TheoremBase

Infima come from suprema of negatives; the Archimedean property follows because a supremum of the images nFn_F of all natural numbers would be exceeded by some (n+1)_F; density scales the gap by a natural number, picks the least natural number above the shifted point, and maps the resulting rational through the canonical embedding, using that nFn_F equals its image under the embedding; rational approximation from below follows from density.

Proof

Each result cited is universally quantified over the data in its own statement.

Notation. Below we write rr for FF, 0r0_{r} and 1r1_{r} for the zero and the unit of FF, and κ\kappa for the canonical embedding κF:Q→r\kappa_{F}:\mathbb{Q}\to r. Operations and orders applied to rational numbers are those of Q\mathbb{Q}, and those applied to elements of rr, in particular to values of κ\kappa, are those of rr, as The Natural Numbers and the Natural Numbers with Zero: Arithmetic, Order, Induction and Recursion §overloading allows. By Sets and Maps: Ordinary Notation §orders, << is the strict relation of ≤\le, and bounds, suprema and infima of subsets of rr are as in Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order, formed in rr. For n∈N0n\in\mathbb{N}_{0} we write nrn_{r} for the image of nn in rr, which is the element of rr that nn denotes by Commutative Rings, Fields and Ordered Fields: Standard Notation §numerals; thus clause archimedean of the statement says that x<nrx<n_{r} for some n∈Nn\in\mathbb{N}. A natural number standing where a rational number is required, such as an operand of a difference or a quotient in Q\mathbb{Q} or an argument of κ\kappa, denotes its image in Q\mathbb{Q}, by The Integers and the Rational Numbers, with the Natural Numbers and the Integers Identified with Subsets of the Rationals §identification. Clauses of the statement are cited below by their names.

Preliminaries. By hypothesis, rr is an ordered field; so rr is a set, ++ and ⋅\cdot are binary operations on it, and ≤\le is a total order on rr, by Fields §field and Ordered Fields §ordered-field. Hence Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order and Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders apply with a=ra=r, and Rules of Arithmetic and Order in an Ordered Field applies to rr, with 0r0_{r}, 1r1_{r} in place of 00, 11. Negatives, differences, reciprocals and quotients in rr are as in Negatives, Differences, Reciprocals and Quotients §negative and Negatives, Differences, Reciprocals and Quotients §reciprocal. Rearrangements in rr that use only associativity, commutativity, distributivity, z+0r=zz+0_{r}=z, z⋅1r=zz\cdot1_{r}=z (all from Commutative Rings §ring), z+(−z)=0rz+(-z)=0_{r} (Negatives, Differences, Reciprocals and Quotients §negative) and z⋅z−1=1rz\cdot z^{-1}=1_{r} for z≠0rz\neq0_{r} (Negatives, Differences, Reciprocals and Quotients §reciprocal) are called ring rearrangements below.

Facts about κ\kappa. By The Canonical Embedding of the Rational Numbers into an Ordered Field §embedding, κ\kappa is the map of The Rational Numbers Embed in Exactly One Way into Every Ordered Field §unique, so for all u,v∈Qu,v\in\mathbb{Q}: u<vu<v if and only if κ(u)<κ(v)\kappa(u)<\kappa(v) by The Rational Numbers Embed in Exactly One Way into Every Ordered Field §order; κ(0)=0r\kappa(0)=0_{r} by The Rational Numbers Embed in Exactly One Way into Every Ordered Field §zero; κ(u−v)=κ(u)−κ(v)\kappa(u-v)=\kappa(u)-\kappa(v) by The Rational Numbers Embed in Exactly One Way into Every Ordered Field §negative; and, if v≠0v\neq0, κ(u/v)=κ(u)/κ(v)\kappa(u/v)=\kappa(u)/\kappa(v) with κ(v)≠0r\kappa(v)\neq0_{r} by The Rational Numbers Embed in Exactly One Way into Every Ordered Field §reciprocal. Here Q\mathbb{Q} is an ordered field by The Integers and the Rational Numbers, with the Natural Numbers and the Integers Identified with Subsets of the Rationals §rationals, so its order is a total order.

