Boolean Algebras, Boolean Rings and Stone’s

Representation Theorem
Hongtaek Jung
December 27, 2017

Abstract
This is a part of a supplementary note for a Logic and Set Theory
course. The main goal is to constitute equivalences between the category
of Boolean algebras, Boolean rings and Stone spaces.

Contents
1 Boolean Algebras 1
1.1 Definition and First Properties . . . . . . . . . . . . . . . . . . . 1
1.2 Stone’s Representation theorem . . . . . . . . . . . . . . . . . . . 3

2 Boolean Rings and its Spectrum 5

2.1 Boolean Rings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Spec of Boolean Rings . . . . . . . . . . . . . . . . . . . . . . . . 7

1 Boolean Algebras
1.1 Definition and First Properties
Definition 1.1.1. A boolean algebra B is a set together with complement
operation ¬ and two binary operations ∨ and ∧ subject to
• Commutativity
a ∨ b = b ∨ a, a∧b=b∧a

• Associativity

a ∨ (b ∨ c) = (a ∨ b) ∨ c, a ∧ (b ∧ c) = (a ∧ b) ∧ c

• Distributivity

a ∨ (b ∧ c) = (a ∨ b) ∧ (a ∨ c), a ∧ (b ∨ c) = (a ∧ b) ∨ (a ∧ c)

• Absorption
a ∨ (a ∧ b) = a, a ∧ (a ∨ b) = a

1
 • There are two different elements 0 and 1 in B such that

a ∨ 0 = a, a∧1=a

and
a ∨ ¬a = 1, a ∧ ¬a = 0

Definition 1.1.2. Given two Boolean algebras B1 and B2 the function f : B1 →

B2 is called a Boolean algebra homomorphism (or simply homomorphism if no
confusion occurs) if f satisfies the following properties: f (a ∨ b) = f (a) ∨ f (b),
f (a ∧ b) = f (a) ∧ f (b), f (0) = 0 and f (1) = 1.
Proposition 1.1.3. Let f : B1 → B2 be an homomorphism between Boolean
algebras B1 and B2 . For any x ∈ B1 , we have f (¬x) = ¬f (x).
Proof. Observe:
1 = f (1) = f (x ∨ ¬x) = f (x) ∨ f (¬x)
and
0 = f (0) = f (x ∧ ¬x) = f (x) ∧ f (¬x).
Now the proposition is a consequence of the following lemma.
Lemma 1.1.4. If x ∧ y = 0 and x ∨ y = 1 then y = ¬x.
Proof. To this end, we first observe that

¬x = ¬x ∧ 1
= ¬x ∧ (x ∨ y)
= (¬x ∧ x) ∨ (¬x ∧ y)
= 0 ∨ (¬x ∧ y)
= ¬x ∧ y.

Since 0 = x ∧ y we finally have

¬x = ¬x ∨ 0
= (¬x ∧ y) ∨ (x ∧ y)
= ((¬x ∧ y) ∨ x) ∧ ((¬x ∧ y) ∨ y)
= ((¬x ∨ x) ∧ (y ∨ x)) ∧ y
= (y ∨ x) ∧ y
= y.

We used the absorption property several times.

Example 1.1.5 (Two-element Boolean Algebra). This is the one of motivating
examples of Boolean algebras. Let T = {0, 1}. Interpret 0, 1, ∨, ∧, ¬ as “false”,
“true”, “or”, “and”, “not” respectively. Then we get a Boolean algebra which
is called a two-element Boolean algebra.
Example 1.1.6 (Power Set Boolean Algebra). Let X be a set and consider
its power set 2X . Let 0 = ∅ and 1 = X and interpret ∨, ∧, ¬ as ∪, ∩, X \ −
respectively. Then (2X , ∨, ∧, ¬) becomes a Boolean algebra.

