ M and hence, A ∩ M 6= φ which is a contradiction.
(2) If x ∈ A, then there exists a maximal element y in A such that x ≤ y, so
[ [
↓ x ⊆↓ y, and this implies that A = ↓x⊆ ↓ x =↓ M (A).
x∈A x∈M (A)
The other inclusion is obvious, since {↓ x : x is maximal in A} ⊆ {↓ x : x ∈ A} .
(3) Since X satisfies the ACC, it follows that

A = A\{x : x is maximl in A}
[
. = (↓ x) \M (A)
x∈M (A)
[
= ↓ x \{x}.
x∈M (A)

(4) (⇒) Suppose that A is dense, and let x ∈ M . Then ↑ x ∩ A 6= φ (by Theorem
1.3.19). But ↑ x = {x}, so x ∈ A.
(⇐) Suppose that M ⊆ A, so M (A) = M. By part (2),
[ [
A= ↓x= ↓x=X
x∈M (A) x∈M

o
(5)(⇒) Suppose that A is nowhere dense, i.e., A = φ, so by part(1) of this Theorem,
M ∩ A = φ, and hence M ∩ A = φ.
(⇐) Suppose that M ∩ A = φ, so no maximal element of X is in A. By Proposition
o
2.3.2, no maximal elements of X is in A, and hence A = φ.
(6) Let M = {>}, and let A be a subset of X. Then either > ∈ A or > ∈
/ A, and
by parts (4) and (5) above, either A is dense or nowhere dense.

29
Chapter 3

Scott Topology

In this chapter, we introduce first the definition of the Scott topology followed by

some illustrative examples of Scott open sets and Scott topology. Furthermore, we

shed some light on some properties of the Scott topology such as being a T0 -space

or a sober space. Also, we seek in the relation between the Scott topology and the

Alexandroff topology. The base of this topology is given later.

3.1 Scott open sets

We begin this section by giving the definition of the Scott-open set.

Definition 3.1.1. [8] A subset U of a poset P is Scott open if:

(i) U is an up set (some authors use the term “upper set” instead of up set with the
same definition), and
(ii) U is inaccessible by directed suprema. That is; for any directed subset S ⊆ P
W W
with a supremum S, if S ∈ U , there exists s0 ∈ S such that s0 ∈ U
(i.e., S ∩ U 6= φ).

In this case, s ∈ U for each s ∈ P with s0 ≤ s.

30
Remarks 3.1.2. (1) From (i) in the above definition, we observe that every Scott
open is Alexandroff open.
(2) Some authors define the Scott topology over a dcpo and then S in (ii) of the
above definition is supposed to be just directed (see [4]).
(3) A subset F ⊆ P is Scott closed if its complement is Scott open. That is; if F C
W
is an up set and for any directed subset D of P that has a supremum D with
D ∈ F C , we have D ∩ F C 6= φ and so D * F .
W

This leads us to the following lemma:

Lemma 3.1.3. A subset F of a poset P is Scoot closed if it satisfies:

(i) F =↓ F (F is a down set) and
(ii) if D is a directed set contained in F and sup D exists, then sup D ∈ F .

Proof. We get this result by the contrapositive of part (3) of Remarks 3.1.2, together
with the fact that the complement of an up set is a down set.

Example 3.1.1. Let P = [0, 10], and define ≤ on P to be the usual order.
It is clear that P is a dcpo. Let U = (1, 10] be a subset of P (clearly U is an up set)
W W
and let S be any directed subset of P such that S ∈ U . Therefore, 1 < S ≤ 10.
W
Clearly, S * [0, 1)(otherwise S ∈ / U ). Thus S ∩ U 6= φ. Hence, the set U is Scott
open.

Lemma 3.1.4. [17] If D is a dcpo, then the set Ux = {z ∈ D : z  x} is a Scott

open set.

We can generalis the above lemma as follows:

Lemma 3.1.5. Let D be a poset. Then, the set Ux = {z ∈ D : z  x} = D− ↓ x

is a Scott open set.

Proof. Let z ∈ Ux and let y ∈ D such that z ≤ y. Suppose that y ∈

/ Ux . Then,
y ≤ x and hence, z ≤ x. So, z ∈
/ Ux which is a contradiction. Hence, y ∈ Ux and

31
hence, Ux is an up set.
W W
Now, let S be any directed subset of D such that S exists and S ∈ Ux . Then,
W
S  x. Assume that for any s ∈ S, s ∈
/ Ux . So, s ≤ x for all s ∈ S. Thus, x is an
W
upper bound of S and hence, S ≤ x which is a contradiction. Thus, S ∩ Ux 6= φ.
Hence, Ux is a Scott open.

Proposition 3.1.6. [4] Let P be a dcpo and let KP denote the set of all compact
elements in P . Then, for any a ∈ KP , ↑ a = {x : a ≤ x} is Scott open.

Proof. It is clear that ↑ a is an up set. Let S be any directed subset of P such that
W W
S ∈ ↑ a. So, a ≤ S. Since a is compact, then there is s ∈ S such that a ≤ s
and hence ↑ a ∩ S 6= φ. Therefore, ↑ a is a Scott open.

Example 3.1.2. The set of all real numbers R, is Scott closed.

W
Proof. Clearly R is a down set. If D ⊆ R is a directed subset and D exists, then
W
D ∈ R. Therefore, R is closed.

Recalling that an ideal I is a directed-lower set (see Definition 1.2.20), we have:

Proposition 3.1.7. In finite posets, every ideal is Scott closed.

Proof. Let P be a finite poset and let I be an ideal subset of P . Since P is finite,
then it is a dcpo. Since I is an ideal, then I is a directed-lower set. So, I =↓ I.
Now, let D be any directed subset of I such that sup D exists (It does, since D ⊆ P
and P is a dcpo). Again by the finiteness of P and by Proposition 1.2.4, sup D ∈ D
and hence sup D ∈ I. This proves the desired result.

Proposition 3.1.8. [4] Let P be a poset. Then, the collection of all Scott open sets
forms a topology on P . This topology is called the Scott topology and is denoted by
σ(P ).

Proof. Let σ(P ) be the collection of all Scott open subsets of P . It is clear that
both φ and P are Scott open.

