1.4. Orders
Definition 1.4.1. A relation \(<\) on a set \(A\) is an order on \(A\) if:
- (Irreflexivity) \(\forall x\in A \ \neg(x<x),\)
- (Asymmetry) \(\forall x,y \in A\ (x<y\rightarrow \neg(y<x)),\)
- (Transitivity) \(\forall x,y,z\in A\ ((x<y \land y<z) \rightarrow x<z).\)
In this case, the set equipped with the order \(<\) is called an ordered set.
Example 1.4.2. Suppose that \(A\) and \(B\) are two sets with orders \(<_A\) and \(<_B\) respectively. A relation \(<_{\text{lex}}\) on \(A\times B\) is the lexicographic order on \(A\times B\) if:
\[(a,b)<_{\mathrm{lex}}(a',b') \quad\Longleftrightarrow\quad a<_A a' \lor (a=a' \land b<_B b').\]Definition 1.4.3. An order \(<\) on a set \(A\) is a total order on \(A\) if it has comparability:
\[\forall x,y\in A\ (x\neq y\Rightarrow ((x<y)\lor (y<x))).\]Definition 1.4.4. Let \(A\) be an ordered set. \(a\in A\) is a least element of \(A\) if
\[\forall x\in A\ (a\le x).\]\(a\in A\) is a minimal element of \(A\) if
\[\forall x\in A\ (x\le a \Rightarrow x=a).\]So least means smaller than everything else, while minimal means nothing else is smaller.
Proposition 1.4.5. Let \(A\) be an ordered set.
- If there exists a least element of \(A,\) then it is minimal and it is unique.
- If \(<\) is a total order, then any minimal element is the least element.
Proof. (1) Let \(a\) be a least element if \(A.\) Then, for all \(x,\) we have \(a\le x.\) If \(x\le a,\) then \(x=a,\) since \(<\) is asymmetric. Hence \(a\) is minimal. Suppose that \(a'\in A\) is a least element. Then we have \(a\le a'\) and \(a'\le a,\) so that \(a=a'.\) Thus \(a\) is the unique element. (2) Let \(a\) be a minimal element. Then we have \(a<x\) if \(x\ne a.\) Also, for all \(x\in A,\) if \(x\ne a\) then \(x<a\) or \(a<x,\) since \(<\) is comparable. Consequently, for all \(x\in A,\) we have \(a\le x.\) Therefore \(a\) is the least element. \(\square\)
Definition 1.4.6. An element \(a\) of the ordered set \(A\) is greatest if \(x\le a\) for all \(x\in A;\) and \(a\) is maximal if \(a\le x\) implies \(x=a.\)
Definition 1.4.7. Let \(A\) be an ordered set. If \(a_1,a_2\in A\) and \(a_1<a_2,\) an open interval is the set \((a_1,a_2)\) defined by
\[(a_1,a_2) = \lbrace x\in A : a_1<x<a_2 \rbrace.\]If this set is empty, \(a_1\) is called the immediate predecessor of \(a_2,\) and \(a_2\) is called the immediate successor of \(a_1.\) The corresponding closed interval is the set \([a_1,a_2]\) defined by
\[[a_1,a_2] = \lbrace x\in A : a_1 \le x \le a_2 \rbrace.\]Definition 1.4.8. Let \(A\) and \(B\) are two sets with order relations \(<_A\) and \(<_B\) respectively. \(A\) and \(B\) have the same order type if there exists a bijective function \(f:A\to B\) such that
\[a_1<_Aa_2 \Longrightarrow f(a_1)<_B f(a_2).\]Definition 1.4.9. Let \(A\) be an ordered set, and \(A_0\subseteq A.\) If there is an element \(a\in A\) such that \(x\leq a\) for all \(x\in A_0,\) \(A_0\) is bounded above and \(a\) is an upper bound for \(A_0.\) If \(a\in A_0,\) then \(a\) is the greatest element of \(A_0.\) Similarly, if there is an element \(a\in A\) such that \(a\leq x\) for all \(x\in A_0,\) \(A_0\) is bounded below and \(a\) is a lower bound for \(A_0.\) If \(a\in A_0,\) then \(a\) is the least element of \(A_0.\)
Definition 1.4.10. Let \(A\) be an ordered set and \(A_0\subseteq A.\) Suppose that \(A_0\) is bounded above and \(a\) is an upper bound for \(A_0.\) If \(a'<a\) implies that \(a'\) is not an upper bound for \(A_0,\) then \(a\) is the least upper bound or the supremum for \(A_0.\) In this case we write
\[a = \sup {A_0}.\]We now suppose that \(A_0\) is bounded below and \(a\) is a lower bound for \(A_0.\) If \(a<a'\) implies that \(a'\) is not a lower bound for \(A_0,\) then \(a\) is the greatest lower bound or the infimum for \(A_0.\) In this case we write
\[a = \inf {A_0}.\]Lemma 1.4.11. Let \(A\) be an ordered set and \(A_0\subseteq A.\) Suppose that \(A_0\) is bounded above and \(a=\sup A_0.\) Then for all \(\varepsilon>0\), there exists \(x\in A_0\) such that
\[a-\varepsilon<x.\]Proof. If there are no elements in the interval \((a-\varepsilon, a],\) then \(a-\varepsilon\) is an upper bound for \(A_0,\) which is a contradiction to the premise that \(a\) is the supremum.\(\square\)
Definition 1.4.12. An ordered set \(A\) has the least-upper-bound property if, for every nonempty subset \(A_0\) that is bounded above, there exists \(a\in A\) such that \(a = \sup {A_0}.\) Similarly, \(A\) has the greatest-lower-bound property if, for every nonempty subset \(A_0\) that is bounded below, there exists \(a\in A\) such that \(a = \inf {A_0}.\)
Theorem 1.4.13. Suppose that \(A\) is an ordered set with the least-upper-bound property, \(A_0\) is a nonempty subset of \(A,\) and \(A_0\) is bounded below. Let \(L\) be the set of all lower bounds for \(A_0.\) Then
\[\exists a\in A \ (a = \sup L) \qquad \text{and} \qquad a = \inf A_0 .\]Proof. Since \(A_0\) is bounded below, \(L\) is not empty. Since every \(y\in A_0\) is an upper bound for \(L,\) \(L\) is bounded above. Given that \(A\) has the least-upper-bound property, there exists \(x\in L\) such that \(x=\sup L;\) we let \(a = \sup L.\) Since every \(y\in A_0\) is an upper bound for \(L\) and \(a\) is the least upper bound for \(L,\) \(a\) is a lower bound for \(A_0.\) Let \(x\) be any lower bound of \(A_0.\) By the definition of \(L,\) \(x\in L.\) Since \(a=\sup L,\) \(x\le a.\) Consequently, \(a\) is the greatest lower bound for \(A_0.\) \(\square\)
Exercises
Exercise 1. Prove that the lexicographic order is an order relation.
Proof. Assume that \(<_A\) is an order relation on \(A,\) \(<_B\) is an order relation on \(B,\) and \(<_{\mathrm{lex}}\) is the lexicographic order on \(A\times B.\) Firstly, to check whether the lexicographic order is irreflexive, we determine the truth value of the statement \((a,b)<_{\mathrm{lex}}(a,b).\) It is equivalent to
\[a<_A a \lor \bigl(a=a \land b<_B b\bigr).\]Since \(<_A\) and \(<_B\) are irreflexive, both alternatives are false so that \(<_{\mathrm{lex}}\) is irreflexive. We now check the comparability of \(<_{\mathrm{lex}}.\) If \((a,b)\neq (a',b')\) then \(a\neq a',\) or \(a=a'\) and \(b\neq b'.\) We first suppose that \(a\neq a'.\) Since \(<_A\) is comparable, \(a<_Aa'\) or \(a'<_Aa.\) If \(a<_Aa'\) then \((a,b)<_{\mathrm{lex}}(a',b').\) Similarly, if \(a'<_Aa\) then \((a',b')<_{\mathrm{lex}}(a,b).\) We now suppose that \(a=a'\) and \(b\neq b'.\) Since \(<_B\) is comparable, \(b<_Bb'\) or \(b'<_Bb.\) If \(b<_Bb'\) then \((a,b)<_{\mathrm{lex}}(a',b').\) If \(b'<_Bb\) then \((a',b')<_{\mathrm{lex}}(a,b).\) Therefore \(<_{\mathrm{lex}}\) is comparable. Finally we check the transitivity. We suppose that \((a,b)<_{\mathrm{lex}} (a',b')\) and \((a',b')<_{\mathrm{lex}} (a'',b'').\) It splits into four cases. Firstly, if \(a<_A a'\) and \(a' <_A a''\) then \(a<_Aa''.\) SecondIy, if \(a=a'\) and \(a' <_A a''\) then \(a<_Aa''.\) Thirdly, if \(a' = a''\) then \(a<_Aa''.\) Lastly, if \(a=a'\) and \(b<_B b',\) and \(a'=a''\) and \(b'<_B b'',\) then \(a=a''\) and \(b<_B b''.\) Therefore \(<_{\mathrm{lex}}\) is transitive. Since \(<_{\mathrm{lex}}\) is irreflexive, comparable, and transitive, \(<_{\mathrm{lex}}\) itself is an order relation on \(A\times B.\)\(\square\)
Exercise 2. Define a relation on \(\mathbb R^2\) by setting \(\left(x_0, y_0\right)<\left(x_1, y_1\right)\) if either \(y_0-x_0^2<y_1-x_1^2,\) or \(y_0-x_0^2=y_1-x_1^2\) and \(x_0<x_1.\) Show that this is an order relation on \(\mathbb R^2,\) and describe it geometrically.
Proof. Define \(F:\mathbb R^2\to \mathbb R^2\) by
\[F(x,y)=(y-x^2,x).\]Then, by the definition of the given relation,
\[(x_0,y_0)<(x_1,y_1) \quad\Longleftrightarrow\quad F(x_0,y_0)<_{\mathrm{lex}}F(x_1,y_1).\]Thus the relation \(<\) on \(\mathbb R^2\) is the relation obtained by pulling back the lexicographic order through \(F\). We first observe that \(F\) is injective. Suppose \(F(x_0,y_0)=F(x_1,y_1).\) Then \((y_0-x_0^2,x_0)=(y_1-x_1^2,x_1).\) Hence \(x_0=x_1\), and therefore \(y_0-x_0^2=y_1-x_1^2\) implies \(y_0=y_1\). Thus \((x_0,y_0)=(x_1,y_1)\), so \(F\) is injective. We now prove that \(<\) is an order relation. First, \(<\) is irreflexive. For any \((x,y)\in\mathbb R^2\), \(F(x,y)=F(x,y),\) so \(F(x,y)<_{\mathrm{lex}}F(x,y)\) is false because \(<_{\mathrm{lex}}\) is irreflexive. Therefore \((x,y)<(x,y)\) is false, and \(<\) is irreflexive. Next, \(<\) is transitive. Suppose \((x_0,y_0)<(x_1,y_1)\) and \((x_1,y_1)<(x_2,y_2).\) Then \(F(x_0,y_0)<_{\mathrm{lex}}F(x_1,y_1)\) and \(F(x_1,y_1)<_{\mathrm{lex}}F(x_2,y_2).\) Since \(<_{\mathrm{lex}}\) is transitive, \(F(x_0,y_0)<_{\mathrm{lex}}F(x_2,y_2).\) Therefore \((x_0,y_0)<(x_2,y_2).\) Thus \(<\) is transitive. It remains to check comparability. Let \((x_0,y_0)\) and \((x_1,y_1)\) be distinct points of \(\mathbb R^2\). Since \(F\) is injective, \(F(x_0,y_0)\neq F(x_1,y_1).\) Since \(<_{\mathrm{lex}}\) is a strict total order on \(\mathbb R^2\), \(F(x_0,y_0)<_{\mathrm{lex}}F(x_1,y_1)\) or \(F(x_1,y_1)<_{\mathrm{lex}}F(x_0,y_0).\) Therefore \((x_0,y_0)<(x_1,y_1)\) or \((x_1,y_1)<(x_0,y_0).\) Hence \(<\) is a strict total order on \(\mathbb R^2\). Geometrically, the ordered pairs correspond to the points on a plane with Cartesian coordinates system. The order first compares the value of \(y-x^2\). Given that the point \((x_0,y_0)\) is on the parabola \(y=x^2+c,\) where \(c=y_0-x_0^2,\) the order is understood to have points on lower parabola \(y=x^2+c_0\) first, and points on higher parabola \(y=x^2+c_1\) next, when \(c_0<c_1.\) In the next step, the points that lie on the same parabola are ordered by their \(x\)-coordinates from left to right.\(\square\)
References
- Munkres, J. R. (2000). Topology (2nd ed.). Pearson.
- Rudin, W. (1976). Principles of mathematical analysis (3rd ed.). McGraw-Hill Education.
- Cameron, P. J. (1998). Sets, logic and categories. Springer. https://doi.org/10.1007/978-1-4471-0589-3