LATTICES

We shall consider some properties of partially ordered sets (called posets) and the
properties that qualifies a poset to be called a lattice.
We shall also discuss chains, antichains and the decomposition theorem.
Definition:
A partially ordered set (A, ≼) is reflexive, antisymmetric and transitive.
i.e.,
A poset (or partially ordered set) is a set A with a binary relation that is reflexive,
antisymmetric and transitive.

≼ denotes “binary relation”.

In other words, if the reflexive, antisymmetric and transitive properties hold for a set A
with a binary relation, the set is called a poset.

A poset is represented by a poset diagram also known as a Hasse diagram.

EXAMPLE: Consider a set A = {1, 2, 3, 4, 5, 6} with a binary relation 𝑎 𝑑𝑖𝑣𝑖𝑑𝑒𝑠 𝑏

i.e., aRb. The poset diagram is shown below.

6
4

2 3 5

1
[Note: All the self-loops and transitive arcs are removed]

In this example, the following result hold:

• The least (minimum) element is 1.
• The greatest (maximum) element does not exist because least and
greatest elements are meant to be unique
• The maximal elements are 4, 6 and 5 because they are the topmost in
their corresponding chains.
• The minimal element is 1.
• For the pair 〈2, 3〉, the join (or least upper bound) is 6 and
the meet (or greatest lower bound) is 1.
• For the pair 〈3, 5〉, the meet is 1 but there is no join.
• Also, for 〈4, 3〉, the meet is 1 but there is no join.
 • Therefore, this poset is not a lattice because every pair of elements of a
lattice has a meet (greatest lower bound) and a join (least upper bound).

Note:
The maximal and minimal elements are different from the maximum and minimum
elements. An element is a maximum if it is larger than every single element in the set,
whereas an element is maximal if it is not smaller than any other element in the set.
(The terms “smaller” and “larger” are determined by the partial order ≼).

If the maximum (greatest) element of a poset exists, then it is unique, i.e., a poset can
have at most one maximum element. The maximum of a poset is the greatest element
in the entire poset or subset under consideration. The minimum (least) element is also
defined dually.

The maximal element is the highest element in a chain of a poset. There can be more
than one maximal element in a poset. The minimal element is defined dually.

Therefore, maximum element is definitely a maximal but a maximal is not necessarily a

maximum.

EXAMPLE: Consider a set A = {1, 2, 3, 4, 5, 6, 7, 8, 9} with a partial order or binary

relation 𝑎 𝑑𝑖𝑣𝑖𝑑𝑒𝑠 𝑏
i.e., aRb. The poset diagram is shown below.

6 9
4

2 3 5 7

The minimum is 1
The maximum is 8
The minimal is 1
The maximal is 8
4, 6 and 9 are equal [antichains]
2, 3, 5 and 7 are equal [they form another antichain]
 〈1, 2, 4, 8〉 is the longest chain with a length of 4 because it has 4
nodes/elements
The length of a chain is the number of elements in that chain / subset

EXAMPLE: Consider a set A = {1, 2, 3, 4, 5, 6, 7, 9} with a partial order or binary

relation 𝑎 𝑑𝑖𝑣𝑖𝑑𝑒𝑠 𝑏
i.e., aRb. The poset diagram is shown below.

6 9
4

2 3 5 7

The minimum is 1
The maximum does not exist
The minimal is 1
The maximal are 4, 6 and 9

Definition:
Let [P, ≦] be any poset, and let 𝑎, 𝑏 ∈ P be given. Then an element 𝑑 ∈ P is called the
greatest lower bound (or meet) of 𝑎 and 𝑏 (symbolized as d = 𝑎 ˄ 𝑏) when
𝑑 ≦ 𝑎, 𝑑 ≦ 𝑏
and
𝑥 ≦ 𝑎, 𝑥 ≦ 𝑏
imply 𝑥 ≦ 𝑑

Dually, the element s ∈ P is called the least upper bound (or join) of 𝑎 and 𝑏 (symbolized
as s = 𝑎 ˅ 𝑏) when
𝑎 ≦ 𝑠, 𝑏 ≦ 𝑠
and
𝑎 ≦ 𝑥, 𝑏 ≦ 𝑥
imply 𝑠 ≦ 𝑥

Note:
The symbols ≦ and ≼ are used interchangeably.
Note:
It is not necessary that the meet and join should exist however if each of them exists, it
must be unique.
In the same vein, if you consider a poset and its subset, the meet and join of any pair of
elements may not belong to the subset.