32
Next, let A, B ∈ σ(P ) and let x ∈ A ∩ B and x ≤ y. Therefore, x ∈ A, x ∈ B and
x ≤ y. Since A, B are up sets, y ∈ A and y ∈ B. That is; y ∈ A ∩ B.
W
Now, let S be any directed subset of P with supremum such that S ∈ A ∩ B.
W W
Therefore, S ∈ A and S ∈ B. Since both A and B are Scott open, then
∃s1 , s2 ∈ S such that s1 ∈ A and s2 ∈ B. Since S is directed, then ∃s ∈ S such
that s1 ≤ s and s2 ≤ s and hence, s ∈ A and s ∈ B (by hereditability of A and B).
Hence, s ∈ A ∩ B and hence A ∩ B ∈ σ(P ).
[
Finally, let {Aα : α ∈ ∆} be a family of Scott open subsets of P . Let x ∈ Aα
α∈∆
with x ≤ y. Therefore x ∈ Aα , for some α ∈ ∆. Since Aα is Scott open and x ≤ y,
[
then y ∈ Aα ⊆ Aα .
α∈∆ [
W W
Now, let S be a directed subset of P with supremum S ∈ Aα . Then, S ∈ Aβ
α∈∆ [
for some β ∈ ∆. But Aβ is Scott open, so ∃s ∈ S such that s ∈ Aβ ⊆ Aα . This
α∈∆
completes the proof.

Example 3.1.3. The right ray topology on R is the Scott topology of usual order.

Proof. Consider R with its usual order and let τr denote the right ray topology on
R, that is; τr = {φ, R} ∪ {(x, ∞) : x ∈ R}. Clearly, for any x ∈ R, U = (x, ∞) ∈ τr
W
is an up set. Now, let S be any directed subset of R with supremum S exists such
W W
that S ∈ U . Thus, x < S < ∞ and hence, x is not an upper bound of S. So,
W
there is s ∈ S such that x < s ≤ S. So, s ∈ U and so, U ∩ S 6= φ. Hence, U is
Scott open.

3.2 Scott Topology and Alexandroff Topology

One consequence of the definition of the Scott topology is that, any algebraic dcpo

D (see Definition 1.2.11) is completely determined by its Scott topology.

Lemma 3.2.1. [4] The order relation on an algebraic dcpo D can be completely
recovered by setting x ≤ y if and only if ∀ O ∈ σ(D), x ∈ O we have that y ∈ O.

33
Proof. (⇒) It is obvious since O is an up set.
(⇐) To prove the other implication, one should note that for any compact element
a ∈ KD , ↑ a = {x ∈ D : a ≤ x} is a Scott open (see Proposition 3.1.6). So, by the
hypothesis that ∀ O ∈ σ(D), x ∈ O ⇒ y ∈ O, choosing a ∈ KD such that x ∈↑ a
implies that y ∈↑ a. Hence a ≤ x ⇒ a ≤ y and so, by Proposition 1.2.15, this is
equivalent to x ≤ y.

Corollary 3.2.2. [4] On any algebraic dcpo D, the Scott topology σ(D) is To .

Proof. It is enough to show that, if ∀ O ∈ σ(D), the condition that “ x ∈ O if

and only if y ∈ O” implies that x = y. Now, by the above lemma, the condition “
x ∈ O if and only if y ∈ O” implies that x ≤ y and y ≤ x and hence, x = y.

Lemma 3.2.3. Every Scott-open set is a Alexandroff open. That is; on any poset
P , the Scott topology is coarser than the Alexandroff topology.

Proof. The proof follows immediately from Definition 3.1.1(i) and Theorem 2.2.14.

The following examples are an illustrative examples. We show that the Scott

topology on a poset may be proper subcollection of the T0 -Alexandroff topology.

Example 3.2.1. Consider the set R of real numbers with the usual order. Let
Ba = {[x, ∞) : x ∈ R}. It is clear that Ba is a base for the T0 -Alexandroff topology
on R. Now, let σ(R) denote the right ray topology on R. Then, σ(R) is the Scott
topology on R (see Example 3.1.3). That is; σ(R) = {φ, R} ∪ {(y, ∞) : y ∈ R}.
[ 1
Since (y, ∞) = [y + , ∞) ∈ τa , where τa is the T0 -Alexandroff topology, then
n∈N
n
σ(R) ⊆ τa .
Claim: U = [x, ∞) is not Scott open. To see this, let S = [b, x) ⊆ R, where b < x.
W
Then S is a directed set and S = x ∈ U . Moreover, S ∩ U = φ. Thus, U is not
Scott open.

34
Example 3.2.2. Recall Example 1.2.9. Let U ⊆ D be such that U = {1, 3}. Clearly
U is an up set and hence an Alexandroff-open.
Claim: U is not Scott open. For, if S = [2, 3), then S is a directed subset of D with
W
S = 3 ∈ U and S ∩ U = φ. Hence, U is not Scott open.

Proposition 3.2.4. [21] On a finite poset D, the T0 -Alexandroff topology agrees

with the Scott topology. That is; τ (≤) = σ(D).

Proof. Let D be a finite poset. Therefore, D is a dcpo (see Example 1.2.3). Let
U be any Alexandroff-open set and let S be any directed subset of D such that
W
S ∈ U . Since S is directed and D is finite, by Proposition 1.2.4, S has a top
W
element > = S ∈ S. So, U ∩ S 6= φ and hence, U is Scott open.

Proposition 3.2.5. [4] Let D be an algebraic dcpo. Then, the family

↑ KD = {↑ a : a ∈ KD }

is a base for the Scott topology σ(D) on D.

Proof. For any x ∈ D, there exists a compact element a ∈ KD such that a ≤ x, that
[ [
is; x ∈↑ a (see Lemma 1.2.12). So, D ⊆ ↑ a and hence, D = ↑ a.
a∈KD a∈KD
Next, let x ∈↑ a ∩ ↑ b for a, b ∈ KD . So, a, b ∈↓K x. Since ↓K x is directed, then
there exists c ∈↓K x such that a ≤ c and b ≤ c. This implies that x ∈↑ c ⊆↑ a∩ ↑ b.
Hence, by Theorem 1.3.16, ↑ KD is a base for σ(D).

Example 3.2.3. Let P be as in Example 1.2.5. Then P = KP and P is algebraic

dcpo. So, B = {↑ a : a ∈ KP } = {↑ x : x ∈ P } is a base for the Scott topology
on P . It is clear that the Scott topology on P is the same as the T0 -Alexandroff
topology.

The following property is an interesting and a strong compactness property.

Lemma 3.2.6. [4] Let D be a poset. Then, for any U ⊆ D and for any a ∈ D,
[
↑a⊆ ↑ b if and only if ∃ b ∈ U such that ↑ a ⊆↑ b.
b∈ U

35
 [ [
Proof. (⇒) Suppose that ↑ a ⊆ ↑ b. Then a ∈ ↑ b (since a ∈↑ a). Therefore,
b∈ U b∈ U
∃ b ∈ U such that a ∈↑ b (and so, b ≤ a) which gives that ↑ a ⊆↑ b.
(⇐) Obvious.

Lemma 3.2.7. [4] Let D be an algebraic dcpo.Then, any Scott open O ⊆ D is the
union of the basic open sets such that
[
O= ↑ a.
a∈O∩KD

Proof. Let O be any Scott open.

