4 views

Uploaded by Yesudas Mallipeddi

save

- Matlab
- PartI2008Solns.pdf
- Matrix Sol
- logic.ps
- Equivalence Relations
- Ch2
- SEMANTIC STUDIES OF A SYNCHRONOUS APPROACH TO ACTIVITY RECOGNITION
- Ramanujan's Notebooks, Part III, Verlag New York, 1991
- Semi Lattice
- complex numbers answer key
- 06_4_10
- IIJEC-2014-09-28-10
- Exterior Derivative - Wikipedia, The Free Encyclopedia
- Basic Matrix Operations
- Test 4
- MIII04sample
- From the Point of View of Inter-universal Teichmuller Theory
- Smarandache N-structure
- BMat59
- T8-9 T.pdf
- Partition Congruences and the Andrews–Garvan–Dyson Crank
- Lesson Plan Polynomial
- 151456
- mongoc
- sql tut
- SQL Tutorial
- ch8
- (Www.entrance-exam.net)-Infosys Placement Sample Paper 7
- Brochure
- dwdm

You are on page 1of 8

**312: Algebraic Combinatorics Lionel Levine
**

Lecture 8

Lecture date: March 1, 2011 Notes by: Christopher Policastro

Remark: In the last lecture, someone asked whether all posets could be constructed from

a point using the operations of disjoint union, ordinal sum, cartesian product and expo-

nentiation. This is not possible in general. Consider the poset N := ¦a, b, c, d¦ with Hasse

diagram:

We can see that this poset is not representable as P + Q, P ⊕ Q, P Q, or P

Q

for any

posets P and Q.

The class of posets that can be constructed using disjoint union and ordinal sum are called

series parallel posets. In exercise 3.13 of Stanley, it is shown that a poset is series parallel

iﬀ its Hasse diagram does not contain N as a subdiagram.

Lattices

Recall from last lecture the deﬁnition of a lattice:

Deﬁnition 1 A poset L is a lattice if every pair of elements x, y has

(i) a least upper bound x ∨ y (called join), and

(ii) a greatest lower bound x ∧ y (called meet);

that is

z ≥ x ∨ y ⇐⇒ z ≥ x and z ≥ y

z ≤ x ∧ y ⇐⇒ z ≤ x and z ≤ y.

For a later result, we will need the following deﬁnition.

8-1

Deﬁnition 2 An element z of a lattice L is called join irreducible if z ,= z

1

∨ z

2

for

z

1

, z

2

< z.

Observe that the meet and join operations are associative; we note that for meet

w ≤ (x ∧ y) ∧ z ⇔ w ≤ z, x ∧ y ⇔ w ≤ x, y, z

⇔ w ≤ x, y ∧ z ⇔ w ≤ x ∧ (y ∧ z)

and similarly for join

w ≥ (x ∨ y) ∨ z ⇔ w ≥ z, x ∨ y ⇔ w ≥ x, y, z

⇔ w ≥ x, y ∨ z ⇔ w ≥ x ∨ (y ∨ z).

Moreover observe that if L is ﬁnite, then L has a unique minimal element

ˆ

0 :=

x∈L

x

and unique maximal element

ˆ

1 :=

x∈L

. By deﬁnition

ˆ

0 ∨ x = x,

ˆ

0 ∧ x =

ˆ

0,

and

ˆ

1 ∨ x =

ˆ

1,

ˆ

1 ∧ x = x.

We might want to show that a poset is a lattice, but only know about one operation. The

following lemma tells us that sometimes this is enough.

Lemma 3 If P is a ﬁnite poset such that

(i) Every x, y ∈ P have a greatest lower bound.

(ii) P has a unique maximal element

ˆ

1.

then P is a lattice.

Proof: Consider x, y ∈ P. Let S = ¦z ∈ P : z ≥ x, y¦. Note that S ,= ∅ since

ˆ

1 ∈ S. Take

x ∨ y :=

z∈S

z. We see that

z ≥ x, y ∀z ∈ S =⇒ x ∨ y ≥ x, y,

and

w ≥ x, y =⇒ w ∈ S =⇒ w ≥

z∈S

z = x ∨ y.