THEOREM:
○ In any poset P, if the meets and joins exist, then they must satisfy the following lemma
(or dual laws):
1. 𝑥˄𝑥=𝑥 and 𝑥˅𝑥=𝑥 Idempotent law

2. 𝑥˄𝑦=𝑦˄𝑥 and 𝑥˅𝑦=𝑦˅𝑥 Commutative law

3. (𝑥 ˄ 𝑦) ˄ 𝑧 = 𝑥 ˄ (𝑦 ˄ 𝑧) Associative law
and (𝑥 ˅ 𝑦) ˅ 𝑧 = 𝑥 ˅ (𝑦 ˅ 𝑧)

4. 𝑥 ˄ (𝑥 ˅ 𝑦) = 𝑥 and 𝑥 ˅ (𝑥 ˄ 𝑦) = 𝑥 Absorption law

Definition:
Let S be an arbitrary subset of a poset [P ≦].
Then 𝑎 = glb S means that
(i) 𝑎 ≦ 𝑥 for all 𝑥 ∈ S
(ii) µ ≦ 𝑥 for all 𝑥 ∈ S
both imply µ ≦ 𝑎

Read as: “any other lower bound is lesser than the glb, which is 𝑎.

Note:
Glb stands for “greatest lower bound” or “meet”, denoted by ˄
Lub stands for “least upper bound” or “join”, denoted by ˅
𝑎 = glb S is read as “𝑎 is the greatest lower bound of the set S”.
𝑎 ˅ 𝑏 is read as “the join of element a and b”.

Definition:
A lattice is a poset in which any two elements 𝑎 and 𝑏 have a greatest lower bound
called the meet 𝑎 ˄ 𝑏 and a least upper bound called the join 𝑎 ˅ 𝑏.

In other words, if you consider any pair of elements of a poset, it will have a meet and
join.
EXAMPLE: From the following Hasse diagrams, state if each is or isn’t a lattice and
justify your answer.

(a)

2 3 5

1
Answer: This is not a lattice because some pairs have no join or meet. For
instance, 〈3, 5〉 has no join. Also, 〈4, 6〉, has join.

(b)

M5

Answer: M5 is a lattice because any pair of elements has a meet and a join.

(c)

N5
Answer: It is a lattice because any pair of elements has a meet and a join.
 (d) e

a b

c d

P6

Answer: P6 is not a lattice because elements 𝑎 and 𝑏 have a unique

join, which is 𝑒 but they have no unique meet because elements c and d
are both meets of the pair 〈𝑎, 𝑏〉. In the same vein, the pair 〈𝑐, 𝑑〉 has a
unique meet but has two joins which are 𝑎 and 𝑏. And, if the meet and join
exist in a lattice, they must be unique.

(e)

23
Answer: It is a lattice because any pair of elements has a unique
meet and a join. This is a power set of a set, which consists of three
elements.

Later, we shall see that 23 is a distributive lattice while N5 and P6 are non-distributive
lattices.
EXAMPLE: Consider the power set of a set A = {a, b, c}. The poset diagram is
represented below.

{a,b,c}

{b,c}
{a,b} {a,c}

{b} {c}
{a}

You can observe that this is a lattice.

Hint: The power set of this set with 3 elements will have 23 elements, so the poset
diagram will be a cube. If the original set had 2 elements the power set would
have 22 elements so the poset diagram would have been a square.
Note: If you carefully observe, you will notice that a power set is a set under
intersection and union.

Distributive Lattices
In the five-element lattices M5 and N5, it is easy to find triples of elements, which do not
satisfy the dual law of distributivity:
𝑥 ˄ (𝑦 ˅ 𝑧) = (𝑥 ˄ 𝑦) ˅ (𝑥 ˄ 𝑧) and Distributive law
𝑥 ˅ (𝑦 ˄ 𝑧) = (𝑥 ˅ 𝑦) ˄ (𝑥 ˅ 𝑧)

Note:
• The distributive laws are satisfied by any sets under intersection and union;
hence the distributive laws hold [or are identities] in the lattice 2 3.

• However, these distributive laws do not hold as identities in the lattices N5 and M5
in example above.

• It is not necessary that the law of distributivity should be satisfied in a lattice but it
is necessary that a lattice must satisfy the four laws of idempotence,
commutativity, associativity and absorption.
 • A lattice which satisfies the dual law of distributivity is called a distributive lattice,
otherwise, it is called a non-distributive lattice.

• Clearly, the distributive laws are inherited from a lattice to a sublattice moreover,
these laws hold in the power set P(U) of all subsets of any set U.

• In a given lattice, the minimum (or least element) is denoted with O while the
maximum (or greatest element) is denoted with I.

• 𝑥 ˄ 𝑦 is read as 𝑥 meets 𝑦. This is the glb of 𝑥 and 𝑦.