(⊆) Let x ∈ O. Then (by Lemma 1.2.12) ∃a ∈ KD such that a ≤ x. That is; x ∈↑ a.
[
Thus, x ∈↑ a ∩ O ⊆↑ a and hence x ∈ ↑ a.
a∈O∩KD
[
(⊇) Let x ∈ ↑ a. Therefore, x ∈↑ a for some a ∈ O ∩ KD . That is; a ≤ x for
a∈O∩KD
[
some a ∈ O ∩ KD ⊆ O. Since O is a Scott open, then x ∈ O. Thus, ↑a⊆
a∈O∩KD
O.

What about the subspace of a Scott space?. Here is the answer:

Proposition 3.2.8. A subspace of a Scott topology is a Scott subspace.

Proof. Let (X, σ(X)) be a Scott topological space and let A be any subset of X.
Then, A has the relative topology TA = {A ∩ U : U ∈ σ(X)}. Let B ∈ TA . So,
there exists U ∈ σ(X) such that B = A ∩ U . Let x ∈ B and y ∈ A such that x ≤ y.
Since x ∈ U ∈ σ(X), then y ∈ U and hence y ∈ A ∩ U = B. Thus, B is an up set
with respect to A. Now, let S be any directed subset of A such that sup S exists
and sup S ∈ B. Then, S ∩ A = S 6= φ. Since U is Scott open, then S ∩ U 6= φ.
Therefore, S ∩ B = S ∩ (A ∩ U ) = (S ∩ A) ∩ U = S ∩ U 6= φ. Hence B is a Scott-open
with respect to A.

The following theorem is a generalization of Corollary 3.2.2:

36
Theorem 3.2.9. On any poset, the Scott topology is T0 .

Proof. Let X be a poset and let x, y ∈ X such that x 6= y. Then either x ∈↓

/ y or
y ∈↓ / y. Then, x ∈ (↓ y)c and y ∈
/ x. Suppose that x ∈↓ / (↓ y)c . By Lemma 3.1.5,
(↓ y)c is Scott open. Similarly if y ∈↓
/ x. So, we are done.

Note that the Scott topology is not T1 , since, WLOG, if x < y, then any up set

contains x must contain y.

Now, we conclude the current chapter by giving the definition of the “sober topol-

ogy”, but before going forward we have some thing to introduce.

Definition 3.2.10. [4] Given a topological space (X, τ ) and any base Bτ for τ , we
define the set Pt(Bτ ) of formal points of the topology consisting of elements of non-
empty subcollections α of Bτ such that:
(1) φ ∈
/ α. (That is; U 6= φ, ∀ U ∈ α).
(2) For any U, V ∈ α, there exists W ∈ Bτ such that W ∈ α and W ⊆ U ∩ V .
[
(3) For any U ∈ α such that U ⊆ Vi , ∃ i ∈ ∆ such that Vi ∈ α.
{Vi ∈Bτ : i∈∆}

Note that condition (2) above insures that for any U, V ∈ α, U ∩ V 6= φ.

Moreover, the Vi ’s in condition (3) are elements of Bτ .

Example 3.2.4. Consider R with the standard topology τ . Then, Bτ = {(a, b) : a, b ∈

R, a < b} forms a base for τ . For each x ∈ R, let αx = {(x − ε, x + δ) : ε, δ > 0}.
Then αx is a formal point. To see this:
(1) Clearly αx ⊆ Bτ and φ ∈
/ αx .
(2) Let U, V ∈ αx such that U = (x − ε1 , x + δ1 ) and V = (x − ε2 , x + δ2 ).
Now, U ∩ V = (x − ε, x + δ) where ε = min{ε1 , ε2 } and δ = min{δ1 , δ2 }. Set
W = U ∩ V = (x − ε, x + δ). Then, W ∈ Bτ , W ∈ αx and W ⊆ U ∩ V.
[
(3) Let U ∈ αx such that U = (x − ε, x + δ). Now, U ⊆ Vi implies that
{Vi ∈Bτ : i∈∆}

37
x ∈ Vi = (ai , bi ) for some i ∈ ∆. Set εi = x − ai > 0, δi = bi − x > 0. So,
Vi = (x − εi , x + δi ) ∈ αx .

Now, we will define the canonical map by means of the formal points (since

this map will be our tool to define the sober Scott topology), but after proving the

following lemma:

Lemma 3.2.11. [4] Let (X, τ ) be a topological space and let Bτ be a base for τ .
Then, the collection Bx = {U ∈ Bτ : x ∈ U } of basic nhoods forms a formal point
for each x ∈ X.

Proof. Let us denote the set {U ∈ Bτ : x ∈ U } by Bx . We will show that Bx is a

formal point. Clearly Bx = {U ∈ Bτ : x ∈ U } ⊆ Bτ . Since x ∈ U ∀ U ∈ Bx , then
U 6= φ ∀ U ∈ Bx . Now, Let U, V ∈ Bx . Then, x ∈ U ∩ V . Since Bτ is a base, then
there exists W ∈ Bτ such that x ∈ W ⊆ U ∩ V . So, x ∈ W . Thus, W ∈ Bx . Finally,
[
Let U ∈ Bx such that U ⊆ Vi , where Vi ∈ Bτ ∀ i. Since U ∈ Bx , then x ∈ U .
[ i∈∆
That is; x ∈ U ⊆ Vi . So x ∈ Vi for some i ∈ ∆ which implies that Vi ∈ Bx .
i∈∆

Definition 3.2.12. [4] For any topological space (X, τ ), the canonical map φ : X →
Pt(Bτ ) is defined by putting:

φ(x) = Bx = {U ∈ Bτ : x ∈ U }.

The canonical map has the following property:

Lemma 3.2.13. [4] Let (X, τ ) be a topological space and let φ : X → Pt(Bτ ) be the
canonical map. If τ is T0 , then the map φ is injective.

Proof. ∀ x, y ∈ X such that φ(x) = φ(y) we have, ∀ U ∈ Bτ , x ∈ U ↔ y ∈ U . Since

τ is T0 , then x = y. Hence, φ is 1-1.

Definition 3.2.14. [4] Let (X, τ ) be a topological space and let φ : X → Pt(Bτ )
be the canonical map defined above. If φ is bijective, then τ is called sober. In this

38
case, no new formal point is added in Pt(Bτ ) which is not already an image of a
point in X.

Proposition 3.2.15. [4] For any algebraic dcpo (D, ≤), the Scott topology σ(D)
with base ↑ KD is sober. In this case, the formal points are the non-empty subcol-
lections α of ↑ KD such that:
(i) ∀ ↑ a, ↑ b ∈ α, ∃ c ∈ KD such that ↑ c ∈ α and a ≤ c and b ≤ c.
(ii) For any a, b ∈ D such that ↑ a ∈ α, b ≤ a, we have ↑ b ∈ α.