2
Example 1.1.7. Let X be a topological space. Let B(X) be the set of clopen
subsets of X. With the same interpretations of ∨, ∧, ¬ as in the power set alge-
bra, we get a Boolean algebra (B(X), ∨, ∧, ¬). For a connected space X, B(X)
is isomorphic to the two-element Boolean algebra. To get a non-trivial B(X),
the space X should have many connected components. B(X) is particularly in-
teresting when X is totally disconnected, Hausdorff and compact. In that case,
we will see that the Boolean algebra B(X) is universal.
Example 1.1.8 (Duality). Let (B, ∨, ∧, ¬) be a given Boolean algebra. We can
construct another Boolean algebra B ∗ with is dual to B as follows. As a set, B ∗
is the same as B. The operations ∨∗ , ∧∗ , ¬∗ of B ∗ are defined by a ∨∗ b = a ∧ b,
a ∧∗ b = a ∨ b and ¬∗ a = ¬a. 0 in B ∗ is 1 in B and 1 in B ∗ is 0 in B. Then
B ∗ = (B, ∨∗ , ∧∗ , ¬∗ ) is a Boolean algebra.
We can define a filter for a Boolean algebra.
Definition 1.1.9. Let B be a Boolean algebra. A filter F of B is a subset of B
such that for all x, y ∈ F , x ∧ y ∈ F and for all x ∈ F and all z ∈ B, x ∨ z ∈ F .
A subset F is a prime filter if F is a filter and for all x, y ∈ B, x ∨ y ∈ F implies
either x ∈ F or y ∈ F . A ultrafilter is a proper filter which is maximal with
respect to the inclusion.
The definition of a filter for a Boolean algebra is consistent with that of a
set if we regard 2X as a Boolean algebra in the sense of example 1.1.6.
There is a dual notion of a filter which is called an ideal.
Definition 1.1.10. An ideal of a Boolean algebra B is a filter of its dual B ∗ .
Prime and maximal ideal is prime, ultrafilter of B ∗ respectively.
Proposition 1.1.11. Let M be an maximal ideal of a Boolean algebra B. Then
there is a homomorphism f : B → T into the two-element Boolean algebra T
such that M = f −1 (0). Conversely, given a homomorphism f : B → T into the
two-element Boolean algebra T , f −1 (0) is a maximal ideal.
Proof. Assume that f : B → T is given. It is not difficult to show that f −1 (0) is
an ideal. To show that f −1 (0) is maximal, suppose that there is another ideal I
containing f −1 (0). Then there is a ∈ I such that a ∈
/ f −1 (0), that is, f (a) = 1.
−1
Since f (¬a) = ¬f (a) = 0 we see that ¬a ∈ f (0). Then since I is an ideal,
a ∧ ¬a = 1 ∈ I. But if 1 ∈ I we see that for all x ∈ B, 1 ∧ x = x ∈ B. Namely,
I = B. Therefore, f −1 (0) is a maximal ideal.
Conversely, given a maximal ideal M we construct a homomorphism f :
B → T into the two-elements Boolean algebra. For this, we construct a quotient
Boolean algebra B/M and prove that B/M is two-element Boolean algebra. We
do not prove this in detail.
Corollary 1.1.12. A filter U of a Boolean algebra B is a ultrafilter if and only
if for each x ∈ B, either x ∈ U or ¬x ∈ U .

1.2 Stone’s Representation theorem

Let B be a Boolean algebra. Let S(B) be the set of all ultrafilters. We give
a topology on S(B) be declaring the open sets are unions of sets of the form
Ua = {ξ ∈ S(B) | a ∈ ξ}.
Let us investigate properties of the topological space S(B).

3
Proposition 1.2.1. Let B be a Boolean algebra and a, b ∈ B.
1. Ua∧b = Ua ∩ Ub and Ua∨b = Ua ∪ Ub .

2. Ua is closed. Consequently Ua are all clopen subsets.

3. For each point ξ ∈ S(B), the singleton {ξ} is closed.
Proof. (1) follows from the absorption property.
(2) is because UTa = S(B) \ U¬a .
(3) since {ξ} = a∈ξ Ua is closed.
Definition 1.2.2. A topological space X is totally disconnected if no subspace
other than singleton is connected.
Definition 1.2.3. A nonempty totally disconnected compact Hausdorff space
is called a Stone space.

A Cantor set is an example of a Stone space.

Theorem 1.2.4. For a Boolean algebra B, S(B) is a Stone space.
Proof. Let C be any subset of S(B) containing at least two elements. Let ξ and
η be distinct elements in C. Then we can choose z ∈ ξ \ η. One can observe that
{Uz ∩ C, U¬z ∩ C} is a separation of C. Therefore, C is not connected proving
that S(B) is totally disconnected.
We prove the compactness of S(B). Suppose that we are given a set of open
sets {Oα }α∈Λ . We have to find a finite subcover, that is, finite subset Λ0 of Λ
such that {Oα }α∈λ0 covers S(B). Since each Oα is a union of base open sets Ua ,
a ∈ B, it is enough to show the following statement
For any subset B 0 of B such thatS a∈B 0 Ua = S(B), one can find a
S
finite subset B 00 of B 0 such that a∈B 00 Ua = S(B).
Since S \ Ua = U¬a , the above statement is equivalent to

For any subset B 0 of B suchTthat a∈B 0 Ua = ∅, one can find a finite

T
subset B 00 of B 0 such that a∈B 00 Ua = ∅.
We will prove the last statement. Given a subset A of B, denote by hAi the
smallest filter of B containing A. Then one can prove that

hAi = {(a1 ∨ b1 ) ∧ (a2 ∨ b2 ) · · · ∧ (an ∨ bn ) | ai ∈ A, bi ∈ B, n ∈ N}.

T
Moreover, we have Ua ∩ Ub = Ua∧b = Uh{a,b}i . Now since a∈B 0 Ua = UhB 0 i = ∅
we see that hB 0 i contains 0. Therefore, 0 can be written as 0 = (a1 ∨b1 )∧(a2 ∨b2 )∧
· · ·∧(an ∨bn ) for ai ∈ B 0 and bi ∈ B. Define a finite 00
Tsubset B = {a1 , a2 , · · · , an }
0 00
of B . Then we have that 0 is in hB i so that b∈B 00 Ub = ∅. Hence S(B) is
compact.
To show that S(B) is Hausdorff, choose two points ξ and η in S(B). Then
there is a ∈ ξ \ η. Recall that Ua and U¬a are disjoint open sets. Since a ∈ / η,
¬a ∈ η. Therefore, Ua and U¬a separate ξ and η. So, S(B) is Hausdorff.
Corollary 1.2.5. All clopen subset of S(B) is of the form Ua for some a ∈ B.