This shows that P admits a join operation. 2

Notice that by an analogous argument, the same conclusion would follow from the existence

of least upper bounds and a unique minimal element. However the meet operation can

oftentimes be more natural than the join operation. We will see this in Example 6.

8-2

Examples of Lattices

Example 4 (n) Recall that n (handwritten n) is the set [n] with the usual order relations.

We see that i ∧ j = min(i, j) and i ∨ j = max(i, j).

Example 5 (Boolean algebras) Recall that B

n

= T([n]). It is a lattice with meet S∧T =

S ∩ T, and join S ∨ T = S ∪ T. This example reinforces our notation.

Example 6 (Partitions) Let π

n

= ¦partitions of [n]¦ ordered by reﬁnement. Given par-

titions σ = (σ

1

, . . . , σ

k

) and τ = (τ

1

, . . . , τ

l

) deﬁne σ ∧ τ as

(nonempty intersections σ

i

∩ τ

j

: 1 ≤ i ≤ k, 1 ≤ j ≤ l).

Since [n] is the unique maximal element of π

n

, Lemma 3 implies that π

n

is a lattice. Showing

that π

n

is a lattice without Lemma 3 would be more diﬃcult because the join of two partitions

has a messy formula.

Example 7 (Vector Spaces) Let V be a vector space, and L be the set of linear subspaces

ordered by inclusion. L is lattice with meet S ∧ T = S ∩ T, and join S ∨ T = S + T =

¦v + w : v ∈ S, w ∈ T¦.

Example 8 (Groups) Let G be a group, and L be the set of subgroups ordered by inclusion.

L is a lattice with meet H ∧ K = H ∩ K, and join H ∨ K = ¸H, K¸.

Note that Examples 7 and 8 are two cases of lattice constructions common to any algebraic

object.

Example 9 (Order-preserving maps) Let P be a poset, and L be a lattice. Recall that

L

P

= ¦order preserving maps P → L¦. L

P

is a lattice with join

(f ∧ g) : P ¸ x → f(x) ∧ g(x) ∈ L,

and meet

(f ∨ g) : P ¸ x → f(x) ∨ g(x) ∈ L.

Distributive Lattices

In this lecture, we will focus on lattices with a certain property that makes them simi-

lar to Boolean algebras. The structure of these lattices is suﬃciently nice to allow for a

classiﬁcation theorem (cf. Theorem 15).

8-3

Deﬁnition 10 A lattice L is distributive if the meet and join operations distribute over

each other i.e. for x, y, z ∈ L

(i) (x ∨ y) ∧ z = (x ∧ z) ∨ (y ∧ z)

(ii) (x ∧ y) ∨ z = (x ∨ z) ∧ (y ∨ z).

Notice that a Boolean algebra is distributive since union and intersection distribute over

each other.

Deﬁnition 11 Given a poset P, an order ideal of P is a subset I ⊂ P such that for all

x ∈ I, if y ≤ x, then y ∈ I.

Consider for instance the following order ideal of B

3

expressed in terms of it Hasse diagram:

Figure 1: Order ideal of B

3

We will denote the set of order ideals as J(P). Note that J(P) is a poset under inclusion.

We will call an order ideal I ⊂ P principal if it is of the form I = ¸x¸ := ¦y ∈ P : y ≤

P

x¦

for some x ∈ P.

Example 12 (Boolean algebras) Let +

n

1 = 1+. . . +1 be the disjoint union of n copies

of 1. Since +

n

1 has no order relations, this means that every subset is an order ideal. So

J(+

n

1) = B

n

.

Example 13 (n) Note that every order ideal of n is of the form ¸b¸ for some b ∈ n. Since

a ≤ b iﬀ ¸a¸ ⊂ ¸b¸, this means that J(n) · n+1.