• 𝑥 ˅ 𝑦 is read as 𝑥 join 𝑦. This the lub of 𝑥 and 𝑦.

EXAMPLE: Determine which of the following posets or lattices is a distributive lattice.

(a)
I

a b c

Solution: Now, let’s check if the distributive law holds.

Taking three elements 𝑎, 𝑏, 𝑐, if the distributive law holds then:
𝑎 ˄ (𝑏 ˅ 𝑐) = (𝑎 ˄ 𝑏) ˅ (𝑎 ˄ 𝑐)
↕ ↕
𝑎˄I O˅O
𝑎 O
𝑎 ≠ O
r.h.s ≠ l.h.s
 the distributive law does not hold in this lattice. It is not a
distributive lattice.
(b)
I

b a

c
O

Check: Considering three elements 𝑎, 𝑏, 𝑐

𝑎 ˄ (𝑏 ˅ 𝑐) = (𝑎 ˄ 𝑏) ˅ (𝑎 ˄ 𝑐)
𝑎˄I O˅O
𝑎 𝑐
𝑎 ≠ 𝑐
r.h.s ≠ l.h.s
 the distributive law does not hold in this lattice. It is not a distributive lattice.
(c)

g
w f

b m
n

Check:
Considering the triple 𝑛, 𝑏, 𝑚
𝑛 ˄ (𝑏 ˅ 𝑚) = (𝑛 ˄ 𝑏) ˅ (𝑛 ˄ 𝑚)
𝑛˄I O˅O
O O
O = O
l.h.s = r.h.s
Therefore, the distributive law holds

Also, considering another triple w, f, b

𝑤 ˄ (𝑓 ˅ 𝑏) = (𝑤 ˄ 𝑓) ˅ (𝑤 ˄ 𝑏)
𝑤˄I 𝑛˅𝑏
 𝑤 𝑤
𝑤 = 𝑤
l.h.s = r.h.s
 the distributive law holds throughout the lattice. It is a distributive lattice.

(d)
I

a b

c d

Solution:
This not even a lattice.

Note:
If the (original) lattice is distributive, its sub-lattice will also be distributive.

LEMMA:
In any distributive lattice,
𝑎˄𝑥=𝑎˄𝑦 and 𝑎˅𝑥=𝑎˅𝑦
both imply 𝑥 = 𝑦

Explanation:
If there exists a distributive lattice such that the meet of 𝑎 and 𝑥 equals the meet
of 𝑎 and 𝑦; and the join of 𝑎 and 𝑥 equals the join of 𝑎 and 𝑦, it implies that 𝑥 = 𝑦.

Proof:
To prove that “if 𝑎 ˄ 𝑥 = 𝑎 ˄ 𝑦 and 𝑎 ˅ 𝑥 = 𝑎 ˅ 𝑦 then imply 𝑥 = 𝑦”

By commutativity, the lemma can be restated below

𝑥˄𝑎=𝑦˄𝑎
𝑥˅𝑎 =𝑦˅𝑎
Show that 𝑥 = 𝑦.
 Proof:
𝑥 = 𝑥 ˅ (𝑥 ˄ 𝑎) Absorption law
𝑥 = 𝑥 ˅ (𝑦 ˄ 𝑎) Given lemma states that 𝑥 ˄ 𝑎 = 𝑦 ˄ 𝑎
𝑥 = (𝑥 ˅ 𝑦) ˄ (𝑥 ˅ 𝑎) Distributivity
𝑥 = (𝑦 ˅ 𝑥) ˄ (𝑥 ˅ 𝑎) Commutativity
𝑥 = (𝑦 ˅ 𝑥) ˄ (𝑦 ˅ 𝑎) Given lemma states that 𝑥 ˅ 𝑎 = 𝑦 ˅ 𝑎
𝑥 = 𝑦 ˅ (𝑥 ˄ 𝑎) Distributivity
𝑥 = 𝑦 ˅ (𝑦 ˄ 𝑎) Given lemma states that 𝑥 ˄ 𝑎 = 𝑦 ˄ 𝑎
𝑥 =𝑦 Absorption law: 𝑦 = 𝑦 ˅ (𝑦 ˄ 𝑎)
Q.E.D

Complement
Corollary:
In a distributive lattice, L, a given element 𝑎 can have at most one complement.

Any isomorphism of order between Boolean algebras is an isomorphism for ˄, ˅, and ’.

(˄, ˅, and ’ stand for the operations meet, join and complement.)