4
Proof. We proved that subsets Ua are all clopenSin proposition 1.2.1. It U is
an open set then U can be written as a union x∈B 0 Ub for some subset B 0
of B. Suppose that U is closed as well. Then since S(B) is compact, U must
be compact.
Sn , · · · , xn } of B 0 such that
Therefore, there is a finite subset {x1S
n
U = i=1 Uxi . By proposition 1.2.1, we see that U = i=1 Uxi = Ux1 ∨···∨xn .
Let f : B1 → B2 be a homomorphism of Boolean algebras B1 and B2 . We
have seen that a ultrafilter ξ of B2 is the kernel of a homomorphism g : B2∗ → T
into the two-element Boolean algebra T . Since g ◦ f ∗ is a homomorphism from
B1∗ to T , its kernel, denoted by S(f )(ξ), is an ultrafilter of B1 . One can prove
that S(f )(ξ) = f −1 (ξ). Hence given a homomorphism f : B1 → B2 we get a
map S(f ) : S(B2 ) → S(B1 ). Since S(B1 ) and S(B2 ) are given a topology, we
expect that S(f ) is continuous.
Theorem 1.2.6. If f : B1 → B2 is a homomorphism of Boolean algebras B1
and B2 then S(f ) : S(B2 ) → S(B1 ) is continuous.
Proof. For a basic open set U1 of S(B1 ), a ∈ B1 , its inverse image S(f )−1 (Ua )
is a basic open set Uf (a) of S(B2 ).
Now we do a reverse engineering. Given a Stone space X, we could associate
a Boolean algebra B(X). If f : X1 → X2 is a continuous function between Stone
spaces, there is a map B(f ) : B(X2 ) → B(X1 ) sending a clopen subset U of X2
to its inverse image f −1 (U ). This map is a homomorphism of Boolean algebras.
As we mentioned before Boolean algebras of the type B(X) is universal in
the following sense.
Theorem 1.2.7 (Stone’s Representation Theorem). Let B be a Boolean algebra.
Then B is isomorphic to a Boolean algebra B(S(B)).
Proof. We have seen that Ux is clopen for each x ∈ B. Hence define a map
F : B → B(S(B)) by F (x) = Ux . This map is bijection by corollary 1.2.5.
Proving that F is homomorphism is routine.
Remark 1.2.8. The meaning of the Stone’s representation theorem is that,
there is essentially only one type of Boolean algebra, that is B(X). We have
seen a similar theorem in linear algebra course: Every finite dimensional vector
space over a field F is isomorphic to F n . However, there is a big difference
between these two theorems. When we construct an isomorphism between a
given vector space and F n , we have to choose some basis. Such a choice of a basis
is very non-canonical so your isomorphism is not natural. On the other hand,
Stone’s representation theorem does not involve any choice in the construction of
the isomorphism. In this sense, the Stone’s isomorphism is a god-given nature
of Boolean algebras and Stone’s representation theorem reveals that intrinsic
property.

2 Boolean Rings and its Spectrum

2.1 Boolean Rings
We consider the the functors S and B defined in the previous section. The
categories we are interested are the category BA of Boolean algebras and the

5
category Stone of Stone spaces. We proved that S : BA → Stone is a contravari-
ant functor. Moreover, there is another contravariant functor B : Stone → BA.
What the Stone’s representation theorem 1.2.7 tells us is that these two functors
S and B are inverse of one another. In other words, the category BA and Stone
are categorically equivalent. Since they are isomorphic one can say that
Stone spaces can be classified up to homeomorphisms by the invari-
ant B

or
B is a complete invariant of the category of Stone spaces.
Moreover, every categorical term in one category can be translated into the
other category. Let us give one example.

Definition 2.1.1. The object 0 of a category C is called initial if, for each
object X of C, there is unique morphism from 0 to X. Its dual notion is the
final object 1 which is defined by the property that for each object X, there is
unique morphism from X to 1.
Not every category has an initial/final object but if exists, it must be unique
up to isomorphism. Note that the initial and final object may coincide.
Example 2.1.2. In the category BA, the initial object of BA is the two-element
Boolean algebra T and the corresponding final (recall that the functor S is
contravariant) object in Stone is a singleton {pt}.
Now, we introduce another category which is naturally arisen from BA.

Definition 2.1.3. A Boolean ring is a ring R with unity such that x2 = x·x = x
for all x ∈ R.
Proposition 2.1.4. Let R be a Boolean ring. For all x, y ∈ R we have
1. x + x = 2x = 0.

2. x · y = y · x.
In other words, R is necessarily commutative and has characteristic 2.
Proof. We have 2x = (2x)2 = (x + x)2 = x2 + 2x + x2 = 4x which proves (1).
Observe that

x + y = (x + y)2 = x2 + x · y + y · x + y 2 = x + y + x · y + y · x.

Hence, x · y = −y · x. But by (1), we have −y · x = y · x. So, we get (2).

Proposition 2.1.5. Let R be a Boolean ring. The following statements are
equivalent.