Example 14 (N) Consider N = ¦a, b, c, d¦ deﬁned at the beginning of lecture. We have

J(N) = ¦∅, ¦a¦, ¦b¦, ¦a, b, c¦, ¦b, d¦, ¦a, b, d¦, ¦a, b, c, d¦, ¦a, b¦¦.

This gives the following Hasse diagram:

8-4

Figure 2: J(N)

Note that if I

1

, I

2

⊂ P are order ideals, then I

1

∩ I

2

and I

1

∪ I

2

are order ideals. Since

unions and intersections distribute over each other, this means that J(P) is a distributive

lattice for any poset P. The following result tells us that if a distributive lattice is ﬁnite,

then it must be of this form.

Theorem 15 (Birkoﬀ’s Theorem)

1

Let L be a ﬁnite distributive lattice. There exists

a poset P unique up to isomorphism such that L · J(P).

Proof:[Uniqueness] Given J(P) we want to recover P. The subset of principal order ideals

is a copy of P sitting inside J(P). However, we want to describe this subset intrinsically

without reference to P.

Fact 1. Let S ⊂ J(P) be the subset of join irreducible elements. S · P:

Let T ⊂ J(P) be the subset of principal order ideals. Note that T · P, because ¸x¸ ⊂ ¸y¸

iﬀ x ≤

P

y. So it is enough to show that S = T.

(T ⊂ S): Suppose that ¸x¸ = I

1

∪ I

2

for I

1

, I

2

∈ J(P). This means that either x ∈ I

1

or

x ∈ I

2

. So ¸x¸ ⊂ I

1

or ¸x¸ ⊂ I

2

, implying ¸x¸ = I

1

or ¸x¸ = I

2

.

(S ⊂ T): Assume that I ∈ J(P) is not principal. So we can ﬁnd distinct maximal elements

x and y in I. Note that I −¦x¦ and I −¦y¦ are order ideals. Since

I = (I −¦x¦) ∪ (I −¦y¦)

this implies that I is not join irreducible. 2

We can now show uniqueness. Suppose that J(P) · L · J(Q) for posets P and Q. This

implies that the join irreducible elements of J(P) and J(Q) are isomorphic. So by Fact 1,

we conclude that P · Q. 2

Proof:[Existence] From the proof of uniqueness, we have to show that L = J(P) for P the

join irreducible elements of L.

1

This result is sometimes called the fundamental theorem of ﬁnite distributive lattices.

8-5

Fact 2. For y ∈ L there exist y

1

, . . . , y

n

∈ P such that y = y

1

∨ . . . ∨ y

n

. For n minimal,

this expression is unique up to permutation:

An element y ∈ L is either join irreducible, or expressible as y = y

1

∨y

2

for y

1

, y

2

< y. Since

L is ﬁnite, we see by induction that an element y ∈ L can be written as y = y

1

∨ . . . ∨ y

n

for y

i

∈ P.

Choose n minimal with this property. Suppose that y

1

∨ . . . ∨ y

n

= y = z

1

∨ . . . ∨ z

n

for

z

j

, y

i

∈ P. Note that y ∧ z

i

= z

i

because z

i

≤ z

1

∨ . . . ∨ z

n

= y. By distributivity

z

i

= z

i

∧ y =

n

j=1

z

i

∧ y

j

.

Since z

i

∈ P, there exists some y

j

such that z

i

= z

i

∧ y

j

. So z

i

≤ y

j

.

Suppose that z

i

, z

j

≤ y

k

for i ,= j. Since join is commutative and associative, we can assume

w.l.o.g that i = 1 and j = 2. Replace z

1

∨z

2

by y

k

in z

1

∨. . . ∨z

n

. So y = y

k

∨(z

3

∨. . . ∨z

n

)

contradicting minimality of n. Therefore σ ∈ S

n

.

Switching the roles of y

i

and z

j

, we obtain τ ∈ S

n

such that y

τ(i)

≤ z

σ(i)

≤ y

i

. By minimality

of n, we see that y

τ(i)

= y

i

. This gives the result. 2

Deﬁne a map f : J(P) ¸ I →

x∈I

x ∈ L. By Fact 2, f is surjective.

