TheoremBase

The field laws are first proved for the product of nonnegative cuts, together with a reciprocal for positive cuts built from the rationals below reciprocals of non-elements; they are then transferred to all cuts by showing that the sign-case product commutes with negation, and the rational clause follows from the nonnegative case and the identity that the negative of the cut of u is the cut of minus u.

Proof

Throughout, << is the strict relation of the order ≤\le of Q\mathbb{Q}. By The Real Numbers §operations and The Real Numbers §constants, x+y=x⊕yx+y=x\oplus y, 0R=0∗0_{\mathbb{R}}=0^{*} and 1R=1∗1_{\mathbb{R}}=1^{*}; x≤Cyx\le_{C}y means x⊆yx\subseteq y (Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts); and ⊕\oplus, ⊖\ominus and ⊙\odot are as in Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts. Two classes with the same elements are equal by Axiom of Extensionality for Classes; every equality of cuts below that is not a computation is proved by comparing elements. Let P={x∈C:0∗⊆x}P=\{x\in C:0^{*}\subseteq x\}, the class of nonnegative cuts.

Step 0 (facts about Q\mathbb{Q}). By The Rational Numbers Form an Archimedean Ordered Field Containing the Integers §ordered-field, Q\mathbb{Q} is an ordered field and ≤\le is a total order on the set Q\mathbb{Q}; the field laws (associativity, commutativity, distributivity, 1⋅c=c1\cdot c=c, and c⋅c−1=1c\cdot c^{-1}=1 for c≠0c\neq0 by Negatives, Differences, Reciprocals and Quotients §reciprocal) are used without further mention. Let a,b,c,d∈Qa,b,c,d\in\mathbb{Q}.

(Q1) If a≤ba\le b and b<cb<c, or a<ba<b and b≤cb\le c, then a<ca<c, by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §weak-strict and Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §strict-transitive. Moreover a≤ba\le b fails if and only if b<ab<a (Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §total-negation), and a<aa<a fails (Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §strict-irreflexive).

(Q2) If a≤ba\le b and 0≤c0\le c, then a⋅c≤b⋅ca\cdot c\le b\cdot c; if a<ba<b and 0<c0<c, then a⋅c<b⋅ca\cdot c<b\cdot c. Indeed, if c=0c=0 both sides of the first claim are 00 by Rules of Arithmetic and Order in an Ordered Field §zero; otherwise 0<c0<c by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §weak-strict, and both claims are Rules of Arithmetic and Order in an Ordered Field §order-product. Taking a=0a=0: if 0≤b0\le b and 0≤c0\le c, then 0≤b⋅c0\le b\cdot c.

(Q3) We use Rules of Arithmetic and Order in an Ordered Field §order-sum (a<ba<b if and only if a+c<b+ca+c<b+c; a≤ba\le b and c≤dc\le d give a+c≤b+da+c\le b+d), the sign rules Rules of Arithmetic and Order in an Ordered Field §signs, 0<10<1 from Rules of Arithmetic and Order in an Ordered Field §squares, 0<c−10<c^{-1} whenever 0<c0<c from Rules of Arithmetic and Order in an Ordered Field §positive-reciprocal, and midpoints from Rules of Arithmetic and Order in an Ordered Field §midpoint. In particular, if 0<d0<d then −d<0-d<0 (add −d-d to both sides), and if a<1a<1 then 0<1−a0<1-a (add −a-a).

Step 1 (facts about cuts). Let x∈Cx\in C. By the definition of CC in Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts, x⊆Qx\subseteq\mathbb{Q}, x≠∅x\neq\emptyset, x≠Qx\neq\mathbb{Q}, every v∈Qv\in\mathbb{Q} with v<uv<u for some u∈xu\in x lies in xx (downward closure), and every u∈xu\in x has some v∈xv\in x with u<vu<v (openness).

(C1) If c∈xc\in x, a∈Qa\in\mathbb{Q} and a≤ca\le c, then a∈xa\in x: by Uniqueness of Least and Greatest Elements, Properties of the Strict Order, and Trichotomy for Total Orders §weak-strict, a<ca<c or a=ca=c.

(C2) If c∈xc\in x and a∈Q∖xa\in\mathbb{Q}\setminus x, then c<ac<a: otherwise a≤ca\le c by (Q1), and a∈xa\in x by (C1).