1. The cardinality of R is 2.
2. R is a field.
3. R is an integral domain.

6
 4. R = F2 = Z/2Z.
Proof. The only nontrivial implication is (3) ⇒ (4). For any x ∈ R, we have
x2 = x. So, x(x − 1) = 0. Since R is assumed to be an integral domain, we see
that either x = 0 or x = 1.
Let BR be the category of Boolean rings. Define a functor R from the cat-
egory of Boolean algebras BA into the category of Boolean rings BR as follows.
For a given Boolean algebra B = (B, ∨, ∧, ¬), R(B) is, as a set, the same as B.
The operations +, · are defined respectively by a + b = (a ∨ b) ∧ ¬(a ∧ b) and
a · b = a ∧ b. One can show that R(B) = (B, +, ·) is a Boolean ring. For instance,
given any x ∈ B, we have
x2 = x · x = x ∧ x = x ∧ (x ∨ 0) = x.
Finally, R sends a morphism f : B1 → B2 to itself (as a set function), i.e.
R(f )(x) = f (x) ∈ R(B2 ), x ∈ R(B1 ).
Example 2.1.6. Consider a power set algebra 2X . The + operation of associ-
ated Boolean ring is nothing but the symmetric difference: A + B = (A ∪ B) \
(A ∩ B).
Now we construct a functor A from BR to BA. As before, A sends a Boolean
ring R = (R, +, ·) to itself as a set. The operations ∨, ∧, and ¬ are defined by
a ∨ b = a + b + a · b, a ∧ b = a · b, and ¬a = 1 + a respectively. We have to
check that A(R) = (R, ∨, ∧, ¬) so defined satisfies all the axioms of a Boolean
algebra. In particular, we have, by proposition 2.1.4, that
a ∨ ¬a = a ∨ (1 + a) = a + (1 + a) + a · (a + 1) = 1 + 3a + a2 = 1 + 4a = 1
and
a ∧ ¬a = a ∧ (1 + a) = a · (1 + a) = a + a2 = a + a = 0.
A proof of the following statement is straightforward.
Theorem 2.1.7. The functors R and A are inverse of each other.
In other words, two categories BA and BR are equivalent.

2.2 Spec of Boolean Rings

We have the notion of ideals, prime ideals and maximal ideals in a Boolean
algebra. They are in one to one correspondence with ideals, prime ideals and
maximal ideals in the associated Boolean ring.
Proposition 2.2.1. A subset I of a Boolean algebra B is an ideal (prime,
maximal) if and only if I is an ideal (prime, maximal respectively) in R(B).
Proof. Proving that I is an ideal (prime ideal) in B if and only if I is an ideal
(prime ideal respectively) in R(B) is straightforward.
Let M be a maximal ideal in B. Then we have a morphism f : B → T into
a two-element Boolean algebra T such that M = f −1 (0). Then, M = ker R(f ).
Observe that R(T ) is a two-element Boolean ring, which is a field F2 . It is
general fact that the kernel of ring homomorphism into a field is a maximal
ideal. Therefore, M is maximal ideal in R(B).
For the converse, it suffices to show that if a Boolean ring R is a field, then
R = F2 . But it is the case by proposition 2.1.5.

7
Proposition 2.2.2. Let I be an ideal of a Boolean ring. Then I is prime if and
only if I is maximal.

Proof. I is prime (maximal) if and only if I is a kernel of a ring homomorphism

f : M → D into an integral domain (field, respectively) D. Due to proposition
2.1.5, D is a field if and only if D is an integral domain. So, I is prime if and
only if I is maximal.
We have seen that BA and Stone are equivalent categories. Since BA and
BR are equivalent, BR and Stone must be equivalent. Can we find an explicit
functor that gives the equivalence between BR and Stone?
Definition 2.2.3. Let R be a commutative ring with unity. Define a topological
space Spec R, the (prime) spectrum of R, as follows:
• As a set, Spec R is the set of prime ideals.

• We give a topology on Spec R by declaring that the closed sets are subsets
of the type V (I) = {p ∈ Spec R | I ⊂ p} where I is an ideal of R. The
topology defined in this way is called the Zariski topology.
Remark 2.2.4. A spectrum of ring is a central object in algebraic geometry.

In fact the functor Spec from BR to Stone can be written as Spec = S ◦ A.

BAa
S / Stone
;

A Spec
BR