Deﬁne a map g : L ¸ y →

y

i

¸y

i

¸ ∈ J(P) where y = y

1

∨. . . ∨y

n

as in Fact 2. Since union

commutes, this map is well-deﬁned by Fact 2. An order ideal I ⊂ P can be expressed as

I =

i

¸y

i

¸ for ¦y

1

, . . . y

n

¦ ⊂ P the maximal elements of I. Note that

g(y

1

∨ . . . ∨ y

n

) =

i

¸y

i

¸ = I

because n is minimal by maximality of the y

i

. Therefore gf is the identity on J(P). This

implies that f is injective.

(f order-preserving): Suppose that I ⊂ I

for I, I

∈ J(P). We have

f(I

) =

y∈I

y = (

y∈I

y) ∨ (

y∈I

−I

y)

= f(I) ∨ z

for some z ∈ L. This implies that f(I

) ≥ f(I).

8-6

(g order-preserving): Suppose that for I, I

∈ J(P), we have

x∈I

x ≤

y∈I

y. We want to

show that I ⊂ I

. Note

x∈I

x ≤

y∈I

y =⇒ x ≤

y∈I

y ∀x ∈ I

=⇒ x = x ∧ x = (

y∈I

y) ∧ x

=⇒ x =

y∈I

(y ∧ x)

by distributivity. Since x is join irreducible, this implies that x = y ∧x for some y ∈ I

. So

x ≤ y for some y ∈ I

. This means that x ∈ I

because I

is an order ideal. Hence I ⊂ I

.

Therefore J(P)

∼

−→ L. 2

Note that we did not make full use of distributivity; we only needed to know that (x∨y)∧z =

(x ∧ z) ∨ (y ∧ z) for x, y, z ∈ L. Hence property (i) of Deﬁnition 10 implies property (ii).

This is a convenient fact for showing that a lattice is distributive, because only property (i)

of Deﬁnition 10 needs to be checked.

We will end this lecture with a proposition that gives us another way of thinking about

order ideals that will be useful when we try to compare posets.

Deﬁnition 16 Given a poset P, deﬁne P

∗

to be the set P with reversed order relations i.e.

x ≤

P

y iﬀ x ≥

P

∗ y.

Proposition 17 J(P) · (2

P

)

∗

.

Proof: Given a map f ∈ 2

P

, let ϕ(f) = f

−1

(1) ⊂ P. Since f is order-preserving, this

implies that ϕ(f) ∈ J(P). So we obtain a map ϕ : 2

P

→ J(P).

Suppose that g ≤

2

P h. If h(x) = 1 for some x ∈ P, then

g ≤

2

P h =⇒ g(x) ≤

P

h(x) =⇒ g(x) = 1.

So if x ∈ ϕ(h) then x ∈ ϕ(g), or equivalently ϕ(h) ≤

J(P)

ϕ(g). Therefore ϕ is order-

reversing.

Given an order ideal I ⊂ P, deﬁne a map ψ(I) : P → 2 by

ψ(I)(x) :=

2 if x / ∈ I

1 if x ∈ I

.

Suppose that x ≤

P

y. There exist three cases:

x, y ∈ I =⇒ ψ(I)(x) = 1 = ψ(I)(y),

8-7

x, y / ∈ I =⇒ ψ(I)(x) = 2 = ψ(I)(y),

x ∈ I, y / ∈ I =⇒ ψ(I)(x) = 1 <

2

2 = ψ(I)(y).

So ψ(I)(x) is order-preserving, and we obtain a map ψ : J(P) → 2

P

.

Suppose that I ≤

J(P)

I

**. For x ∈ P there exist three cases:
**

x ∈ I, x ∈ I

=⇒ ψ(I)(x) = 1 = ψ(I

)(x),

x / ∈ I, x ∈ I

=⇒ ψ(I)(x) = 2 ≥

2

1 = ψ(I

)(x),

x / ∈ I, x / ∈ I

=⇒ ψ(I)(x) = 2 = ψ(I

)(x).