(C3) If 0∈x0\in x, then x∈Px\in P and x≠0∗x\neq0^{*}: every v<0v<0 lies in xx by downward closure, and 0∉0∗0\notin0^{*} by (Q1). Conversely, if x∈Px\in P and x≠0∗x\neq0^{*}, then 0∈x0\in x, there is a∈xa\in x with 0<a0<a, and 0<q0<q for every q∈Q∖xq\in\mathbb{Q}\setminus x. Indeed, since 0∗⊆x0^{*}\subseteq x and x≠0∗x\neq0^{*}, some c∈xc\in x is not in 0∗0^{*}, so 0≤c0\le c by (Q1) and 0∈x0\in x by (C1); aa exists by openness; and 0<q0<q by (C2).

(C4) For every d∈Qd\in\mathbb{Q} with 0<d0<d there are c∈xc\in x and q∈Q∖xq\in\mathbb{Q}\setminus x with q<c+dq<c+d. Indeed, ⊖x∈C\ominus x\in C by Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §negative and x⊕⊖x=0∗x\oplus\ominus x=0^{*} by Addition and Order of Dedekind Cuts: an Ordered Abelian Group in Which Every Nonempty Set Bounded Above Has a Supremum §group; as −d<0-d<0 (Q3), −d∈x⊕⊖x-d\in x\oplus\ominus x, so −d=c+v-d=c+v with c∈xc\in x and v∈⊖xv\in\ominus x. By the definition of ⊖x\ominus x in Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts there is r∈Qr\in\mathbb{Q} with 0<r0<r and q:=−v−r∉xq:=-v-r\notin x. Then q+r=−v=c+dq+r=-v=c+d, and q=q+0<q+rq=q+0<q+r by (Q3), so q<c+dq<c+d.

(C5) For x,y∈Cx,y\in C and w∈Qw\in\mathbb{Q}: w∈x⊙yw\in x\odot y if and only if w<0w<0 or w=u⋅vw=u\cdot v for some u∈xu\in x and v∈yv\in y with 0≤u0\le u and 0≤v0\le v (definition of ⊙\odot in Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts). Hence x⊙y=y⊙xx\odot y=y\odot x; x⊙0∗=0∗x\odot0^{*}=0^{*}, since no u∈0∗u\in0^{*} has 0≤u0\le u (Q1); and if y⊆y′y\subseteq y' with y′∈Cy'\in C, then x⊙y⊆x⊙y′x\odot y\subseteq x\odot y'. If x,y∈Px,y\in P, then x⊙y∈Px\odot y\in P by Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §product, so (C1) applies to x⊙yx\odot y. Finally 1∗∈P1^{*}\in P: 1∗∈C1^{*}\in C by Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §rational, and every v<0v<0 has v<1v<1 by 0<10<1 and (Q1); likewise 0∗∈P0^{*}\in P.

Step 2 (group facts). For a,b∈Ca,b\in C, a+b∈Ca+b\in C and ⊖a∈C\ominus a\in C by Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §sum and Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §negative, and by Addition and Order of Dedekind Cuts: an Ordered Abelian Group in Which Every Nonempty Set Bounded Above Has a Supremum §group the operation ++ is associative and commutative with a+0∗=aa+0^{*}=a and a+⊖a=0∗a+\ominus a=0^{*}. Let a,b,c∈Ca,b,c\in C.

(G1) If a+b=0∗a+b=0^{*} and a+c=0∗a+c=0^{*}, then b=cb=c: b=b+(a+c)=(b+a)+c=(a+b)+c=0∗+c=cb=b+(a+c)=(b+a)+c=(a+b)+c=0^{*}+c=c.

(G2) ⊖⊖a=a\ominus\ominus a=a, ⊖0∗=0∗\ominus0^{*}=0^{*}, ⊖(a+b)=⊖a+⊖b\ominus(a+b)=\ominus a+\ominus b, and (a+b)+⊖b=a=(a+⊖b)+b(a+b)+\ominus b=a=(a+\ominus b)+b. Indeed ⊖a+a=0∗=⊖a+⊖⊖a\ominus a+a=0^{*}=\ominus a+\ominus\ominus a; 0∗+0∗=0∗=0∗+⊖0∗0^{*}+0^{*}=0^{*}=0^{*}+\ominus0^{*}; and (a+b)+(⊖a+⊖b)=(a+⊖a)+(b+⊖b)=0∗=(a+b)+⊖(a+b)(a+b)+(\ominus a+\ominus b)=(a+\ominus a)+(b+\ominus b)=0^{*}=(a+b)+\ominus(a+b) by associativity and commutativity; apply (G1) in each case. The last two identities follow from associativity, b+⊖b=⊖b+b=0∗b+\ominus b=\ominus b+b=0^{*} and a+0∗=aa+0^{*}=a.