Proof. [4] The first condition on formal points holds since all the elements of ↑ KD
are non-empty. (To see this, let ↑ a ∈↑ KD , then a ∈ KD . So for any directed subset
W
U of D such that a ≤ U, ∃ u ∈ U such that a ≤ u. So, u ∈↑ a. That is; ↑ a 6= φ).
Moreover, for any a, b ∈ KD , ↑ a ⊆↑ b if and only if b ≤ a (note that ↑ a, ↑ b are
Scott open and hence up sets) and hence the first condition in the proposition, (in
(i)), is just re-writing of the second condition on the formal points’ definition of a
generic topological space. That is, we can restate the first condition here as follows:
for any ↑ a, ↑ b ∈ α, ∃ ↑ c ∈ KD such that ↑ c ∈ α and ↑ c ⊆ (↑ a ∩ ↑ b), so
a ≤ c and b ≤ c. This shows that (i) of the proposition holds.
The third condition in the definition of the formal points is here substituted by a
simpler one, because of the strong compactness property (Lemma 3.2.6) applied on
the base ↑ KD . (To see this, let a, b ∈ D such that ↑ a ∈ α, b ≤ a. Then,↑ a ⊆↑ b
and hence ↑ b is a cover for ↑ a. So, by the compactness property of KD , ↑ b ∈ α).
This shows that (ii) of the proposition holds.
Now, observe that, for any formal point α, the subset Uα = {a ∈ KD | ↑ a ∈ α} is
directed, because: if a, b ∈ Uα , then ↑ a, ↑ b ∈ α. So, there exists c ∈ KD such that
↑ c ∈ α. Hence, c ∈ Uα such that a ≤ c and b ≤ c (from (i) here). This shows that
W
Uα is directed. Since D is a dcpo and Uα is directed, Uα exists.
Since the Scott topology is a To -space on D, then the canonical map φ is injective
(by Lemma 3.2.13). So, to show that the Scott topology σ(D) is sober we have only
to show that φ is surjective.

39
Now, we can prove that the map φ : D → Pt(↑ KD ) is surjective by showing that:
W
for any formal point α, α = φ( Uα ). To see this:
W _
Let a ∈ KD such that ↑ a ∈ φ( Uα ). So, a ≤ b. Since a is a compact element
↑b∈α
and Uα is directed and
_ _
a≤ b= Uα
↑b∈α

, we obtain that ∃ b ∈ KD such that ↑ b ∈ α(≡ b ∈ Uα ) and a ≤ b. That is, ∃ b ∈ KD

such that ↑ b ∈ α and ↑ b ⊆↑ a, which shows that ↑ a ∈ α, since α is a formal point
(by the second condition of its definition).
For the other inclusion, ↑ a ∈ φ( Uα ) = BW Uα implies that Uα ∈↑ a. This implies
W W
W
that a ≤ Uα . Then,
_ _
φ( Uα ) = {U ∈↑ KD : Uα ∈ U }
_
= {↑ b : Uα ∈↑ b, b ∈ KD }
_
= {↑ b : b ≤ Uα , b ∈ KD }

= α.

This completes the proof of the proposition. Hence the Scott topology is sober.

40
Chapter 4

Scott Topology and

Approximation relation

In this chapter we introduce the definition of the approximation relation and some

of its important properties. Then we will look to the Scott topology through this

relation. The definition of the continuity of a poset will be introduced and through

this we shall see some of the applications of the Scott topology.

4.1 The approximation relation

A significant contribution of the theory of continuous partially ordered sets has

been the explicit definition and use of a new order relation, one that sharpens the

traditional notion of order, namely the approximation relation. Here we introduce

the definition of this relation and browse some of its properties.

Definition 4.1.1. [8] Let (D, ≤) be a poset. Then, for any x, y ∈ D we write x  y
W W
if and only if for all directed sets S ⊆ D with a supremum S, y ≤ S implies
that there exists s ∈ S such that x ≤ s. In other words, x  y if and only if every

41
directed set with join above y has a member above x, [10].
For the symbol “  ”, read “approximates” (Some authors prefer the term “way-
below”[3]). If x  y, we say x approximates y (or x is way-below y). When
confusion may arise, the relation  in a poset D will be specifically written D .
We define ⇓ x = {a ∈ D : a  x} and ⇑ x = {a ∈ D : x  a}.

Remark 4.1.2. The set ⇓ x (resp. ⇑ x) is the set of all elements that approximate
x (resp. that x approximates). In some texts (see [7]), ⇓ x (resp. ⇑ x) is called the
way- below (-above) set of x.

Example 4.1.1. Let P = [1, 3]. Then, (P, ≤)- where ≤ is the usual order - is a
W
dcpo. Clearly 1  2 since for any directed subset U with 2 ≤ U , there exists
u ∈ U such that 1 ≤ u. Actually, 1  x for each x ∈ P .

Some basic properties of the approximation relation are given in the following

proposition:

Proposition 4.1.3. [10] In any partially ordered set P ,

(1) x  y implies x ≤ y.
(2) z ≤ x  y ≤ w implies z  w.
(3) If ⊥ is the least element, then ⊥ x, ∀ x ∈ P .

W
Proof. (1) Suppose that x  y. Since {y} is a directed subset with {y} = y, y ≤
W
{y} = y and x  y, then there exists s ∈ {y} such that x ≤ s. So, x ≤ s = y.
(2) Suppose that z ≤ x  y ≤ w. Let S be any directed subset with supremum
W W W W
S such that w ≤ S. Since y ≤ w, then y ≤ S. Now, x  y and y ≤ S,
then ∃ s ∈ S such that x ≤ s. Finally, z ≤ x implies z ≤ s. Therefore, z  w.
W
(3) Let x ∈ P and let S be any directed subset with supremum S such that
W
x ≤ S. Since S is directed, then S 6= φ. So, there exists s ∈ S such that ⊥ ≤ s
(since ⊥ ≤ s for all s ∈ P ) . Hence, ⊥  x for all x ∈ P .

42
 The converse of part (1) of the above proposition need not be true as the following

example shows.

Example 4.1.2. Recall Example 1.2.9. One can easily check that D is a dcpo and
W
1 ≤ 3. Let S = [2, 3), then S is a directed subset of D with S = 3. Clearly
W
3 ≤ 3 = S. But for any x ∈ S, 1  x. Hence, 1 6 3.

More properties for the approximation relation are given in the following propo-

sition:

Proposition 4.1.4. [3] Let (D, ≤) be a poset. Then, the approximation relation on
D has the following properties :
(1) If x  y and y  x, then x = y.
(2) If x  y and y  z then x  z.
(3) x ∈⇓ y if and only if x  y.
(4) x ∈⇑ y if and only if y  x.
(5) For any x ∈ D, ⇓ x ⊆↓ x and ⇑ x ⊆↑ x.
(6) For any x, y ∈ D, if x ≤ y then ⇓ x ⊆⇓ y and ⇑ y ⊆⇑ x.