So ψ(I)(x) ≥

2

ψ(I

**)(x). Since x was arbitrary, this implies that ψ(I) ≥
**

2

P ψ(I

). Therefore

ψ is order-reversing.

We claim that ϕ and ψ are inverses. Consider f ∈ 2

P

. By deﬁnition for x ∈ P,

[ψϕ(f)](x) = [ψ(f

−1

(1))](p)

=

2 if x / ∈ f

−1

(1)

1 if x ∈ f

−1

(1)

=

2 if f(x) = 2

1 if f(x) = 1

= f(x)

Therefore ψϕ(f) = f. Conversely, consider I ∈ J(P). By deﬁnition

ϕψ(I) = ψ(I)

−1

(1) = I.

Therefore ϕψ(I) = I. 2

8-8

y ∈ P have a greatest lower bound. Take 1 x ∨ y := z∈S z. y}. We will see this in Example 6. The following lemma tells us that sometimes this is enough. then P is a lattice. and ˆ ∨ x = ˆ ˆ ∧ x = x. We see that z ≥ x. and w ≥ x. Proof: Consider x. x ∧ y ⇔ w ≤ x. we note that for meet w ≤ (x ∧ y) ∧ z ⇔ w ≤ z. y. y ∀z ∈ S =⇒ x ∨ y ≥ x. However the meet operation can oftentimes be more natural than the join operation. Lemma 3 If P is a ﬁnite poset such that (i) Every x. By deﬁnition ˆ ∨ x = x. y ∨ z ⇔ w ≥ x ∨ (y ∨ z). (ii) P has a unique maximal element ˆ 1. ˆ ∧ x = ˆ 0 0 0. 2 Notice that by an analogous argument. y =⇒ w ∈ S =⇒ w ≥ z∈S x∈L x z = x ∨ y. x ∨ y ⇔ w ≥ x. y ∈ P . y ∧ z ⇔ w ≤ x ∧ (y ∧ z) and similarly for join w ≥ (x ∨ y) ∨ z ⇔ w ≥ z. This shows that P admits a join operation. Let S = {z ∈ P : z ≥ x. but only know about one operation. Note that S = ∅ since ˆ ∈ S. z ⇔ w ≤ x. then L has a unique minimal element ˆ := 0 ˆ := and unique maximal element 1 x∈L . 1 We might want to show that a poset is a lattice. z2 < z. 1 1. z ⇔ w ≥ x. the same conclusion would follow from the existence of least upper bounds and a unique minimal element. y. Observe that the meet and join operations are associative. Moreover observe that if L is ﬁnite.Deﬁnition 2 An element z of a lattice L is called join irreducible if z = z1 ∨ z2 for z1 . 8-2 . y.

. It is a lattice with meet S∧T = S ∩ T . Example 8 (Groups) Let G be a group. j). Distributive Lattices In this lecture. . and L be a lattice. 1 ≤ j ≤ l). . L is lattice with meet S ∧ T = S ∩ T . w ∈ T }. Lemma 3 implies that πn is a lattice. . Recall that LP = {order preserving maps P → L}. 8-3 . . Theorem 15). Note that Examples 7 and 8 are two cases of lattice constructions common to any algebraic object. K . and L be the set of subgroups ordered by inclusion. Given partitions σ = (σ1 . j) and i ∨ j = max(i. σk ) and τ = (τ1 . L is a lattice with meet H ∧ K = H ∩ K. We see that i ∧ j = min(i. . LP is a lattice with join (f ∧ g) : P and meet (f ∨ g) : P x → f (x) ∨ g(x) ∈ L. τl ) deﬁne σ ∧ τ as (nonempty intersections σi ∩ τj : 1 ≤ i ≤ k. The structure of these lattices is suﬃciently nice to allow for a classiﬁcation theorem (cf. and join H ∨ K = H. Example 5 (Boolean algebras) Recall that Bn = P([n]). Showing that πn is a lattice without Lemma 3 would be more diﬃcult because the join of two partitions has a messy formula. Since [n] is the unique maximal element of πn . . and L be the set of linear subspaces ordered by inclusion. Example 7 (Vector Spaces) Let V be a vector space. . and join S ∨ T = S ∪ T . This example reinforces our notation. Example 6 (Partitions) Let πn = {partitions of [n]} ordered by reﬁnement. Example 9 (Order-preserving maps) Let P be a poset. and join S ∨ T = S + T = {v + w : v ∈ S. x → f (x) ∧ g(x) ∈ L.Examples of Lattices Example 4 (n) Recall that n (handwritten n) is the set [n] with the usual order relations. we will focus on lattices with a certain property that makes them similar to Boolean algebras.