Natural numbers in rr. Let n∈Nn\in\mathbb{N}. Then n+1∈Nn+1\in\mathbb{N}, as 1∈N1\in\mathbb{N} by The Natural Numbers and the Natural Numbers with Zero: Arithmetic, Order, Induction and Recursion §sets and sums of natural numbers are natural numbers by The Natural Numbers and the Natural Numbers with Zero: Arithmetic, Order, Induction and Recursion §operations. As rr is a commutative ring by Fields §field, the image of 11 in rr is the unit 1r1_{r} of rr, by The Image of the Natural Numbers with Zero in a Commutative Ring Respects Zero, One, Sums, Products, Differences, Powers, and Finite Sums and Products §constants, and

(n+1)r=nr+1r,0r<nr,nr=κ(n),(n+1)_{r}=n_{r}+1_{r},\qquad0_{r}<n_{r},\qquad n_{r}=\kappa(n),

the first by The Image of the Natural Numbers with Zero in a Commutative Ring Respects Zero, One, Sums, Products, Differences, Powers, and Finite Sums and Products §sum and The Image of the Natural Numbers with Zero in a Commutative Ring Respects Zero, One, Sums, Products, Differences, Powers, and Finite Sums and Products §constants, the second by Inequalities in an Ordered Field: Mixed Transitivity, Strict Sums, Signs, Products, Natural Numbers, Halving, Reciprocals, Absolute Values and Squares §naturals, and the third, with nn read as a rational number, by Natural Numbers Read in an Ordered Field Agree with the Canonical Embedding of the Rationals §numerals. These three facts, together with the fact that the image 1r1_{r} of 11 in rr is the unit of rr, are the facts (N).

Clause infimum. Let ss be a nonempty subset of rr that is bounded below, and fix a lower bound b∈rb\in r of ss (Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order §bounded) and an element v0∈sv_{0}\in s. Let

t={z∈r:∃v (v∈s∧z=−v)},t=\{z\in r:\exists v\,(v\in s\wedge z=-v)\},

formed by class abstraction from a formula quantifying over sets only, in which −v-v is used only for v∈s⊆rv\in s\subseteq r, where it is defined; tt is a subset of rr by Sets and Maps: Ordinary Notation §set-builder, as are the sets formed in the same way below. It is nonempty, since −v0∈t-v_{0}\in t. It is bounded above by −b-b: if z∈tz\in t, say z=−vz=-v with v∈sv\in s, then b≤vb\le v, so −v≤−b-v\le-b by Rules of Arithmetic and Order in an Ordered Field §order-negative. By the hypothesis on rr, tt has a supremum mm, that is, a least element of Ub⁡(t)\operatorname{Ub}(t) (Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order §supremum). We show that −m-m is a greatest element of Lb⁡(s)\operatorname{Lb}(s), i.e. an infimum of ss by Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order §supremum.

First, −m∈Lb⁡(s)-m\in\operatorname{Lb}(s): for v∈sv\in s we have −v∈t-v\in t, so −v≤m-v\le m, hence −m≤−(−v)=v-m\le-(-v)=v by Rules of Arithmetic and Order in an Ordered Field §order-negative and Rules of Arithmetic and Order in an Ordered Field §signs. Second, let c∈Lb⁡(s)c\in\operatorname{Lb}(s). Then −c∈Ub⁡(t)-c\in\operatorname{Ub}(t): for z∈tz\in t, z=−vz=-v with v∈sv\in s, so c≤vc\le v and z=−v≤−cz=-v\le-c by Rules of Arithmetic and Order in an Ordered Field §order-negative. As mm is a least element of Ub⁡(t)\operatorname{Ub}(t), m≤−cm\le-c, hence c=−(−c)≤−mc=-(-c)\le-m by Rules of Arithmetic and Order in an Ordered Field §order-negative and Rules of Arithmetic and Order in an Ordered Field §signs. So −m-m is a greatest element of Lb⁡(s)\operatorname{Lb}(s), and ss has an infimum.

Clause archimedean. Suppose, for a contradiction, that x<nrx<n_{r} fails for every n∈Nn\in\mathbb{N}. Then nr≤xn_{r}\le x for every n∈Nn\in\mathbb{N} by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §total-negation. Let

K={z∈r:∃n (n∈N∧z=nr)},K=\{z\in r:\exists n\,(n\in\mathbb{N}\wedge z=n_{r})\},

formed from a formula quantifying over sets only, with nrn_{r} defined for every n∈Nn\in\mathbb{N} by The Image of a Natural Number in a Commutative Ring §image; KK is a subset of rr. It is nonempty, as 1r∈K1_{r}\in K (1∈N1\in\mathbb{N} by The Natural Numbers and the Natural Numbers with Zero: Arithmetic, Order, Induction and Recursion §sets), and xx is an upper bound of it. By the hypothesis on rr it has a supremum mm, that is, a least element of Ub⁡(K)\operatorname{Ub}(K).