Proof. (1) The proof follows immediately by applying (1) in Proposition 4.1.3 above
together with the anti-symmetric property of ≤ on D.
W W
(2) Let S be a directed subset of D with supremum S such that z ≤ S. Since
y  z then ∃ s ∈ S such that y ≤ s. Since x  y implies x ≤ y, therefore x ≤ s.
That is; ∃ s ∈ S such that x ≤ s. Hence, x  z.
(3),(4) Just re-stating the definitions.
(5) Let z ∈⇓ x. Therefore, z  x and hence z ≤ x. That is, z ∈↓ x. Hence,
⇓ x ⊆↓ x.
Now, let w ∈⇑ x. Therefore, x  w and hence x ≤ w. That is, w ∈↑ x. Hence,
⇑ x ⊆↑ x.
(6) Suppose that x ≤ y and let z ∈⇓ x. So, z  x. Therefore, z ≤ z  x ≤ y.

43
Thus, by Proposition 4.1.3 part (2), z  y. Hence, ⇓ x ⊆⇓ y.
For the second statement, let w ∈⇑ y. So, y  w. Now, x ≤ y  w ≤ w implies
that x  w. Hence, ⇑ y ⊆⇑ x.

Lemma 4.1.5. In a finite poset P , x ≤ y if and only if x  y. That is; ↓ y

coincides with ⇓ y.

Proof. We will prove only the first direction. So, let P be a finite poset. Therefore,
P is a dcpo. Now, Suppose that x ≤ y and let S be any directed subset of P such
W W W
that y ≤ S. Therefore, x ≤ S. Since S has a top element > = S ∈ S, then,
take s = > ∈ S so that x ≤ s. Hence, x  y.

The following example shows that the approximation relation need not be re-

flexive.

Example 4.1.3. Recall Example 1.2.9 . Let S = [2, 3). Clearly S is directed subset
W W
with S = 3 and 1 ≤ S = 3. Since ∀x ∈ S, 1kx, then there is no x ∈ S such
that 1 ≤ x. Thus, 1 6 1.

Proposition 4.1.6. [6] For any dcpo P , an element k ∈ P is compact if and only
if k  k.

Proof. (⇒) Let P be any dcpo and let k ∈ P be a compact element. Let U be any
W
directed subset such that k ≤ U . Then, by the definition of the compact element,
there is u ∈ U such that k ≤ u. That is, k  k.
(⇐) Straightforward from the definitions of the approximation relation and the
compact element.

Due to the above proposition, the element 1 in Example 4.1.3 is not compact.

In general, we have the following theorem:

Theorem 4.1.7. Let (P, ≤) be a dcpo.Then,

KP = P if and only if ⇓ x =↓ x.

44
Proof. (⇒) Suppose that for each x ∈ P, x is compact. It is clear that ⇓ x ⊆↓ x.
W
Now, let y ∈↓ x and let D be any directed subset of P such that x ≤ D. Since
W
y ∈↓ x, then y ≤ x. Thus, y ≤ D. Since y is compact, then there exists d ∈ D
such that y ≤ d. Hence, y  x.
(⇐) Suppose that ⇓ x =↓ x. Then, for each x, y ∈ P, x  y if and only x ≤ y.
Thus, ∀ x ∈ P, x  x and hence, x is compact.

Corollary 4.1.8. In finite posets, each element is compact.

Proof. Let P be a finite poset. Then, for any x ∈ P, x ≤ x. Now, from Lemma
4.1.5, x  x and hence by Proposition 4.1.6, x is compact.

Corollary 4.1.9. In any dcpo, the approximation relation  over any subset of
compact elements forms a partial order. In particular, over finite posets, the ap-
proximation relation defines a partial order.

Proof. Let P be a dcpo and let A be a subset of compact elements in P . Now, for
any a, b, c ∈ A:
(i) a  a ( by Proposition 4.1.6). That is;  is reflexive.
(ii) If a  b and b  a, then a = b (by part (1) of Proposition 4.1.4). That is;  is
anti-symmetric.
(iii) If a  b and b  c, then a  c (by part (2) of Proposition 4.1.4). That is; 
is transitive. This completes the proof of the first statement. The second statement
is straightforward from the above corollary.

For an application of the approximation relation over ideals (see Definition

1.2.20), we finish this section by the following proposition:

Proposition 4.1.10. [7] Let P be a poset. Then the following two statements are
equivalent:
(1) x  y.
(2) For each ideal I of P with supremum, the relation y ≤ sup I implies x ∈ I.

45
Proof. (1)⇒(2) Suppose that x  y and let I be an ideal subset of P such that
sup I exists and y ≤ sup I. Since I is directed and x  y, then there is i ∈ I such
that x ≤ i. Since I is a lower set, then x ∈ I.
W
(2)⇒(1) Let U be any directed subset with supremum such that y ≤ U . Consider
I =↓ U . So, I is a directed down set; i.e., I is an ideal. To see this, notice first that
I is a down set. Now, let a, b ∈ I. Then there exist a1 , b1 ∈ U such that a ≤ a1
and b ≤ b1 . Since U is directed, then there exists w ∈ U ⊆ I such that a1 ≤ w
W W W
and b1 ≤ w. Therefore, I is directed. Moreover, I exists and I = U . This
W W
is because for any a ∈ I, there exists a1 ∈ U such that a ≤ a1 ≤ U . So, U is
an upper bound of I. Consider v is any other upper bound of I. So, v is an upper
W W W W
bound of U . Therefore, U ≤ v and hence U = I. So, y ≤ I. Since (2)
holds, then x ∈ I. Therefore, there exists u ∈ U such that x ≤ u. Hence, x  y.

4.2 Continuous posets

One of the most important applications of the definition of the approximation rela-

tion is the definition of the continuous poset that plays a big role in domain theory.

Definition 4.2.1. [9] A poset P is said to be continuous if every element in P is

the directed supremum of elements that approximate it. That is;, a poset P is said
to be continuous if for every x ∈ P there is a directed subset Dx ⊆⇓ x such that
W
x = Dx .

Remark 4.2.2. [7] We can introduce an equivalent definition for P to be continuous as

follows: A poset P is said to be continuous if for all x ∈ P , the set ⇓ x = {y : y  x}
is directed and has x as its join.

Proof. It is clear that the second definition implies the first. Conversely, suppose
W
y  x and z  x (i., e., y, z ∈⇓ x). Let x = Dx where Dx is directed and d  x
for each d ∈ Dx . By definition of , y ≤ d1 and z ≤ d2 for some d1 , d2 ∈ Dx .

46
Suppose d3 is larger than both d1 and d2 in Dx . Then y ≤ d3 , z ≤ d3 and d3  x.
Hence the set ⇓ x is directed.
Since y ≤ x for each y ∈⇓ x, then x is an upper bound of ⇓ x. Suppose that
v is any other upper bound of ⇓ x, then v is an upper bound of Dx . Therefore,
W W
Dx = x ≤ v. Thus, x is the least upper bound of ⇓ x. Hence, x = ⇓ x.