(G3) If b∈Pb\in P, then a⊆a+ba\subseteq a+b. Indeed 0∗≤Cb0^{*}\le_{C}b gives 0∗+a≤Cb+a0^{*}+a\le_{C}b+a by Addition and Order of Dedekind Cuts: an Ordered Abelian Group in Which Every Nonempty Set Bounded Above Has a Supremum §order, where 0∗+a=a0^{*}+a=a and b+a=a+bb+a=a+b. Hence if a,b∈Pa,b\in P, then 0∗⊆a⊆a+b0^{*}\subseteq a\subseteq a+b, so a+b∈Pa+b\in P, and also b⊆a+bb\subseteq a+b.

(G4) If a∈Pa\in P and ⊖a∈P\ominus a\in P, then a=0∗a=0^{*}: by (G3), a⊆a+⊖a=0∗⊆aa\subseteq a+\ominus a=0^{*}\subseteq a. Consequently, by Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §absolute, if a≠0∗a\neq0^{*} then exactly one of a∈Pa\in P and ⊖a∈P\ominus a\in P holds.

Step 3 (the product ⊙\odot on PP). Let x,y,z∈Px,y,z\in P. By (C5), x⊙y=y⊙x∈Px\odot y=y\odot x\in P and x⊙0∗=0∗x\odot0^{*}=0^{*}. Each class compared below is an element of PP or of the form s⊙ts\odot t, so contains 0∗0^{*}; hence for each inclusion only elements ww with 0≤w0\le w need checking (Q1), and for such ww, w∈s⊙tw\in s\odot t means w=u⋅vw=u\cdot v with u∈su\in s, v∈tv\in t, 0≤u0\le u and 0≤v0\le v.

(3a) (x⊙y)⊙z=x⊙(y⊙z)(x\odot y)\odot z=x\odot(y\odot z). Let 0≤w∈(x⊙y)⊙z0\le w\in(x\odot y)\odot z: w=e⋅cw=e\cdot c with e∈x⊙ye\in x\odot y, c∈zc\in z, 0≤e0\le e, 0≤c0\le c. As 0≤e0\le e, e=u⋅ve=u\cdot v with u∈xu\in x, v∈yv\in y, 0≤u0\le u, 0≤v0\le v. Then w=u⋅(v⋅c)w=u\cdot(v\cdot c) with v⋅c∈y⊙zv\cdot c\in y\odot z and 0≤v⋅c0\le v\cdot c (Q2), so w∈x⊙(y⊙z)w\in x\odot(y\odot z). Conversely, if 0≤w=u⋅e0\le w=u\cdot e with u∈xu\in x, e∈y⊙ze\in y\odot z, 0≤u0\le u, 0≤e0\le e, then e=v⋅ce=v\cdot c with v∈yv\in y, c∈zc\in z, both nonnegative, and w=(u⋅v)⋅cw=(u\cdot v)\cdot c with 0≤u⋅v∈x⊙y0\le u\cdot v\in x\odot y, so w∈(x⊙y)⊙zw\in(x\odot y)\odot z.

(3b) x⊙1∗=xx\odot1^{*}=x. If 0≤w∈x⊙1∗0\le w\in x\odot1^{*}, then w=u⋅vw=u\cdot v with u∈xu\in x, 0≤u0\le u and 0≤v<10\le v<1, so w≤1⋅u=uw\le1\cdot u=u by (Q2) and w∈xw\in x by (C1). If 0≤w∈x0\le w\in x, choose u∈xu\in x with w<uw<u (openness); then 0<u0<u by (Q1). Let v=w⋅u−1v=w\cdot u^{-1}. Then 0≤v0\le v by (Q2) and (Q3), v<u⋅u−1=1v<u\cdot u^{-1}=1 by (Q2), and u⋅v=wu\cdot v=w; so v∈1∗v\in1^{*} and w∈x⊙1∗w\in x\odot1^{*}.

(3c) x⊙(y+z)=x⊙y+x⊙zx\odot(y+z)=x\odot y+x\odot z. Here y+z∈Py+z\in P by (G3), and both sides lie in PP by (C5) and (G3). If y=0∗y=0^{*}, then y+z=zy+z=z and x⊙y+x⊙z=0∗+x⊙z=x⊙zx\odot y+x\odot z=0^{*}+x\odot z=x\odot z, so both sides are x⊙zx\odot z; the case z=0∗z=0^{*} is symmetric. So let y≠0∗≠zy\neq0^{*}\neq z; then 0∈y0\in y and 0∈z0\in z by (C3).