c. d}.Deﬁnition 10 A lattice L is distributive if the meet and join operations distribute over each other i. Example 13 (n) Note that every order ideal of n is of the form b for some b ∈ n. c}. . b. this means that every subset is an order ideal. Note that J(P ) is a poset under inclusion. d}. We will call an order ideal I ⊂ P principal if it is of the form I = x := {y ∈ P : y ≤P x} for some x ∈ P . {a. z ∈ L (i) (x ∨ y) ∧ z = (x ∧ z) ∨ (y ∧ z) (ii) (x ∧ y) ∨ z = (x ∨ z) ∧ (y ∨ z). {a}. an order ideal of P is a subset I ⊂ P such that for all x ∈ I. b. {a. + 1 be the disjoint union of n copies of 1. b. Since a ≤ b iﬀ a ⊂ b . Example 12 (Boolean algebras) Let +n 1 = 1 + . So J(+n 1) = Bn .e. d} deﬁned at the beginning of lecture. c. Deﬁnition 11 Given a poset P . Consider for instance the following order ideal of B3 expressed in terms of it Hasse diagram: Figure 1: Order ideal of B3 We will denote the set of order ideals as J(P ). b. y. {a. Notice that a Boolean algebra is distributive since union and intersection distribute over each other. {a. This gives the following Hasse diagram: 8-4 . if y ≤ x. {b}. Since +n 1 has no order relations. b}}. d}. We have J(N) = {∅. this means that J(n) n+1. {b. Example 14 (N) Consider N = {a. then y ∈ I. for x. .

This implies that the join irreducible elements of J(P ) and J(Q) are isomorphic. So x ⊂ I1 or x ⊂ I2 . (S ⊂ T ): Assume that I ∈ J(P ) is not principal. Let S ⊂ J(P ) be the subset of join irreducible elements. Note that T iﬀ x ≤P y. this means that J(P ) is a distributive lattice for any poset P . implying x = I1 or x = I2 . The subset of principal order ideals is a copy of P sitting inside J(P ). Theorem 15 (Birkoﬀ ’s Theorem) 1 Let L be a ﬁnite distributive lattice. then I1 ∩ I2 and I1 ∪ I2 are order ideals. 8-5 . we want to describe this subset intrinsically without reference to P . P: P . There exists a poset P unique up to isomorphism such that L J(P ). we have to show that L = J(P ) for P the join irreducible elements of L. So we can ﬁnd distinct maximal elements x and y in I. Note that I − {x} and I − {y} are order ideals. The following result tells us that if a distributive lattice is ﬁnite. So by Fact 1. Since unions and intersections distribute over each other. 1 This result is sometimes called the fundamental theorem of ﬁnite distributive lattices. 2 Proof:[Existence] From the proof of uniqueness. Since I = (I − {x}) ∪ (I − {y}) this implies that I is not join irreducible. we conclude that P Q. 2 We can now show uniqueness. S Let T ⊂ J(P ) be the subset of principal order ideals. However. I2 ∈ J(P ). I2 ⊂ P are order ideals. This means that either x ∈ I1 or x ∈ I2 . Proof:[Uniqueness] Given J(P ) we want to recover P . then it must be of this form.Figure 2: J(N) Note that if I1 . So it is enough to show that S = T . Fact 1. Suppose that J(P ) L J(Q) for posets P and Q. because x ⊂ y (T ⊂ S): Suppose that x = I1 ∪ I2 for I1 .