Example 4.2.1. The set of real numbers R under the usual order is a continuous
poset.

Proof. Let x ∈ R be arbitrary fixed point and let z ∈ R. Then we have three cases:
W W
Case 1: z < x. Let U be any directed subset with join U such that x ≤ U .
W
Then, z < U and hence z is not an upper bound of U in R under usual order.
Therefore, there is u ∈ U such that z ≤ u and hence z  x.
Case 2: z = x. Then, z 6 x. For if we take U = (−∞, x) which is directed, then
W W
U = x. So, x ≤ U but there is no u ∈ U such that z = x ≤ u.
Case 3: z > x. Then, by the contrapositive of part (1) in Proposition 4.1.3, we have
z 6 x.
Thus, from the three cases we have ⇓ x = (−∞, x) which is a directed subset with
join x. Hence, R is continuous.

Here are two examples of non-continuous posets:

Example 4.2.2. Recall Example 1.2.9.

Claim: ⇓ 1 = φ. To see this, assume that there is x ∈ D such that x  1. Then,
x ≤ 1 (otherwise x 6 1) and hence x = 1. Therefore, 1  1. Now, let S = (2, 3).
W W
Therefore, S is a directed subset with S = 3 and that 1 ≤ S. Since there is no
s ∈ S such that 1 ≤ s, then 1 6 1, which is a contradiction. So, ⇓ 1 = φ. Thus,
for the element 1 ∈ D, ⇓ 1 is not directed and hence D is not continuous.

Example 4.2.3. [10] A constructive example of a non-continuous poset is obtained

by adding a top element ∞ to the natural numbers under their natural order and

47
adding an element a with 0 < a ≤ ∞ where a is incomparable with all other elements.
W
Now, let P = N ∪ {∞} ∪ {a} ∪ {0}. Then, ⇓ a = {0}. Thus, ⇓ a 6= a. Hence, P
is not a continuous poset.

Recalling the definition of the algebraic dcpo (see Definition 1.2.11), we have the

following result:

Proposition 4.2.3. Every algebraic dcpo is continuous.

Proof. Let D be an algebraic dcpo. Then, for each x ∈ D we have that the set
W
↓K x = {a ∈ KD : a ≤ x} is directed and x = ↓K x. Only we have to show
that↓K x ⊆⇓ x. So, let a ∈↓K x. Then, a is compact and a ≤ x. So, by Proposition
4.1.6, we have a ≤ a  a ≤ x and hence by Proposition 4.1.3 part (2), a  x.
Hence ↓K x ⊆⇓ x. By Definition 4.2.1, D is continuous.

Corollary 4.2.4. Every finite linearly ordered set is continuous.

Proof. Let P be a finite linearly ordered set. Then, by Lemma 1.2.14, P is an

algebraic dcpo. By Proposition 4.2.3, P is continuous.

Theorem 4.2.5. [10] Any finite partially ordered set is continuous, with the ap-
proximation relation () coinciding with the partial order.

Proof. Let (P, ≤) be a finite poset. Then, by Lemma 4.1.5, the approximation
relation() coincides with the partial order relation(≤). Thus, for any x ∈ P , the
set
⇓ x =↓ x = {y : y ≤ x}

is directed with x as its join (see Lemma 1.2.3). Hence, P is continuous.

Example 4.2.4. [10] The plain is a continuous poset under its coordinatewise order
(i.e, for x, y ∈ R2 , x ≤ y if and only if x1 ≤ y1 and x2 ≤ y2 ), with approximation
relation given by x  y if and only if x1 < y1 and x2 < y2 .

48
 We close this section by the following lemma:

Lemma 4.2.6. [10] In a continuous poset P , the set ⇓ x is an ideal for each x ∈ P .

Proof. Let P be a poset. Then, for any x ∈ P , the set ⇓ x is directed and has x as
its join. Let y ∈⇓ x and let z be any element of P such that z ≤ y. So, z ≤ y and
y  x. By Proposition 4.1.3 we have z  x and hence z ∈⇓ x. Thus, ⇓ x is a down
set.

4.3 Scott bases

We saw in Proposition 3.2.5 that the family ↑ KD = {↑ a : a ∈ KD } is a base for the

Scott topology σ(D) on an algebraic dcpo D. In this section, we’ll give more than

one base for the Scott topology over a poset in the shadow of the approximation

relation.

We begin first by the definition of the base for a poset.

Definition 4.3.1. [9] Let P be a poset and B ⊆ P . B is called a basis for P

if ∀ x ∈ P , there is a directed set Dx ⊆ B such that ∀ d ∈ Dx , d  x (i.e.,
Dx ⊆ B∩ ⇓ x) and sup Dx = x.
In other words, a subset B ⊆ P is a basis for P if each element in P is the directed
supremum of elements in B that approximate it.

This together with the definition of a continuous poset lead us to the following

lemma:

Lemma 4.3.2. [8] A poset is continuous if and only if it has a basis.

Proof. (⇒) LetP be a continuous poset. Then, for each x ∈ P , ⇓ x is directed and
W
x = ⇓ x. In the above definition put B = P and Dx =⇓ x to get the result.
(⇐) Let P be a poset with B ⊆ P as its basis. So, ∀ x ∈ P, ∃Dx ⊆ B such that Dx

49


is directed with d
x∀ d ∈ Dx and x = Dx . That is, x is a directed supremum
of elements that approximate it. Hence, P is continuous.

Example 4.3.1. The rational points,Q, forms a basis for R under usual order.

Proof. Let x ∈ R be arbitrary. Then, ⇓ x = (−∞, x) (see Example 4.2.1). Let

B = Q ∩ ⇓ x. Thus, B = {q ∈ Q : q < x}. It is clear that x = B. Also, B is a
subset of Q, which is linearly ordered. So B is directed (see Lemma 1.2.2).

Lemma 4.3.3. [10] The set P of all elements of a continuous poset is a basis for
P.

Proof. Let P be a continuous poset. Then, for any x ∈ P there is a directed subset


Dx ⊆⇓ x = P ∩ ⇓ x such that x = Dx . Hence, P is a basis for P .

Theorem 4.3.4. In a continuous finite poset P , no proper subset is a basis for P .

Proof. Let P be a continuous finite poset and suppose B is a basis for P and let


y ∈ P . Thus, ⇓ y ∩ B contains a directed set Dy such that y = Dy . Since, in
finite posets, each directed subset contains its supremum, therefore y ∈ Dy ⊆⇓ y ∩B
which implies that y ∈ B, and hence B = P .

An important property of the approximation relation on continuous posets is

the interpolation property. Before introducing this property, we have the following

lemma which is valid in the remainder of this chapter.

Lemma 4.3.5. [7] Let P be a continuous poset and let x, y ∈ P such that x
y.
Then, there exists z ∈⇓ x such that z
y.

