## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

This is an abbreviated version of the Combinatorics Study Group notes by Thomas Britz and Peter Cameron.

1

What is a poset?

A binary relation R on a set X is a set of ordered pairs of elements of X, that is, a subset of X × X. We can represent R by a matrix with rows and columns indexed by X, with (x, y) entry 1 if (x, y) ∈ R, 0 otherwise. The term “poset” is short for “partially ordered set”, that is, a set whose elements are ordered but not all pairs of elements are required to be comparable in the order. Just as an order in the usual sense may be strict (as <) or non-strict (as ≤), there are two versions of the deﬁnition of a partial order: A strict partial order is a binary relation S on a set X satisfying the conditions (R−) for no x ∈ X does (x, x) ∈ S hold; (A−) if (x, y) ∈ S, then (y, x) ∈ S; / (T) if (x, y) ∈ S and (y, z) ∈ S, then (x, z) ∈ S. A non-strict partial order is a binary relation R on a set X satisfying the conditions (R+) for all x ∈ X we have (x, x) ∈ R; (A) if (x, y) ∈ R and (y, x) ∈ R then x = y; (T) if (x, y) ∈ R and (y, z) ∈ R then (x, z) ∈ R. Condition (A−) appears stronger than (A), but in fact (R−) and (A) imply (A−). So we can (as is usually done) replace (A−) by (A) in the deﬁnition of a strict partial order. Conditions (R−), (R+), (A), (T) are called irreﬂexivity, reﬂexivity, antisymmetry and transitivity respectively. We often write x < y if (x, y) ∈ S, and x ≤ y if (x, y) ∈ R. We usually prefer the non-strict version. The two deﬁnitions are essentially the same: we get from one to the other in the obvious way, setting x ≤ y if x < y or x = y, and setting x < y if x ≤ y but x = y. Thus, a poset is a set X carrying a partial order (either strict or non-strict). If there is ambiguity about R, we simply write x ≤R y. A total order is a partial order in which every pair of elements is comparable, that is, the following condition (known as trichotomy) holds:

The Encyclopaedia of Design Theory Posets/1

is deﬁned similarly. R) is a subset C of X which is totally ordered by the restriction of R (that is. y]R is empty if x ≤R y. x = y. 2 Properties of posets An element x of a poset (X. there is no “proof” of Zorn’s Lemma. R). replace it by an element y satisfying x <R y. Dually.u. since by the irreﬂexive and transitive laws the chain can never revisit any element.b.b. An element x is an upper bound for a subset Y of X if y ≤R x for all y ∈ Y . y]R to be the set [x. The concept of a greatest lower bound or g.l. Our proof of the existence of maximal elements in ﬁnite posets indicates why this should be so: the construction requires The Encyclopaedia of Design Theory Posets/2 . A chain in a poset (X. In a poset (X. if it is not maximal. Then X contains a maximal element. The process must terminate. We say that x is a least upper bound or l. y ∈ X. Dually. as we remarked. But in a ﬁnite poset there is always at least one maximal element. and models in which it is false). and y <R x holds. By transitivity. or there may be more than one. repeat until a maximal element is found. Inﬁnite posets (such as Z). need not contain maximal elements. x ≤R y if x divides y) is a locally ﬁnite poset. R) be a poset in which every chain has an upper bound. x is minimal if no element satisﬁes y <R x. The set of positive integers ordered by divisibility (that is. In a general poset there may be no maximal element. R) is called maximal if there is no element y ∈ X satisfying x <R y.• for all x. exactly one of x <R y. We say that the poset is locally ﬁnite if all intervals are ﬁnite. Lower bounds are deﬁned similarly. An antichain is a set A of pairwise incomparable elements. y]R = {z ∈ X : x ≤R z ≤R y}. a totally ordered subset of X). Zorn’s Lemma gives a sufﬁcient condition for maximal elements to exist: Let (X. we deﬁne the interval [x. of Y if it is an upper bound and satisﬁes x ≤R x for any upper bound x . As well known. which can be found as follows: choose any element x. the interval [x. since it is equivalent to the Axiom of Choice (and so there are models of set theory in which it is true. a ﬁnite poset must contain minimal elements.