This gives the result. . there exists some yj such that zi = zi ∧ yj . . . or expressible as y = y1 ∨ y2 for y1 . . . yi ∈ P . . . . This implies that f (I ) ≥ f (I). This implies that f is injective. Since join is commutative and associative. By Fact 2. For n minimal. Switching the roles of yi and zj . We have f (I ) = y∈I y=( y∈I y) ∨ ( y∈I −I y) = f (I) ∨ z for some z ∈ L. this map is well-deﬁned by Fact 2. Since L is ﬁnite. ∨ zn = y. . yn ∈ P such that y = y1 ∨ . we see by induction that an element y ∈ L can be written as y = y1 ∨ . ∨ zn ) contradicting minimality of n. zj ≤ yk for i = j. ∨ yn = y = z1 ∨ . Since union commutes. Therefore σ ∈ Sn . Replace z1 ∨ z2 by yk in z1 ∨ . . . yn } ⊂ P the maximal elements of I. we obtain τ ∈ Sn such that yτ (i) ≤ zσ(i) ≤ yi . Deﬁne a map g : L y → yi yi ∈ J(P ) where y = y1 ∨ .g that i = 1 and j = 2. . . . . An order ideal I ⊂ P can be expressed as I = i yi for {y1 . f is surjective. . we can assume w. By minimality of n. ∨ yn as in Fact 2. 2 Deﬁne a map f : J(P ) I→ x∈I x ∈ L. 8-6 . ∨ yn for yi ∈ P . y2 < y. (f order-preserving): Suppose that I ⊂ I for I. So zi ≤ yj . Note that g(y1 ∨ . Suppose that zi . So y = yk ∨ (z3 ∨ . ∨ zn . . . ∨ zn for zj . . Choose n minimal with this property. . Since zi ∈ P . I ∈ J(P ). . we see that yτ (i) = yi . ∨ yn ) = i yi = I because n is minimal by maximality of the yi . Note that y ∧ zi = zi because zi ≤ z1 ∨ . . Suppose that y1 ∨ . . ∨ yn .l. For y ∈ L there exist y1 .o.Fact 2. . By distributivity n zi = zi ∧ y = j=1 zi ∧ yj . this expression is unique up to permutation: An element y ∈ L is either join irreducible. . Therefore gf is the identity on J(P ).

This means that x ∈ I because I is an order ideal. then g ≤2P h =⇒ g(x) ≤P h(x) =⇒ g(x) = 1. 2 Note that we did not make full use of distributivity. this implies that x = y ∧ x for some y ∈ I . There exist three cases: x. or equivalently ϕ(h) ≤J(P ) ϕ(g). Therefore J(P ) −→ L.(g order-preserving): Suppose that for I. Since f is order-preserving. 8-7 . z ∈ L. Suppose that g ≤2P h. this implies that ϕ(f ) ∈ J(P ). deﬁne a map ψ(I) : P → 2 by ψ(I)(x) := 2 if x ∈ I / . Note x≤ x∈I y∈I x∈I x≤ y∈I y. y ∈ I =⇒ ψ(I)(x) = 1 = ψ(I)(y). We will end this lecture with a proposition that gives us another way of thinking about order ideals that will be useful when we try to compare posets. Hence I ⊂ I . Proposition 17 J(P ) (2P )∗ . We want to y =⇒ x ≤ y∈I y ∀x ∈ I y) ∧ x =⇒ x = x ∧ x = ( y∈I =⇒ x = y∈I (y ∧ x) by distributivity. deﬁne P ∗ to be the set P with reversed order relations i. we have show that I ⊂ I . So x ≤ y for some y ∈ I . let ϕ(f ) = f −1 (1) ⊂ P . So we obtain a map ϕ : 2P → J(P ). Given an order ideal I ⊂ P . we only needed to know that (x∨y)∧z = (x ∧ z) ∨ (y ∧ z) for x. Therefore ϕ is orderreversing. I ∈ J(P ). So if x ∈ ϕ(h) then x ∈ ϕ(g). 1 if x ∈ I Suppose that x ≤P y. This is a convenient fact for showing that a lattice is distributive. x ≤P y iﬀ x ≥P ∗ y. Hence property (i) of Deﬁnition 10 implies property (ii). y.e. Since x is join irreducible. ∼ Proof: Given a map f ∈ 2P . Deﬁnition 16 Given a poset P . If h(x) = 1 for some x ∈ P . because only property (i) of Deﬁnition 10 needs to be checked.