From 0r<1r0_{r}<1_{r} (Rules of Arithmetic and Order in an Ordered Field §squares), adding m−1rm-1_{r} by Rules of Arithmetic and Order in an Ordered Field §order-sum and using ring rearrangements, m−1r<mm-1_{r}<m. Hence m≤m−1rm\le m-1_{r} fails by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §total-negation, so m−1r∉Ub⁡(K)m-1_{r}\notin\operatorname{Ub}(K), since otherwise m≤m−1rm\le m-1_{r} as mm is least in Ub⁡(K)\operatorname{Ub}(K). So there is z∈Kz\in K for which z≤m−1rz\le m-1_{r} fails; fix one and fix n∈Nn\in\mathbb{N} with z=nrz=n_{r}. Then m−1r<nrm-1_{r}<n_{r} by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §total-negation, and adding 1r1_{r} by Rules of Arithmetic and Order in an Ordered Field §order-sum, m<nr+1r=(n+1)rm<n_{r}+1_{r}=(n+1)_{r} by (N). But n+1∈Nn+1\in\mathbb{N} by (N), so (n+1)r∈K(n+1)_{r}\in K and (n+1)r≤m(n+1)_{r}\le m, which by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §total-negation contradicts m<(n+1)rm<(n+1)_{r}. Hence some n∈Nn\in\mathbb{N} has x<nrx<n_{r}, which is clause archimedean. Since x∈rx\in r was arbitrary, this holds for every element of rr; it is used below for elements other than xx.

Clause density. Let x<yx<y. The choices are made in the following order.

Step 1: the gap. Let d=y−xd=y-x. By Rules of Arithmetic and Order in an Ordered Field §order-sum, adding −x-x to x<yx<y gives 0r<d0_{r}<d. So d≠0rd\neq0_{r} (Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §strict-characterization), d−1d^{-1} is defined, and 0r<d−10_{r}<d^{-1} by Rules of Arithmetic and Order in an Ordered Field §positive-reciprocal.

Step 2: the scale. By clause archimedean applied to d−1d^{-1}, fix m∈Nm\in\mathbb{N} with d−1<mrd^{-1}<m_{r}, and put M=mrM=m_{r}; 0r<M0_{r}<M by (N). Multiplying by dd, with 0r<d0_{r}<d, by Rules of Arithmetic and Order in an Ordered Field §order-product, 1r=d−1⋅d<M⋅d1_{r}=d^{-1}\cdot d<M\cdot d. By ring rearrangements and Rules of Arithmetic and Order in an Ordered Field §signs, M⋅d=M⋅y−M⋅xM\cdot d=M\cdot y-M\cdot x, and adding M⋅xM\cdot x by Rules of Arithmetic and Order in an Ordered Field §order-sum,

M⋅x+1r<M⋅y.M\cdot x+1_{r}<M\cdot y.

Step 3: the shift. By clause archimedean applied to −(M⋅x)-(M\cdot x), fix p∈Np\in\mathbb{N} with −(M⋅x)<pr-(M\cdot x)<p_{r}, and put w=M⋅x+prw=M\cdot x+p_{r}. Adding M⋅xM\cdot x by Rules of Arithmetic and Order in an Ordered Field §order-sum, 0r<w0_{r}<w.

Step 4: the least natural number above ww. Let A={k∈N:w<kr}A=\{k\in\mathbb{N}:w<k_{r}\}, formed from a formula quantifying over sets only; it is a subset of N\mathbb{N}, and nonempty by clause archimedean applied to ww. By Arithmetic and Order of the Natural Numbers §well-order, fix k∈Ak\in A with k≤vk\le v for every v∈Av\in A. Then w<krw<k_{r}, and we claim kr≤w+1rk_{r}\le w+1_{r}. If k=1k=1, then kr=1r=0r+1r≤w+1rk_{r}=1_{r}=0_{r}+1_{r}\le w+1_{r} by (N), 0r≤w0_{r}\le w (from 0r<w0_{r}<w by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §weak-strict) and Rules of Arithmetic and Order in an Ordered Field §order-sum (with 1r≤1r1_{r}\le1_{r}, by reflexivity of the partial order ≤\le, for the second summand). If k≠1k\neq1, fix e∈Ne\in\mathbb{N} with k=e+1k=e+1 by Arithmetic and Order of the Natural Numbers §predecessor. Then e<ke<k by Arithmetic and Order of the Natural Numbers §successor, so k≤ek\le e fails by Arithmetic and Order of the Natural Numbers §trichotomy and Arithmetic and Order of the Natural Numbers §partial-order; hence e∉Ae\notin A, i.e. w<erw<e_{r} fails, so er≤we_{r}\le w by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §total-negation, and kr=er+1r≤w+1rk_{r}=e_{r}+1_{r}\le w+1_{r} by (N) and Rules of Arithmetic and Order in an Ordered Field §order-sum. In both cases

w<kr≤w+1r.w<k_{r}\le w+1_{r}.