R). Down-sets are. a chain C and an antichain A have at most one element in common. and its width is the largest cardinality of an antichain. .(in general inﬁnitely many) choices of upper bounds for the elements previously chosen (which form a chain by construction). 3 Hasse diagrams Let x and y be distinct elements of a poset (X. and zi+1 covers zi for i = 0. The set of minimal elements in an up-set is an antichain. Then let A2 be the set of maximal elements in X \ A1 . An up-set in a poset (X. let A1 be the set of maximal elements. R). We denote the height and width of (X. x <R y but no element z satisﬁes x <R z <R y. Then x ≤R y if and only if there exist elements z0 . for example). R) by h(X) and w(X) respectively (suppressing as usual the relation R in the notation). zn (for some non-negative integer n) such that z0 = x. there may be no pairs x and y such that y covers x (this is the case in the rational numbers. Conversely. The complement of an up-set is a down-set. harder to prove. if y ∈ Y and y ≤R z. In general. and it meets every maximal chain. if A is an antichain. and iterate this procedure to ﬁnd the other antichains. The Encyclopaedia of Design Theory Posets/3 . Then there is a partition of X into w(X) chains. . R) is a subset Y of X such that. y}. by deﬁnition this is an antichain. so the numbers of up-sets and antichains in a poset are equal. locally ﬁnite posets are determined by their covering pairs: Proposition 2 Let (X. These two correspondences between up-sets and antichains are mutually inverse. known as Dilworth’s Theorem: Theorem 1 Let (X. y]R = {x. . that is. and x. There is a kind of dual statement. We say that y covers x if [x. of course. R) be a ﬁnite poset. then ↑ (A) = {x ∈ X : a ≤R x for some a ∈ A} is an up-set. zn = y. To see this. so there are equally many up-sets and down-sets. . In a ﬁnite poset (X. However. then z ∈ Y . y ∈ X. In fact there is a partition of X into h(X) antichains. . R) be a locally ﬁnite poset. The height of a poset is the largest cardinality of a chain. . n − 1. . . deﬁned dually. Hence the least number of antichains whose union is X is not less than the size h(X) of the largest chain in X.

¨u r rr ¨ r ¨¨ r r u¨ {1. R) is the directed graph whose vertex set is X and whose arcs are the covering pairs (x.3} Figure 1: A Hasse diagram The Hasse diagram of a poset (X. b) ∈ R (that is. 2. y) in the poset. then the point representing y is higher than the point representing x. b be incomparable elements of X.2} ¨ {1.2. This theorem follows by a ﬁnite number of applications of the next result. the Hasse diagram of the poset of subsets of {1. 3} is shown in Figure 1. For example. This view is borne out by the following theorem. Then no arrows are required in the drawing. if y covers x.) The Encyclopaedia of Design Theory Posets/4 .3} u r ¨ r ¨¨rrr¨¨ r¨ rr ¨r ¨ ¨ r r u¨ r¨ {1} ¨ {2} u {3} u r ¨ rr ¨¨ rr ¨ ¨ ru ¨ / 0 {1. We say that one relation extends another if the second relation (as a set of ordered pairs) is a subset of the ﬁrst. A total order extending R in this sense is referred to as a linear extension of R. Then there is a partial order R extending R such that (a. since the directions of the arrows are implicit. Proposition 4 Let R be a partial order on a set X. and let a. (The term “linear order” is an alternative for “total order”. We usually draw the Hasse diagram of a ﬁnite poset in the plane in such a way that.3} u {2. Theorem 3 Any partial order on a ﬁnite set X can be extended to a total order on X. 4 Linear extensions and dimension One view of a partial order is that it contains partial information about a total order on the underlying set. a < b in the order R ).

that is. an . More precisely. . This deﬁnes a partial order (assuming that each attribute gives a total order). . In other words. but it is strictly weaker than the Axiom of Choice. the elements ai and a j are incomparable. . In other words. assuming the axioms consistent. the elements ai and bi are incomparable. we regard the objects as being incomparable. Theorem 5 Every partial order on a ﬁnite set X is the intersection of some set of total orders on X. To motivate this. . In our motivating example. If each beats the other on some attributes. there is a model in which Theorem 3 is false for some inﬁnite poset. Suppose that a number of products are being compared using several different attributes. we use another model of a partial order. b1 . . given a set S of total orders on X. . . we deﬁne a partial order R on X by x <R y if and only if x <s y for every s ∈ S. Now we have the following result: Proposition 6 The crown on 2n elements has dimension n. it is the smallest number of attributes which could give rise to the observed total order R. and for each i. We regard object a as below object b if b beats a on every attribute. If we assume Zorn’s Lemma. but ai < b j . the conclusion follows. bn is the partial order deﬁned as follows: for all indices i = j.b1 u a1 2 3 u $u r ¨ r $$$ ¨ ¨ rr ¨ $$ ¨¨ $ rr ¨ r ¨¨$$ u rrru $$ rr¨¨ $ ¨ u b b a2 a3 Figure 2: A crown This proof does not immediately show that every inﬁnite partial order can be extended to a total order. The Encyclopaedia of Design Theory Posets/5 . Figure 2 shows the Hasse diagram of the 6-element crown. The crown on 2n elements a1 . Now we deﬁne the dimension of a partial order R to be the smallest number of total orders whose intersection is R. and another model in which Theorem 3 is true for all posets but Zorn’s Lemma is false. R is the intersection of the total orders in S. It cannot be proved from the Zermelo–Fraenkel axioms alone (assuming their consistency). The theorem gives us another measure of the size of a partially ordered set. the Axiom of Choice (or Zorn’s Lemma) cannot be proved from the Zermelo–Fraenkel axioms and this assumption. . the elements bi and b j are incomparable.

then there is no point z such that x ≤R z and z ≤R y. Also. y) = z∈[x. where x1 < . A (R) is closed under multiplication and does indeed form an algebra. . for all (x. Incidentally. then we have µR (x. y]R \ {y}. y) = z∈[x. z). an element f ∈ A (R) is invertible if and only if f (x. This relation allows the M¨ bius function of a poset to be calculated recuro sively. and so ( f g)(x. . y) ∈ R. Now. y]R = {z : x ≤R z ≤R y}: ( f g)(x. and so A (R) consists of upper triangular matrices. . We begin with µR (x. multiplication is given by the rule ( f g)(x. y) = 0. z) = 1 0 if x = y. j) entry of A f is zero if i > j. xn }. ζR (x. . z)g(z. < xn (in the linear order S): this is not essential but is convenient later. Thus. Its inverse (which also lies in A (R)) is the M¨ bius function µR of R. We take any linear order extending R.5 The M¨ bius function o Let R be a partial order on the ﬁnite set X.y]R \{y} ∑ µR (x. if x ≤R y. z If we represent f by the n × n matrix A f with (i. we o have. z)g(z. we see that the (i. . 0 otherwise. and write X = {x1 . We could regard it as a function on R. since f and g vanish on pairs not in R. y) = 0 unless x ≤R y holds. x j ). y). x) = 0 for all x ∈ X. y) = ∑ f (x. Also. otherwise. regarded as a set of ordered pairs. The zeta-function ζR is the matrix representing the relation R as deﬁned earlier. y) = − The Encyclopaedia of Design Theory z∈[x. y). then this is precisely the rule for matrix multiplication. Thus. Posets/6 .y]R ∑ f (x. the element of A (R) deﬁned by 1 if x ≤R y. The incidence algebra A (R) of R is the set of all functions f : X × X → R which satisfy f (x. j) entry f (xi .y]R ∑ µ(x. a subset of the matrix algebra Mn (R). z) for all z ∈ [x. Addition and scalar multiplication are deﬁned pointwise. x) = 1 for all x ∈ X. . the sum can be restricted to the interval [x. if x <R y and we know the values of µ(x. Thus. that is.

In particular. The deﬁnition of the incidence algebra and the M¨ bius function extend immeo diately to locally ﬁnite posets. The Encyclopaedia of Design Theory Posets/7 . We may express this identity as f = gζ where we consider f and g as vectors and where ζ is the zeta function for the lattice of positive integer divisors of n. Theorem 7 implies that g = µ f . f = ζg ⇔ g = µ f . n) (in the notation of the third example above). Example: Suppose that f and g are functions on the natural numbers which are related by the identity f (n) = ∑d|n g(d).W ) = (−1)k q(2) for U ⊆ W . In number theory. y]R . it is obvious. 0 otherwise. the classical M¨ bius function is the function of one variable o given by µ(n) = µ(1. n) f (d) = ∑ µ d|n d|n d n f (d). • The subspaces of a vector space V ⊆ GF(q)n : k µ(U. where k = dimU − dimW . y) = −1 if y covers x. since the sums involved are over intervals [x. which is precisely the classical M¨ bius inversion. We may express this identity as f = ζg where ζ is the zeta function for the lattice of subsets of X. µR (x. or g(A) = B⊇A ∑ µ(A. • The (positive) divisors of an integer n: b r µ(a. Theorem 7 implies that g = f µ. The following are examples of M¨ bius functions. or g(n) = ∑ µ(d. Similarly. B) f (B) = ∑ (−1)|B\A| f (B) B⊇A which is a rather general form of the inclusion/exclusion principle. o • The subsets of a set: µ(A. o Example: Suppose that f and g are functions on the subsets of some ﬁxed (countable) set X which are related by the identity f (A) = ∑B⊇A g(B). B) = (−1)|B\A| for A ⊆ B. b) = (−1) if a is the product of r distinct primes. From the o present point of view. The following result is the M¨ bius inversion for locally ﬁnite posets. Theorem 7 f = gζ ⇔ g = f µ.

∨. • x ∧ 0 = 0. of {x. y} by x ∨ y.l. • Commutative laws: x ∧ y = y ∧ x and x ∨ y = y ∨ x. • The subsets of a (ﬁxed) set: A∧B = A∩B A∨B = A∪B The Encyclopaedia of Design Theory Posets/8 . • Idempotent laws: x ∧ x = x ∨ x = x. It is a lattice in its own right. ∧. for ﬁnite lattices we don’t need to specify 0 and 1 separately. We commonly regard a lattice as being a set with two distinguished elements and two binary operations. Lattices can be axiomatised in terms of the two constants 0 and 1 and the two operations ∨ and ∧. R) with the properties In a lattice. Proposition 8 Let X be a set. ∧ and ∨ two binary operations deﬁned on X. • for any two elements x. by x ∧ y. and the g. A simple example of a poset which is not a lattice is the poset r dr A lattice is a poset (X. y ∈ X. Then (X. A sublattice of a lattice is a subset of the elements containing 0 and 1 and closed under the operations ∨ and ∧. 0. since 0 is just the meet of all elements in the lattice and 1 is their join. The axioms given below are not all independent.b. we denote the l. and 0 and 1 two elements of X. • x ∧ (x ∨ y) = x = x ∨ (x ∧ y). The following are a few examples of lattices.b. though the details are not so important for us.6 Lattices • X has an upper bound 1 and a lower bound 0. x ∨ 1 = 1. y}. r r d . 1) is a lattice if and only if the following axioms are satisﬁed: • Associative laws: x ∧ (y ∧ z) = (x ∧ y) ∧ z and x ∨ (y ∨ z) = (x ∨ y) ∨ z. The result is as follows.u. there is a least upper bound and a greatest lower bound of the set {x. In particular. instead of as a special kind of poset.

N5 .• The subspaces of a vector space: U ∧V = U ∩V U ∨V = span(U ∪V ) • The partitions of a set: R∧T = R∩T R∨T = R∪T Here R ∪ T is the partition whose classes are the connected components of the graph in which two points are adjacent if they lie in the same class of either R or T. and g. The poset of all subsets of a set S (ordered by inclusion) is a distributive lattice: / we have 0 = 0. y. Not only are N5 and M3 the smallest lattices with these properties. in a certain sense. z such that x ≤ z. A lattice is modular if it satisﬁes the modular law (M) x ∨ (y ∧ z) = (x ∨ y) ∧ z for all x. The following theorem states this more precisely. and l. M3 . A lattice is distributive if and only if it contains neither the lattice N5 nor the lattice M3 as a sublattice. every ﬁnite distributive lattice is a sublattice of the lattice of subsets of a set. they are. Theorem 9 A lattice is modular if and only if it does not contain the lattice N5 as a sublattice. which is not distributive. 1 = S. The Encyclopaedia of Design Theory Posets/9 .u. Hence every sublattice of this lattice is a distributive lattice. z.l. We describe how this representation works.b. This is important in that it gives us another way to look at posets. are union and intersection respectively. as well as a modular lattice. the only lattices with these properties. y.b. Conversely. 7 Distributive and modular lattices A lattice is distributive if it satisﬁes the distributive laws (D) x ∧ (y ∨ z) = (x ∧ y) ∨ (x ∧ z) and x ∨ (y ∧ z) = (x ∨ y) ∧ (x ∨ z) for all x. Figure 3 presents a lattice. which is not modular.

I: Theory of M¨ bius o functions. On the foundations of combinatorial theory. if y ∈ Y and z ≤R y. and there is of course a dual form of Theorem 10. A non-zero element x ∈ L is called join-irreducible if. Cambridge. then z ∈ Y . Davey and H. A. we have x = y or x = z. with the operations of union and intersection and the distinguished elements / 0 = 0 and 1 = X. Wahrscheinlichkeitstheorie 2 (1964). R) be a ﬁnite poset. (c) These two operations are mutually inverse. Then the set of down-sets in X. 340–368. A. Let L be a lattice. is a distributive lattice. Then the set X of non-zero join-irreducible elements of L is a sub-poset of L. Theorem 10 (a) Let (X.t t t t tt t d dt t d t dt t d dt N5 Figure 3: Two lattices M3 Let (X. Cambridge University Press. Recall that an down-set in X is a subset Y with the property that. References [1] B. 1990. Peter J. (b) Let L be a ﬁnite distributive lattice. Priestley. R) be a poset.-C. Meet-irreducible elements are deﬁned dually. Cameron May 30. 2003 The Encyclopaedia of Design Theory Posets/10 . Introduction to Lattices and Order. [2] G. Rota. whenever x = y ∨ z. Z.

- Unit 05
- BB0027 2012 Solved Assignment
- SCM SET-1
- Mu 0017
- Solved, MF0012 Winter Drive Assignment
- File System
- bb0028
- Milan Mehta --Wagering Contract Complete Information
- Underwriting of Securities
- Management information system "Organization Structure"
- BB0017 Winter Drive Assignment 2011
- BC0058-SLM-Unit-01
- MC0070 - Set 1
- Solved, MB0039
- Lattices
- Training Development Survey at Bsnl Mba Hr Project Report
- Management Information System
- Contract Law I
- Stages of New Service Development
- KNOWLEDGE MANAGEMENT
- Slm Unit 03 Mbf201
- Final Project Report on Waste Management
- A Study on Working Capital Management
- Conceptual Structure
- Project Feasibility Study & Evaluation (Agricultural Product)
- Cinema Information management system
- 44415065 MB0035 Computer Networks
- Solved, BT0088
- MI0035 SET-1 & SET-2
- Project Feasibility

- Simulation and Modeling MCQ'S
- The Dyslexic Reader 2011 - Issue 58
- UT Dallas Syllabus for ee3300.001 06f taught by William Pervin (pervin)
- Detroit-Edison-Co-Sheets-----------B5-1through-C2-3
- tmp6D2F.tmp
- UT Dallas Syllabus for cs2305.001 06f taught by Nancy Van Ness (nancyvn)
- UT Dallas Syllabus for phys4352.501.10f taught by Ervin Fenyves (ezbd, cunning, glosser)
- Tmp 3008
- bls_2001-1_1979.pdf
- The Information
- Simulation and Modeling MCQ'S
- UT Dallas Syllabus for phys2303.001.10f taught by John Hoffman (jhoffman)
- Sunway University College Canadian International Matriculation Programme 2010 Student Guide
- tmpC906
- UT Dallas Syllabus for math5320.501 06f taught by Thomas Butts (tbutts)
- UT Dallas Syllabus for math2312.002.09s taught by (mxf091000)
- tmp9989.tmp
- bls_2075_1980_2.pdf
- Meeting the Balance of Electricity Supply and Demand in Latin America and the Caribbean
- UT Dallas Syllabus for phys4352.501.09f taught by Ervin Fenyves (ezbd, chaneyr)
- SENATE HEARING, 108TH CONGRESS - A NEW KIND OF SCIENCE
- UT Dallas Syllabus for phys5317.501.08f taught by Roy Chaney (chaneyr)
- UT Dallas Syllabus for sci5333.06m.07u taught by Mary Lena Kelly (mlk023000)
- Contour Free Level Set Method
- 45411_1920-1924
- GMAT Practice Set 10 - Quantitative
- UT Dallas Syllabus for cs2305.002 05f taught by Timothy Farage (tfarage)
- Consolidated-Edison-Co-NY-Inc-Temporary-Rate-Adjustment
- Study of the Barrier Probability in a Classical Risk Model for two different claim amount distributions with same Mean
- Video Math Tutor

- Infrastructur
- asasasa
- Defi Ntions
- Activity Note16 17
- Course Plan Conquest of Knowledge the British in India
- LLD.pdf
- Confusable Words Table Only
- Guidelines
- Note Formulae
- The Jats of Northern Indiatheir Traditional Political System
- wr_csm_2015_eng.pdf
- asasasx
- Proba
- Mukherjee Gadgil
- saasas
- Turney
- 1104.3170
- 662-2571-1-PB
- OmarYayenie
- OmarYayenie
- Licensea
- L-34 Foreign Exchanger Rate
- Graphs With 1-Factors
- 317EL2_Nation and State
- Mathematics Syllabus for Ias Main Exam 2012
- NCPD
- 1104.3170
- India-Pak Marine Boundary Dispute
- Ethics in Governance by Supreme Court
- Mathematics Syllabus for Ias Main Exam 2012

Sign up to vote on this title

UsefulNot usefulClose Dialog## Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

Close Dialog## This title now requires a credit

Use one of your book credits to continue reading from where you left off, or restart the preview.

Loading