x ∈ I =⇒ ψ(I)(x) = 1 = ψ(I )(x). y ∈ I =⇒ ψ(I)(x) = 2 = ψ(I)(y). Consider f ∈ 2P . / x ∈ I. Since x was arbitrary.x. and we obtain a map ψ : J(P ) → 2P . x ∈ I =⇒ ψ(I)(x) = 2 ≥2 1 = ψ(I )(x). this implies that ψ(I) ≥2P ψ(I ). Suppose that I ≤J(P ) I . Therefore ψ is order-reversing. / / So ψ(I)(x) ≥2 ψ(I )(x). Conversely. By deﬁnition for x ∈ P . By deﬁnition ϕψ(I) = ψ(I)−1 (1) = I. x ∈ I. [ψϕ(f )](x) = [ψ(f −1 (1))](p) = = 2 if x ∈ f −1 (1) / 1 if x ∈ f −1 (1) 2 if f (x) = 2 = f (x) 1 if f (x) = 1 Therefore ψϕ(f ) = f . 2 8-8 . consider I ∈ J(P ). y ∈ I =⇒ ψ(I)(x) = 1 <2 2 = ψ(I)(y). x ∈ I =⇒ ψ(I)(x) = 2 = ψ(I )(x). / x ∈ I. / So ψ(I)(x) is order-preserving. For x ∈ P there exist three cases: x ∈ I. Therefore ϕψ(I) = I. We claim that ϕ and ψ are inverses.

- MatlabUploaded byJeremy Byers
- PartI2008Solns.pdfUploaded bySahabudeen Salaeh
- Matrix SolUploaded byapi-3742735
- logic.psUploaded byDaniel
- Equivalence RelationsUploaded bygdeepthi
- Ch2Uploaded byAlexander Checya
- SEMANTIC STUDIES OF A SYNCHRONOUS APPROACH TO ACTIVITY RECOGNITIONUploaded byCS & IT
- Ramanujan's Notebooks, Part III, Verlag New York, 1991Uploaded byapi-26401608
- Semi LatticeUploaded bysthirumangalath
- complex numbers answer keyUploaded byapi-232996589
- 06_4_10Uploaded byQfwfq Redipicche
- IIJEC-2014-09-28-10Uploaded byAnonymous vQrJlEN
- Exterior Derivative - Wikipedia, The Free EncyclopediaUploaded byAlbert Quaeda
- Basic Matrix OperationsUploaded byRobert John Lemu
- Test 4Uploaded byMasha Iwqedwe
- MIII04sampleUploaded bysrikanth
- From the Point of View of Inter-universal Teichmuller TheoryUploaded byAdrián López Cruces
- Smarandache N-structureUploaded byscience2010
- BMat59Uploaded byGonzalo Gutierrez
- T8-9 T.pdfUploaded bymyasmreg
- Partition Congruences and the Andrews–Garvan–Dyson CrankUploaded byapi-26401608
- Lesson Plan PolynomialUploaded byMaria Priscillya Pasaribu

- 151456Uploaded byYesudas Mallipeddi
- mongocUploaded byYesudas Mallipeddi
- sql tutUploaded byYesudas Mallipeddi
- SQL TutorialUploaded byYesudas Mallipeddi
- ch8Uploaded byYesudas Mallipeddi
- (Www.entrance-exam.net)-Infosys Placement Sample Paper 7Uploaded byYesudas Mallipeddi
- BrochureUploaded byYesudas Mallipeddi
- dwdmUploaded byYesudas Mallipeddi