Step 5: the rational number. By (N) and the facts about κ\kappa, κ(0)=0r<M=mr=κ(m)\kappa(0)=0_{r}<M=m_{r}=\kappa(m), so 0<m0<m in Q\mathbb{Q} by The Rational Numbers Embed in Exactly One Way into Every Ordered Field §order, and m≠0m\neq0 by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §strict-characterization. Let u=(k−p)/m∈Qu=(k-p)/m\in\mathbb{Q}, with kk, pp and mm read as rational numbers (The Integers and the Rational Numbers, with the Natural Numbers and the Integers Identified with Subsets of the Rationals §identification, The Integers and the Rational Numbers, with the Natural Numbers and the Integers Identified with Subsets of the Rationals §agreement) and the difference and quotient formed in Q\mathbb{Q}. By the facts about κ\kappa and (N), κ(u)=(κ(k)−κ(p))/κ(m)=(kr−pr)/M\kappa(u)=(\kappa(k)-\kappa(p))/\kappa(m)=(k_{r}-p_{r})/M; as M≠0rM\neq0_{r} (from 0r<M0_{r}<M by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §strict-characterization), ring rearrangements give κ(u)⋅M=kr−pr\kappa(u)\cdot M=k_{r}-p_{r}. Adding −pr-p_{r} to the display of Step 4 by Rules of Arithmetic and Order in an Ordered Field §order-sum, and using w−pr=M⋅xw-p_{r}=M\cdot x (ring rearrangements), we get

M⋅x<κ(u)⋅M≤M⋅x+1r.M\cdot x<\kappa(u)\cdot M\le M\cdot x+1_{r}.

With Step 2 and Inequalities in an Ordered Field: Mixed Transitivity, Strict Sums, Signs, Products, Natural Numbers, Halving, Reciprocals, Absolute Values and Squares §mixed, κ(u)⋅M<M⋅y\kappa(u)\cdot M<M\cdot y. Since 0r<M0_{r}<M, Rules of Arithmetic and Order in an Ordered Field §order-product and commutativity turn x⋅M<κ(u)⋅Mx\cdot M<\kappa(u)\cdot M and κ(u)⋅M<y⋅M\kappa(u)\cdot M<y\cdot M into x<κ(u)x<\kappa(u) and κ(u)<y\kappa(u)<y. So u∈Qu\in\mathbb{Q} satisfies x<κ(u)<yx<\kappa(u)<y, which is clause density.

Clause rational-supremum. Let S={u∈Q:κ(u)<x}S=\{u\in\mathbb{Q}:\kappa(u)<x\}, a subset of Q\mathbb{Q} by Sets and Maps: Ordinary Notation §set-builder, and let L=κ(S)L=\kappa(S), the set of the statement. Here κ(S)\kappa(S) denotes the image of SS and not a value of κ\kappa, because an element of Q\mathbb{Q} is used as a number only, never as a set (The Integers and the Rational Numbers, with the Natural Numbers and the Integers Identified with Subsets of the Rationals §numbers-only), so SS is not an element of Q\mathbb{Q} in this reading; by Sets and Maps: Ordinary Notation §images it is the image κ[S]\kappa[S] of The Image and the Preimage of a Class under a Class §image, a set. Its elements are exactly the κ(u)\kappa(u) with u∈Qu\in\mathbb{Q} and κ(u)<x\kappa(u)<x: by The Image and the Preimage of a Class under a Class §image, z∈Lz\in L if and only if (u,z)∈κ(u,z)\in\kappa for some u∈Su\in S, and for u∈S⊆Q=dom⁡κu\in S\subseteq\mathbb{Q}=\operatorname{dom}\kappa (Functions, Values of a Function, and Functions from One Class to Another §map), (u,z)∈κ(u,z)\in\kappa holds if and only if z=κ(u)z=\kappa(u), by Functions, Values of a Function, and Functions from One Class to Another §value. In particular LL is a subset of rr, as κ\kappa maps into rr. We show that xx is a least element of Ub⁡(L)\operatorname{Ub}(L), i.e. a supremum of LL by Bounds, Least and Greatest Elements, Suprema and Infima for a Partial Order §supremum; as LL has at most one supremum by the same clause, xx is then the supremum of LL, which is clause rational-supremum.

x∈Ub⁡(L)x\in\operatorname{Ub}(L): every z∈Lz\in L has z<xz<x, hence z≤xz\le x by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §weak-strict.

xx is least in Ub⁡(L)\operatorname{Ub}(L): let b∈Ub⁡(L)b\in\operatorname{Ub}(L) and suppose x≤bx\le b fails. Then b<xb<x by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §total-negation, and by clause density fix u∈Qu\in\mathbb{Q} with b<κ(u)<xb<\kappa(u)<x. Then κ(u)∈L\kappa(u)\in L, so κ(u)≤b\kappa(u)\le b, contradicting b<κ(u)b<\kappa(u) by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §total-negation. Hence x≤bx\le b, so xx is a least element of Ub⁡(L)\operatorname{Ub}(L), and hence the supremum of LL, as explained above.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…