Proof. Let x, y ∈ P such that x
y and suppose to contrary that for each z ∈⇓
x, z ≤ y. This implies that y is an upper bound of ⇓ x. Since P is continuous, then


x = ⇓ x. Therefore, x ≤ y which is a contradiction. Hence, we are done.

50
Proposition 4.3.6. [7] If P is a continuous dcpo, then the approximation relation
 has the interpolation property: x  z ⇒ ∃ y ∈ P such that x  y  z.

Proof. Let D = {d ∈ P : ∃ b ∈ P such that d  b  z}.

Claim 1: D is directed.
Let d1 , d2 ∈ D. Then there is b1 , b2 ∈ P such that d1  b1  z and d2  b2  z.
Since ⇓ z is directed and b1 , b2 ∈⇓ z then, there is b ∈⇓ z, i.,e., b  z, such that
b1 ≤ b and b2 ≤ b. So, d1  b1 ≤ b ≤ z and d2  b2 ≤ b ≤ z and hence from
Proposition 4.1.3, d1  b and d2  b. That is; d1 and d2 ∈⇓ b. Since ⇓ b is directed,
then there is d ∈⇓ b such that d1 ≤ d and d2 ≤ d. Since d  b  z, then d ∈ D.
Hence, D is directed.
Claim 2: D has z as its join.
By the definition of D, ∀ d ∈ D, d ≤ z. So, z is an upper bound of D. Let p ∈ P
W
such that p = D. Then, p ≤ z. Suppose that p < z, i.e., z  p. Since P is
continuous, then z = sup ⇓ z. Therefore, by Lemma 4.3.5, there exists t  z such
that t  p. Similarly, there exists s ∈⇓ t such that s  p. Thus, s  t  z for some
t ∈ P implies that s ∈ D. However, s ∈ D, p = sup D and s  p is a contradiction.
Hence z ≤ p. Thus, z = sup D.
Now, given x  z, since D is directed and z = sup D, then there exists d ∈ D such
that x ≤ d. So, by the definition of D, there exists b ∈ P such that x ≤ d  b  z.
Hence, again by Proposition 4.1.3, x  b  z and hence the interpolation property
holds.

Corollary 4.3.7. [7] Let P be a continuous dcpo. If x  y and y ≤ sup D, for a

directed subset D of P , then x  d for some d ∈ D. (Note that the desired result is
x  d not x ≤ d as in the definition of the approximation relation.)

Proof. Suppose that x  y and y ≤ sup D for a directed set D. By the interpolation
property, there exists w ∈ P such that x  w  y. Since w  y, there exists d ∈ D
such that w ≤ d. Thus, by Proposition 4.1.3 part(2), x  d.

51
 The interpolation property plays an important role in the theory of continuous

posets; it states that if x  y, then we can interpolate an additional element between

them (note that this element might be x or y when x  x or y  y). Corollary

4.3.7 can be viewed as an alternate version of the interpolation property.

One important application of the interpolation property is stated in the next propo-

sition that associates the approximation relation with the Scott topology. Before

introducing this proposition we pave by the following:

Definition 4.3.8. [7] A filter F in a poset P is called an open filter if it is a Scott

open set. An open filter F is called locally bounded if for all y ∈ F , there exists an
open filter Gy ⊆ F and there exists z ∈ F such that y ∈ Gy and Gy ⊆↑ z.

Lemma 4.3.9. Let X be a poset. Then, for any x, y ∈ X, y ∈int(↑ x) (where

int(↑ x) denotes the interior of ↑ x in the Scott topology) implies x  y.

Proof. Let D be any directed subset of X such that sup D exists and y ≤ sup D.
Since y ∈ int(↑ x), y ≤ sup D and since int(↑ x) is an up set, then sup D ∈
int(↑ x). Since int(↑ x) is Scott-open, there exists d ∈ D such that d ∈ int(↑ x)
⊆↑ x. So, d ∈↑ x and x ≤ d. Hence, x  y.

Proposition 4.3.10. [7] Let P be a continuous dcpo and let x, y ∈ P . Then the
following are equivalent:
(1) x  y
(2) There exists an open filter G with y ∈ G ⊆↑ x.
(3) y ∈ int(↑ x) (where int(↑ x) denotes the interior of ↑ x in the Scott topology).

Proof. (1)(⇒)(2). Since x  y and P is continuous, then by the interpolation

property, there exists x1 such that x  x1  y. Also, there exists x2 such that
x  x2  x1 . Thus, continuing in this process we can obtain {xn }n∈N satisfying
x  xn+1  xn  y for all n. Let G = ∪n (↑ xn ).

52
Claim: G is Scott open.
Clearly, G is an up set. Let D be any directed subset of P such that sup D ∈ G.
Then xn ≤ sup D for some n. Since xn+1  xn , then by the definition of ,
xn+1 ≤ d for some d ∈ D, and thus d ∈↑ xn+1 ⊆ G. Hence, G is Scott open.
Now, from Proposition 4.1.3, we have x ≤ xn+1 ≤ xn ≤ y for each n. Thus, G is a
filtered set and y ∈ G ⊆↑ x.
(2)(⇒)(3) Immediate (since G is Scott open).
(3)(⇒) (1) The proof follows immediately by applying Lemma 4.3.9.

Corollary 4.3.11. [7] Let P be a continuous dcpo and let x ∈ P . Then, ⇑ x =int(↑
x), where the interior of ↑ x is taken with respect to the Scott topology. Hence, ⇑ x
is Scott open.

Proof. According to the above proposition, we can say that, in any continuous dcpo
P , y ∈⇑ x if and only if y ∈ int(↑ x), for x, y ∈ P . Therefore ⇑ x = int(↑ x) and
hence is a Scott open.

Proposition 4.3.12. [7] Let P be a continuous dcpo. Then

(i) {⇑ x : x ∈ P } is a base for the Scott topology, and
(ii) The set of all filters is a base for the Scott topology.

Proof. (i) Let U be a Scott open set and let y ∈ U . Since P is continuous, then ⇓ y
is a directed set and y = sup ⇓ y. Also, U ∩ ⇓ y 6= φ. Thus, there exists x ∈ U such
that x  y, i.e.; y ∈⇑ x ⊆ U . By Corollary 4.3.11, we have ⇑ x = int(↑ x) ∈ σ(P ).
Also, y ∈⇑ x ⊆↑ U = U .
(ii) Let U be a Scott open set and let y ∈ U . Since P is continuous, then ⇓ y is
W
directed and y = ⇓ y. So, U ∩ ⇓ y 6= φ. So, there exists x ∈ U such that x  y.
By Proposition 4.3.10, there exists an open filter F such that y ∈ F ⊆↑ x ⊆ U .
Hence, the collection of open filters forms a base for the Scott topology.