Explanation:
What is a complement?
If you have an element 𝑎 and another element 𝑎’, 𝑎’ is the complement of 𝑎 if the meet
of both of them is the minimum element and the join of both of them is the maximum
element.
i.e.,
𝑎’ is the complement of 𝑎 if
𝑎 ˄ 𝑎’ = O and
𝑎 ˅ 𝑎’ = I

EXAMPLE: Consider the power set of a set with two elements {a, b}.

{a,b} I

{a} {b}

Ø O

• {a} is the complement of {b} because

 {a} ˄ {b} = O
and {a} ˅ {b} = I

• In the same vein {a, b} is the complement of Ø

EXAMPLE: Consider the power set of a set A = {a, b, c}.

{a,b,c}

{b,c}
{a,b} {a,c}

{b} {c}
{a}

Ø
• The complement of {a} = {b, c} because their meet is O and their join is I

• The complement of {a, c} is {b}

• The complement of {c} is {a, b}

• The complement of O is I or Ø is the complement of {a, b, c}

Note:
Complement only occurs in a distributive lattice.

In Boolean algebra we have the following operations

˄ and Binary operation
˅ or
￢ not Unary operation
whereas in lattice we have
˄ meet
˅ join
’ complement
Chains and Antichains
Definition:
○ Let (A, ≼) be a partially ordered set. A subset of set A is called a chain if every two
elements in that subset are related.
Note that because of antisymmetric and transitive properties, for any chain with a finite
number of elements {a1, a2, …, ak}, there is an element 𝑎𝑖1 that is less than every other
element in the chain; there is an element 𝑎𝑖2 that is less than every other element
except 𝑎𝑖1 ; there is an element 𝑎𝑖3 that is less than every other element except 𝑎𝑖1 and
𝑎𝑖2 , etc.

EXAMPLE: Identify the chain(s) in the poset below, where 𝑎 is the least element and 𝑒
is the greatest element.

d c

b
a

Solution:
𝑎𝑏𝑐𝑒 is a chain
𝑎𝑑𝑒 is another chain
In the chain 𝑎𝑏𝑐𝑒,
element 𝑎 is smaller than every other element;
element 𝑏 is smaller than every other element except 𝑎;
element 𝑐 is smaller than every other element except 𝑎 and 𝑏.
The result for 𝑎𝑑𝑒 follows in the same manner.
Convention:
We shall use the notation
a1 ≤ a2 ≤ a3 … ≤ ak
as an observation for the list of ordered pairs
𝑎𝑖1 ≤ 𝑎𝑖2 , 𝑎𝑖1 ≤ 𝑎𝑖3 , …, 𝑎𝑖1 ≤ 𝑎𝑖𝑘 , 𝑎𝑖2 ≤ 𝑎𝑖3 , 𝑎𝑖2 ≤ 𝑎𝑖4 , …, 𝑎𝑖2 ≤ 𝑎𝑖𝑘 , …

Length of a Chain:
The number of elements in a chain is the length of the chain.
In the example above, the length of chain 𝑎𝑏𝑐𝑒 is 4, while the length of chain 𝑎𝑑𝑒 is 3.
Definition:
A subset of A is called an antichain if no two distinct elements in the sunset are related.
EXAMPLE: Identify the chain(s) and antichain(s) in the poset below.

d c

b
a

Solution:
The chains are {a, b, c, e}, {a, b, c}, {a, d, e}, and {a}.
The antichains are {b, d}, {c, d}, and {a}

Note: {a} can be considered a chain or antichain.

Now, we have a small result about decomposition, which is stated in the next theorem.

Theorem:
○ Let (P, ≼) be a partially ordered set. Suppose the length of the longest chain(s) in P is
n, then the elements in P can be partitioned into n disjoint antichains.
EXAMPLE: Consider the poset diagram below.

g
c

b t
n w

k
u
m

In this poset,
 The maximal elements are 𝑐 and 𝑔
The minimal elements are 𝑎, 𝑚 and 𝑢
The chains are 𝑎𝑘𝑏𝑐, 𝑚𝑡𝑔, 𝑚𝑛𝑏𝑐, etc.
The length of the longest chain is four (4)
Therefore, the elements in this poset can be partitioned into four disjoint
antichains
o [𝑐, 𝑔] in one block/antichain
o [𝑏, 𝑡] in one block
o [𝑘, 𝑛, 𝑤] in one block
o [𝑎, 𝑚, 𝑢] in another block

The partitioning is not unique, i.e., you can have

[𝑐], [𝑏, 𝑔], [𝑘, 𝑛, 𝑡, 𝑤], [𝑎, 𝑚, 𝑢].
So,
Number of blocks in a partition is four. Each block is an antichain.
Based on this, the next result follows: