DOCSLIB.ORG

Total-Order and Partial-Order Planning: a Comparative Analysis

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Total-Order and Partial-Order Planning: a Comparative Analysis Journal of Articial Intelligence Research Submitted published TotalOrder and PartialOrder Planning A Comparative Analysis Steven Minton mintonptolemyarcnasagov John Bresina bresinaptolemyarcna sagov Mark Drummond medptolemyarcnasagov Recom Technologies NASA Ames Research Center Mail Stop Moett Field CA USA Abstract For many years the intuitions underlying partialorder planning were largely taken for granted Only in the past few years has there b een renewed interest in the fundamental principles underlying this paradigm In this pap er we present a rigorous comparative analysis of partialorder and totalorder planning by fo cusing on two sp ecic planners that can b e directly compared We show that there are some subtle assumptions that underly the widespread intuitions regarding the supp osed eciency of partialorder planning For instance the sup eriority of partialorder planning can dep end critically up on the search strategy and the structure of the search space Understanding the underlying assumptions is crucial for constructing ecient planners Intro duction For many years the sup eriority of partialorder planners over totalorder planners has b een tacitly assumed by the planning community Originally partialorder planning was intro duced by Sacerdoti as a way to improve planning eciency by avoiding premature commitments to a particular order for achieving subgoals The utility of partialorder planning was demonstrated anecdotally by showing how such a planner could eciently solve blo cksworld examples such as the wellknown Sussman anomaly Since partialorder planning intuitively seems like a go o d idea little attention has b een devoted to analyzing its utility at least until recently Minton Bresina Drummond a Barrett Weld Kambhampati c However if one lo oks closely at the issues involved a numb er of questions arise For example do the advantages of partial order planning hold regardless of the search strategy used Do the advantages hold when the planning language is so expressive that reasoning ab out partially ordered plans is intractable eg if the language allows conditional eects Our work Minton et al a has shown that the situation is much more inter esting than might b e exp ected We have found that there are some unstated assumptions underlying the supp osed eciency of partialorder planning For instance the sup eriority of partialorder planning can dep end critically up on the search strategy and search heuristics employed This pap er summarizes our observations regarding partialorder and totalorder plan ning We b egin by considering a simple totalorder planner and a closely related partial order planner and establishing a mapping b etween their search spaces We then examine c AI Access Foundation and Morgan Kaufmann Publishers All rights reserved Minton Bresina Drummond the relative sizes of their search spaces demonstrating that the partialorder planner has a fundamental advantage b ecause the size of its search space is always less than or equal to that of the totalorder planner However this advantage do es not necessarily translate into an eciency gain this dep ends on the typ e of search strategy used For example we describ e a domain where our partial order planner is more ecient than our total order planner when depthrst search is used but the eciency gain is lost when an iterative sampling strategy is used We also show that partialorder planners can have a second indep endent advantage when certain typ es of op erator ordering heuristics are employed This heuristic advantage underlies Sacerdotis anecdotal examples explaining why leastcommitment works However in our blo cksworld exp eriments this second advantage is relatively unimp ortant compared to the advantage derived from the reduction in search space size Finally we lo ok at how our results extend to partialorder planners in general We describ e how the advantages of partialorder planning can b e preserved even if highly ex pressive languages are used We also show that the advantages do not necessarily hold for all partialorder planners but dep end critically on the construction of the planning space Background Planning can b e characterized as search through a space of p ossible plans A totalorder planner searches through a space of totally ordered plans a partialorder planner is dened analogously We use these terms rather than the terms linear and nonlinear b ecause the latter are overloaded For example some authors have used the term nonlinear when fo cusing on the issue of goal ordering That is some linear planners when solving a conjunctive goal require that all subgoals of one conjunct b e achieved b efore subgoals of the others hence planners that can arbitrarily interleave subgoals are often called nonlinear This version of the linearnonlinear distinction is dierent than the partialordertotal order distinction investigated here The former distinction impacts planner completeness whereas the totalorderpartialorder distinction is orthogonal to this issue Drummond Currie Minton et al a The totalorderpartialorder distinction should also b e kept separate from the distinc tion b etween worldbased planners and planbased planners The distinction is one of mo deling in a worldbased planner each search state corresp onds to a state of the world and in a planbased planner each search state corresp onds to a plan While total order planners are commonly asso ciated with worldbased planners such as Strips several wellknown totalorder planners have b een planbased such as Waldingers regression plan ner Waldinger Interplan Tate and Warplan Warren Similarly partialorder planners are commonly planbased but it is p ossible to have a worldbased partialorder planner Go defroid Kabanza In this pap er we fo cus solely on the totalorderpartialorder distinction in order to avoid complicating the analysis We claim that the only signicant dierence b etween partialorder and totalorder plan ners is planning eciency It might b e argued that partialorder planning is preferable b ecause a partially ordered plan can b e more exibly executed However execution exibil ity can also b e achieved with a totalorder planner and a p ostpro cessing step that removes unnecessary orderings from the totally ordered solution plan to yield a partial order Back TotalOrder and PartialOrder Planning strom Veloso Perez Carb onell Regnier Fade The p olynomial time complexity of this p ostpro cessing is negligible compared to the search time for plan generation Hence we b elieve that execution exibility is at b est a weak justication for the supp osed sup eriority of partialorder planning In the following sections we analyze the relative eciency of partialorder and total order planning by considering a totalorder planner and a partialorder planner that can b e directly compared Elucidating the key dierences b etween these planning algorithms reveals some imp ortant principles that are of general relevance Terminology A plan consists of an ordered set of steps where each step is a unique op erator instance Plans can b e total ly ordered in which case every step is ordered with resp ect to every other step or partial ly ordered in which case steps can b e unordered with resp ect to each other We assume that a library of op erators is available where each op erator has preconditions deleted conditions and added conditions All of these conditions must b e nonnegated prop o sitions and we adopt the common convention that each deleted condition is a precondition Later in this pap er we show how our results can b e extended to more expressive languages but this simple language is sucient to establish the essence of our argument A linearization of a partially ordered plan is a total order over the plans steps that is consistent with the existing partial order In a totally ordered plan a precondition of a plan step is true if it is added by an earlier step and not deleted by an intervening step In a partially ordered plan a steps precondition is possibly true if there exists a linearization in which it is true and a steps precondition is necessarily true if it is true in al l linearizations A steps precondition is necessarily false if it is not p ossibly true A state consists of a set of prop ositions A planning problem is dened by an initial state and a set of goals where each goal is a prop osition For convenience we represent a problem as a twostep initial plan where the prop ositions that are true in the initial state are added by the rst step and the goal prop ositions are the preconditions of the nal step The planning pro cess starts with this initial plan and searches through a space of p ossible plans A successful search terminates with a solution plan ie a plan in which all steps preconditions are necessarily true The search space can b e characterized as a tree where each no de corresp onds to a plan and each arc corresp onds to a plan transformation Each transformation incrementally extends ie renes a plan by adding additional steps or orderings Thus each leaf in the search tree corresp onds either to a solution plan or a deadend and each intermediate no de corresp onds to an unnished plan which can b e further extended Backstrom formalizes the problem of removing unnecessary orderings in order to pro duce a least constrained plan He shows that the problem is p olynomial if one denes a leastconstrained plan as a plan in which no orderings can b e removed without impacting the correctness of the plan Backstrom also shows that the problem of nding a plan with
Load more

Recommended publications
  • Scott Spaces and the Dcpo Category
    SCOTT SPACES AND THE DCPO CATEGORY JORDAN BROWN Abstract. Directed-complete partial orders (dcpo’s) arise often in the study of λ-calculus. Here we investigate certain properties of dcpo’s and the Scott spaces they induce. We introduce a new construction which allows for the canonical extension of a partial order to a dcpo and give a proof that the dcpo introduced by Zhao, Xi, and Chen is well-filtered. Contents 1. Introduction 1 2. General Definitions and the Finite Case 2 3. Connectedness of Scott Spaces 5 4. The Categorical Structure of DCPO 6 5. Suprema and the Waybelow Relation 7 6. Hofmann-Mislove Theorem 9 7. Ordinal-Based DCPOs 11 8. Acknowledgments 13 References 13 1. Introduction Directed-complete partially ordered sets (dcpo’s) often arise in the study of λ-calculus. Namely, they are often used to construct models for λ theories. There are several versions of the λ-calculus, all of which attempt to describe the ‘computable’ functions. The first robust descriptions of λ-calculus appeared around the same time as the definition of Turing machines, and Turing’s paper introducing computing machines includes a proof that his computable functions are precisely the λ-definable ones [5] [8]. Though we do not address the λ-calculus directly here, an exposition of certain λ theories and the construction of Scott space models for them can be found in [1]. In these models, computable functions correspond to continuous functions with respect to the Scott topology. It is thus with an eye to the application of topological tools in the study of computability that we investigate the Scott topology.
    [Show full text]
  • Section 8.6 What Is a Partial Order?
    Announcements ICS 6B } Regrades for Quiz #3 and Homeworks #4 & Boolean Algebra & Logic 5 are due Today Lecture Notes for Summer Quarter, 2008 Michele Rousseau Set 8 – Ch. 8.6, 11.6 Lecture Set 8 - Chpts 8.6, 11.1 2 Today’s Lecture } Chapter 8 8.6, Chapter 11 11.1 ● Partial Orderings 8.6 Chapter 8: Section 8.6 ● Boolean Functions 11.1 Partial Orderings (Continued) Lecture Set 8 - Chpts 8.6, 11.1 3 What is a Partial Order? Some more defintions Let R be a relation on A. The R is a partial order iff R is: } If A,R is a poset and a,b are A, reflexive, antisymmetric, & transitive we say that: } A,R is called a partially ordered set or “poset” ● “a and b are comparable” if ab or ba } Notation: ◘ i.e. if a,bR and b,aR ● If A, R is a poset and a and b are 2 elements of A ● “a and b are incomparable” if neither ab nor ba such that a,bR, we write a b instead of aRb ◘ i.e if a,bR and b,aR } If two objects are always related in a poset it is called a total order, linear order or simple order. NOTE: it is not required that two things be related under a partial order. ● In this case A,R is called a chain. ● i.e if any two elements of A are comparable ● That’s the “partial” of it. ● So for all a,b A, it is true that a,bR or b,aR 5 Lecture Set 8 - Chpts 8.6, 11.1 6 1 Now onto more examples… More Examples Let Aa,b,c,d and let R be the relation Let A0,1,2,3 and on A represented by the diagraph Let R0,01,1, 2,0,2,2,2,33,3 The R is reflexive, but We draw the associated digraph: a b not antisymmetric a,c &c,a It is easy to check that R is 0 and not 1 c d transitive d,cc,a, but not d,a Refl.
    [Show full text]
  • [Math.NT] 1 Nov 2006
    ADJOINING IDENTITIES AND ZEROS TO SEMIGROUPS MELVYN B. NATHANSON Abstract. This note shows how iteration of the standard process of adjoining identities and zeros to semigroups gives rise naturally to the lexicographical ordering on the additive semigroups of n-tuples of nonnegative integers and n-tuples of integers. 1. Semigroups with identities and zeros A binary operation ∗ on a set S is associative if (a ∗ b) ∗ c = a ∗ (b ∗ c) for all a,b,c ∈ S. A semigroup is a nonempty set with an associative binary operation ∗. The semigroup is abelian if a ∗ b = b ∗ a for all a,b ∈ S. The trivial semigroup S0 consists of a single element s0 such that s0 ∗ s0 = s0. Theorems about abstract semigroups are, in a sense, theorems about the pure process of multiplication. An element u in a semigroup S is an identity if u ∗ a = a ∗ u = a for all a ∈ S. If u and u′ are identities in a semigroup, then u = u ∗ u′ = u′ and so a semigroup contains at most one identity. A semigroup with an identity is called a monoid. If S is a semigroup that is not a monoid, that is, if S does not contain an identity element, there is a simple process to adjoin an identity to S. Let u be an element not in S and let I(S,u)= S ∪{u}. We extend the binary operation ∗ from S to I(S,u) by defining u ∗ a = a ∗ u = a for all a ∈ S, and u ∗ u = u. Then I(S,u) is a monoid with identity u.
    [Show full text]
  • Arxiv:1811.03543V1 [Math.LO]
    PREDICATIVE WELL-ORDERING NIK WEAVER Abstract. Confusion over the predicativist conception of well-ordering per- vades the literature and is responsible for widespread fundamental miscon- ceptions about the nature of predicative reasoning. This short note aims to explain the core fallacy, first noted in [9], and some of its consequences. 1. Predicativism Predicativism arose in the early 20th century as a response to the foundational crisis which resulted from the discovery of the classical paradoxes of naive set theory. It was initially developed in the writings of Poincar´e, Russell, and Weyl. Their central concern had to do with the avoidance of definitions they considered to be circular. Most importantly, they forbade any definition of a real number which involves quantification over all real numbers. This version of predicativism is sometimes called “predicativism given the nat- ural numbers” because there is no similar prohibition against defining a natural number by means of a condition which quantifies over all natural numbers. That is, one accepts N as being “already there” in some sense which is sufficient to void any danger of vicious circularity. In effect, predicativists of this type consider un- countable collections to be proper classes. On the other hand, they regard countable sets and constructions as unproblematic. 2. Second order arithmetic Second order arithmetic, in which one has distinct types of variables for natural numbers (a, b, ...) and for sets of natural numbers (A, B, ...), is thus a good setting for predicative reasoning — predicative given the natural numbers, but I will not keep repeating this. Here it becomes easier to frame the restriction mentioned above in terms of P(N), the power set of N, rather than in terms of R.
    [Show full text]
  • Learning Binary Relations and Total Orders
    Learning Binary Relations and Total Orders Sally A Goldman Ronald L Rivest Department of Computer Science MIT Lab oratory for Computer Science Washington University Cambridge MA St Louis MO rivesttheorylcsmitedu sgcswustledu Rob ert E Schapire ATT Bell Lab oratories Murray Hill NJ schapireresearchattcom Abstract We study the problem of learning a binary relation b etween two sets of ob jects or b etween a set and itself We represent a binary relation b etween a set of size n and a set of size m as an n m matrix of bits whose i j entry is if and only if the relation holds b etween the corresp onding elements of the twosetsWe present p olynomial prediction algorithms for learning binary relations in an extended online learning mo del where the examples are drawn by the learner by a helpful teacher by an adversary or according to a uniform probability distribution on the instance space In the rst part of this pap er we present results for the case that the matrix of the relation has at most k rowtyp es We present upp er and lower b ounds on the number of prediction mistakes any prediction algorithm makes when learning such a matrix under the extended online learning mo del Furthermore we describ e a technique that simplies the pro of of exp ected mistake b ounds against a randomly chosen query sequence In the second part of this pap er we consider the problem of learning a binary re lation that is a total order on a set We describ e a general technique using a fully p olynomial randomized approximation scheme fpras to implement a randomized
    [Show full text]
  • Order Relations and Functions
    Order Relations and Functions Problem Session Tonight 7:00PM – 7:50PM 380-380X Optional, but highly recommended! Recap from Last Time Relations ● A binary relation is a property that describes whether two objects are related in some way. ● Examples: ● Less-than: x < y ● Divisibility: x divides y evenly ● Friendship: x is a friend of y ● Tastiness: x is tastier than y ● Given binary relation R, we write aRb iff a is related to b by relation R. Order Relations “x is larger than y” “x is tastier than y” “x is faster than y” “x is a subset of y” “x divides y” “x is a part of y” Informally An order relation is a relation that ranks elements against one another. Do not use this definition in proofs! It's just an intuition! Properties of Order Relations x ≤ y Properties of Order Relations x ≤ y 1 ≤ 5 and 5 ≤ 8 Properties of Order Relations x ≤ y 1 ≤ 5 and 5 ≤ 8 1 ≤ 8 Properties of Order Relations x ≤ y 42 ≤ 99 and 99 ≤ 137 Properties of Order Relations x ≤ y 42 ≤ 99 and 99 ≤ 137 42 ≤ 137 Properties of Order Relations x ≤ y x ≤ y and y ≤ z Properties of Order Relations x ≤ y x ≤ y and y ≤ z x ≤ z Properties of Order Relations x ≤ y x ≤ y and y ≤ z x ≤ z Transitivity Properties of Order Relations x ≤ y Properties of Order Relations x ≤ y 1 ≤ 1 Properties of Order Relations x ≤ y 42 ≤ 42 Properties of Order Relations x ≤ y 137 ≤ 137 Properties of Order Relations x ≤ y x ≤ x Properties of Order Relations x ≤ y x ≤ x Reflexivity Properties of Order Relations x ≤ y Properties of Order Relations x ≤ y 19 ≤ 21 Properties of Order Relations x ≤ y 19 ≤ 21 21 ≤ 19? Properties of Order Relations x ≤ y 19 ≤ 21 21 ≤ 19? Properties of Order Relations x ≤ y 42 ≤ 137 Properties of Order Relations x ≤ y 42 ≤ 137 137 ≤ 42? Properties of Order Relations x ≤ y 42 ≤ 137 137 ≤ 42? Properties of Order Relations x ≤ y 137 ≤ 137 Properties of Order Relations x ≤ y 137 ≤ 137 137 ≤ 137? Properties of Order Relations x ≤ y 137 ≤ 137 137 ≤ 137 Antisymmetry A binary relation R over a set A is called antisymmetric iff For any x ∈ A and y ∈ A, If xRy and y ≠ x, then yRx.
    [Show full text]
  • Partial Orders CSE235 Partial Orders
    Partial Orders CSE235 Partial Orders Slides by Christopher M. Bourke Instructor: Berthe Y. Choueiry Spring 2006 Computer Science & Engineering 235 Introduction to Discrete Mathematics Sections 7.6 of Rosen 1 / 1 [email protected] Partial Orders I Motivating Introduction Partial Orders CSE235 Consider the recent renovation of Avery Hall. In this process several things had to be done. Remove Asbestos Replace Windows Paint Walls Refinish Floors Assign Offices Move in Office-Furniture. 2 / 1 Partial Orders II Motivating Introduction Partial Orders CSE235 Clearly, some things had to be done before others could even begin—Asbestos had to be removed before anything; painting had to be done before the floors to avoid ruining them, etc. On the other hand, several things could have been done concurrently—painting could be done while replacing the windows and assigning office could have been done at anytime. Such a scenario can be nicely modeled using partial orderings. 3 / 1 Partial Orderings I Definition Partial Orders CSE235 Definition A relation R on a set S is called a partial order if it is reflexive, antisymmetric and transitive. A set S together with a partial ordering R is called a partially ordered set or poset for short and is denoted (S, R) Partial orderings are used to give an order to sets that may not have a natural one. In our renovation example, we could define an ordering such that (a, b) ∈ R if a must be done before b can be done. 4 / 1 Partial Orderings II Definition Partial Orders CSE235 We use the notation a 4 b to indicate that (a, b) ∈ R is a partial order and a ≺ b when a 6= b.
    [Show full text]
  • 3 Limits of Sequences and Filters
    3 Limits of Sequences and Filters The Axiom of Choice is obviously true, the well-ordering theorem is obviously false; and who can tell about Zorn’s Lemma? —Jerry Bona (Schechter, 1996) Introduction. Chapter 2 featured various properties of topological spaces and explored their interactions with a few categorical constructions. In this chapter we’ll again discuss some topological properties, this time with an eye toward more fine-grained ideas. As introduced early in a study of analysis, properties of nice topological spaces X can be detected by sequences of points in X. We’ll be interested in some of these properties and the extent to which sequences suffice to detect them. But take note of the adjective “nice” here. What if X is any topological space, not just a nice one? Unfortunately, sequences are not well suited for characterizing properties in arbitrary spaces. But all is not lost. A sequence can be replaced with a more general construction—a filter—which is much better suited for the task. In this chapter we introduce filters and highlight some of their strengths. Our goal is to spend a little time inside of spaces to discuss ideas that may be familiar from analysis. For this reason, this chapter contains less category theory than others. On the other hand, we’ll see in section 3.3 that filters are a bit like functors and hence like generalizations of points. This perspective thus gives us a coarse-grained approach to investigating fine-grained ideas. We’ll go through some of these basic ideas—closure, limit points, sequences, and more—rather quickly in sections 3.1 and 3.2.
    [Show full text]
  • Ordered Sets
    Ordered Sets Motivation When we build theories in political science, we assume that political agents seek the best element in an appropriate set of feasible alternatives. Voters choose their favorite candidate from those listed on the ballot. Citizens may choose between exit and voice when unsatisfied with their government. Consequently, political science theory requires that we can rank alternatives and identify the best element in various sets of choices. Sets whose elements can be ranked are called ordered sets. They are the subject of this lecture. Relations Given two sets A and B, a binary relation is a subset R ⊂ A × B. We use the notation (a; b) 2 R or more often aRb to denote the relation R holding for an ordered pair (a; b). This is read \a is in the relation R to b." If R ⊂ A × A, we say that R is a relation on A. Example. Let A = fAustin, Des Moines, Harrisburgg and let B = fTexas, Iowa, Pennsylvaniag. Then the relation R = f(Austin, Texas), (Des Moines, Iowa), (Harrisburg, Pennsylvania)g expresses the relation \is the capital of." Example. Let A = fa; b; cg. The relation R = f(a; b); (a; c); (b; c)g expresses the relation \occurs earlier in the alphabet than." We read aRb as \a occurs earlier in the alphabet than b." 1 Properties of Binary Relations A relation R on a nonempty set X is reflexive if xRx for each x 2 X complete if xRy or yRx for all x; y 2 X symmetric if for any x; y 2 X, xRy implies yRx antisymmetric if for any x; y 2 X, xRy and yRx imply x = y transitive if xRy and yRz imply xRz for any x; y; z 2 X Any relation which is reflexive and transitive is called a preorder.
    [Show full text]
  • Partial Order Relations
    Partial Order Relations We defined three properties of relations: reflexivity, symmetry and transitivity. A fourth property of relations is anti-symmetry. Definition of Antisymmetry:: Let R be a relation on a set A, R is antisymmetric if , and only if, for all a and b in A, if a R b and b R a then a=b. Testing for Antisymmetry of finite Relations: Let R1 and R2 be the relations on {0, 1, 2} defined as follows: a- R1= { (0,2 ), (1, 2), (2, 0)} b- R2 = { (0,0), (0, 1), (0, 2), (1, 1), (1, 2)} Draw a directed graph for R1 and R2 and indicate which relation is antisymmetric? a- R1 is not antisymmetric b- R2 is antisymmetric Testing for Antisymmetry of “Divides” relations. Let R1 be the “divides” relation on the set of all positive integers, and let R2 be the “divides” relation on the set of all integers. + For all a, b ∈ Z , a R1 b ⇔ a | b For all a, b ∈ Z, a R2 b ⇔ a | b a- Is R1 antisymmetric ? Prove or give a counterexample b- Is R2 antisymmetric ? Prove or give a counter example. Solution: -R1 is antisymmetric -R2 is not antisymmetric Partial Order Relations: Let R be a binary relation defined on a set A. R is a partial order relation if, and only if, R is reflexive, antisymmetric and transitive. Two fundamental partial order relations are the “less than or equal to” relation on a set of real numbers and the “subset” relation on a set of sets. The “Subset” Relation: Let A be any collection of sets and define the subset relation ⊆ on A as follows: For all U, V ∈ A U ⊆ V ⇔ for all x, if x ∈ U then x ∈ V.
    [Show full text]
  • Relations and Partial Orders
    “mcs-ftl” — 2010/9/8 — 0:40 — page 213 — #219 7 Relations and Partial Orders A relation is a mathematical tool for describing associations between elements of sets. Relations are widely used in computer science, especially in databases and scheduling applications. A relation can be defined across many items in many sets, but in this text, we will focus on binary relations, which represent an association between two items in one or two sets. 7.1 Binary Relations 7.1.1 Definitions and Examples Definition 7.1.1. Given sets A and B, a binary relation R A B from1 A W ! to B is a subset of A B. The sets A and B are called the domain and codomain of R, respectively. We commonly use the notation aRb or a R b to denote that .a; b/ R. 2 A relation is similar to a function. In fact, every function f A B is a rela- W ! tion. In general, the difference between a function and a relation is that a relation might associate multiple elements ofB with a single element ofA, whereas a func- tion can only associate at most one element of B (namely, f .a/) with each element a A. 2 We have already encountered examples of relations in earlier chapters. For ex- ample, in Section 5.2, we talked about a relation between the set of men and the set of women where mRw if man m likes woman w. In Section 5.3, we talked about a relation on the set of MIT courses where c1Rc2 if the exams for c1 and c2 cannot be given at the same time.
    [Show full text]
  • Rel.1 Orders
    rel.1 Orders sfr:rel:ord: Many of our comparisons involve describing some objects as being \less than", explanation sec \equal to", or \greater than" other objects, in a certain respect. These involve order relations. But there are different kinds of order relations. For instance, some require that any two objects be comparable, others don't. Some include identity (like ≤) and some exclude it (like <). It will help us to have a taxonomy here. Definition rel.1 (Preorder). A relation which is both reflexive and transi- tive is called a preorder. Definition rel.2 (Partial order). A preorder which is also anti-symmetric is called a partial order. Definition rel.3 (Linear order). A partial order which is also connected is called a total order or linear order. Example rel.4. Every linear order is also a partial order, and every partial order is also a preorder, but the converses don't hold. The universal relation on A is a preorder, since it is reflexive and transitive. But, if A has more than one element, the universal relation is not anti-symmetric, and so not a partial order. ∗ Example rel.5. Consider the no longer than relation 4 on B : x 4 y iff len(x) ≤ len(y). This is a preorder (reflexive and transitive), and even con- nected, but not a partial order, since it is not anti-symmetric. For instance, 01 4 10 and 10 4 01, but 01 6= 10. Example rel.6. An important partial order is the relation ⊆ on a set of sets. This is not in general a linear order, since if a 6= b and we consider }(fa; bg) = f;; fag; fbg; fa; bgg, we see that fag * fbg and fag 6= fbg and fbg * fag.
    [Show full text]