53
 Here we have the following Proposition:

Proposition 4.3.13. [7] Let P be a dcpo. The following statements are equivalent:
(1) P is continuous.
(2) Every open filter is locally bounded, the open filters containing a fixed point form
a descending family and the open filters separate the points of P .
(3) For each x ∈ P, {y ∈ P : x ∈ int(↑ y)} is directed and has x as its supremum.

Proof. (1)⇒(2) Let F be an open filter and y ∈ F . Since y ∈ P and P is continuous,

then ⇓ y is directed and y = sup ⇓ y. Since F is Scott open with y ∈ F , then
F ∩ ⇓ y 6= φ. So, there exists x  y with x ∈ F . By Proposition 4.3.10, there exists
an open filter G with y ∈ G ⊆↑ x ⊆↑ F = F . Thus, F is locally bounded. Since
the open filters are a base for the Scott topology (Proposition 4.3.12), they form a
descending family at each point.
Now, suppose y 6= z, for some z ∈ P . Without loss of generality, we may assume
that y  z. Then, by Lemma 4.3.5, there exists x  y such that x  z and hence
z ∈↑
/ x. By Proposition 4.3.10, there exists an open filter G such that y ∈ G ⊆↑ x.
Hence, z ∈
/ G.
(2)⇒(1) Let x ∈ P and w < x, for some w ∈ P . Since open filters separate points,
then there exists an open filter F such that x ∈ F and w ∈
/ F . Since F is locally
bounded, there exists an open filter G ⊆ F and z ∈ F such that x ∈ G ⊆↑ z. Note
that z  w (since F is an up set and z ∈ F and w ∈
/ F ). Now, by Proposition
4.3.10, x ∈ G ⊆ int(↑ z) ⊆↑ z implies z  x, and hence z ≤ x. Let H denote the
set {z ∈ P : there exists an open filter G such that x ∈ G ⊆↑ z}. Certainly, we
W
have sup H ≤ x, if H exists. We have just shown that for w < x, there exists
such a z with z  w. Hence equality (i.e., x = sup H) must hold, provided the sup
exists. Since for each z ∈ H we have z  x, we complete the proof by showing
H is directed(and hence guaranteeing the existence of the sup) and appealing to
Definition 4.2.1. Let F1 , F2 be open filters, x ∈ F1 ∩ F2 and F1 ⊆↑ zl , F2 ⊆↑ z2 .
By hypothesis, there exists an open filter F such that x ∈ F ⊆ F1 ∩ F2 ; local

54
boundedness implies that there exists an open filter G ⊆ F and z3 ∈ F such that
x ∈ G ⊆↑ z3 . Then z1 ≤ z3 and z2 ≤ z3 .
(1)⇒(3) Since P is continuous, then the set {y ∈ P : x ∈ int(↑ y)} = {y ∈ P : y 
x} =⇓ x. Thus, the required result follows immediately by applying Remark 4.2.2.
(3)⇒(1) Suppose x ∈int(↑ y). Let D be any directed subset of P such that x ≤
sup D. Then, sup D ∈ int(↑ y) since int(↑ y) is Scott open. Thus, D ∩ int(↑ y) 6= φ.
So, d ∈ int(↑ y) for some d ∈ D, and hence y ≤ d. This argument shows y  x (see
Lemma 4.3.9). It follows that y ∈⇓ x, and therefore {y ∈ P : x ∈ int(↑ y)} ⊆⇓ x.
Since x is arbitrary and the set {y ∈ P : x ∈ int(↑ y)} is a directed set of elements
that approximate x with x as its join, the desired conclusion follows by applying
Definition 4.2.1.

55
Conclusion

The aim of this research is to focus on the Scott topology and some of its properties

and furthermore, its relation with the T0 -Alexandroff topology. The research pointed

out the relation that the Scott topology is coarser than the T0 -Alexandroff topology

and they are the same on finite posets.

The researcher looking for more studies in the future Studying application studies

on the Scott topology. Moreover, Looking for a definite definitions of the interior,

exterior, boundary and limit points in the Scott topology.

56
Bibliography

[1] D.N. Georgiou, S.D. Iliadis, A generalization of core compact spaces, University

of Patras, Department of Mathematics,265 00 Patras, Greece, (2004).

[2] F.G. Arenas, Alexandroff spaces, Acta Math. Univ. Comenianae, (1999).

[3] G. Bancerek, The way-below Relation, Warsaw University, Bialystok, For-

malised Mathematics, Volume 6, (1997).

[4] G. Sambin, S. Valentini, Topological Characterization of Scott Domains, Dipart-

mento di Matematica Pura ed Applicata, Universita di Padova,via G. Belzoni

No.7,I-35131 Padova, Italy.

[5] H. Mahdy and M. El atrash, On T0 -Alexandroff spaces, the islamic University

Journal, Volume 13, Number 2,June, (2005).

[6] J.D.Lawson, Problems in Domain Theory and Topology, Department of Math-

ematics, Louisiana State University, (1990).

[7] J.D. Lawson, The Duality of Continuous Posets, Houston Journal of Mathe-

matics, Volume 5, No.3, (1979).

57
 [8] K. Martin, A domain of spacetime intervals in general relativity, Department

of Mathematics, Tulane University, New Orleans,(2006).

[9] L. Xu, Continuity of posets via Scott topology and sobrification, Department of

Mathematics, Yangzhou University, Yangzhou 225002, PR China, (2005).

[10] M. Escardó, Function-space compactifications of function spaces, 4th July,

(2000).

[11] M. Escardó, Topologies on spaces of continuous functions, School of Computer

science, University of Birmingham, (2002).

[12] M.W. Mislove, Local DCPOs, Local CPOs and Local Completions, Department

of Mathematics, Tulane University, New Orleans, (1999).

[13] M.W. Mislove, Toplogy, Domain theory and Theoretical Computer Scince, De-

partment of Mathematics, Tulane University, New Orleans, LA 70118, ( 1997).

[14] P. Alexandroff, Diskrete Räume, Mat. sb. (N.S.)2, 501-518, (1937).

[15] P.E. LONG, An Introduction To General Topology, University of Arkansas,

(1971).

[16] P. Taylor, Computably Based Locally Compact Spaces. Paul Taylor, June 29,

(2004).

[17] R.H. Santiago, Interval Representations, Mat. Apl. Comput., 5, No. 2, (2004).

[18] S. Lipschuts, Theory and Problems of General Topology, Schaum’s Outline se-

ries, (1965).

58
[19] S. Willard, General Topology, University of Alberta, Addison Wesley Publish-

ing Company, London, (1970).

[20] V. Khanna, Lattices and Boolean Algebra, Department of Mathematics, Uni-

versity of Delhi, (1994).

[21] V.R. Pratt, Comonoids in chu: a large cartesian closed sibling of topological

spaces, Computer Science Dept, Stanford University, (2003).

[22] http://planetmath.org/encyclopedia.

[23] http://en.wikipedia.org/wiki/Domain theory.

59