For ⊆\subseteq, let 0≤w∈x⊙(y+z)0\le w\in x\odot(y+z): w=u⋅sw=u\cdot s with u∈xu\in x, s∈y+zs\in y+z, 0≤u0\le u, 0≤s0\le s, and s=e+fs=e+f with e∈ye\in y, f∈zf\in z. We may assume 0≤e0\le e and 0≤f0\le f: if e<0e<0, then s=e+f<0+f=fs=e+f<0+f=f by (Q3), so s∈zs\in z by (C1), and we replace (e,f)(e,f) by (0,s)(0,s), using 0∈y0\in y; if f<0f<0, symmetrically replace (e,f)(e,f) by (s,0)(s,0). Then w=u⋅e+u⋅fw=u\cdot e+u\cdot f with u⋅e∈x⊙yu\cdot e\in x\odot y and u⋅f∈x⊙zu\cdot f\in x\odot z, so w∈x⊙y+x⊙zw\in x\odot y+x\odot z.

For ⊇\supseteq, let 0≤w∈x⊙y+x⊙z0\le w\in x\odot y+x\odot z: w=p+qw=p+q with p∈x⊙yp\in x\odot y and q∈x⊙zq\in x\odot z. If p<0p<0, then w<qw<q by (Q3); as z⊆y+zz\subseteq y+z by (G3), (C5) gives q∈x⊙z⊆x⊙(y+z)q\in x\odot z\subseteq x\odot(y+z), and w∈x⊙(y+z)w\in x\odot(y+z) by (C1). If q<0q<0, the same argument applies with y⊆y+zy\subseteq y+z. Otherwise 0≤p0\le p and 0≤q0\le q, so p=u1⋅ep=u_{1}\cdot e and q=u2⋅fq=u_{2}\cdot f with u1,u2∈xu_{1},u_{2}\in x, e∈ye\in y, f∈zf\in z, all nonnegative. Let uu be the larger of u1u_{1} and u2u_{2} for the total order ≤\le; then u∈xu\in x, and p≤u⋅ep\le u\cdot e, q≤u⋅fq\le u\cdot f by (Q2), so w≤u⋅e+u⋅f=u⋅(e+f)w\le u\cdot e+u\cdot f=u\cdot(e+f) by (Q3). Since e+f∈y+ze+f\in y+z and 0≤e+f0\le e+f (Q3), u⋅(e+f)∈x⊙(y+z)u\cdot(e+f)\in x\odot(y+z), and w∈x⊙(y+z)w\in x\odot(y+z) by (C1).

Step 4 (reciprocals in PP). Let x∈Px\in P with x≠0∗x\neq0^{*}. By (C3), 0∈x0\in x, there is a∈xa\in x with 0<a0<a, and 0<q0<q for every q∈Q∖xq\in\mathbb{Q}\setminus x; since x≠Qx\neq\mathbb{Q}, fix q0∈Q∖xq_{0}\in\mathbb{Q}\setminus x. Let

w={p∈Q:∃q (q∈Q∧q∉x∧p⋅q<1)}.w=\{p\in\mathbb{Q}:\exists q\,(q\in\mathbb{Q}\wedge q\notin x\wedge p\cdot q<1)\}.

As a subclass of the set Q\mathbb{Q}, ww is a set by Subclasses of Sets Are Sets, the Union and Power Set of a Set Exist Uniquely, Binary Unions of Sets Are Sets, and the Universal Class Is Proper §subclass, hence an element of P(Q)\mathcal{P}(\mathbb{Q}) by The Union Set and the Power Set of a Set §power.

(4a) w∈Pw\in P and w≠0∗w\neq0^{*}. If p≤0p\le0, then p⋅q0≤0⋅q0=0<1p\cdot q_{0}\le0\cdot q_{0}=0<1 by (Q2) and (Q1), so p∈wp\in w; in particular 0∈w0\in w and w≠∅w\neq\emptyset. Also a−1∉wa^{-1}\notin w, so w≠Qw\neq\mathbb{Q}: if a−1⋅q<1a^{-1}\cdot q<1 with q∉xq\notin x, then q=(a−1⋅q)⋅a<1⋅a=aq=(a^{-1}\cdot q)\cdot a<1\cdot a=a by (Q2), so q∈xq\in x by downward closure, a contradiction. Downward closure: if p∈wp\in w with witness qq and p′<pp'<p, then p′⋅q<p⋅q<1p'\cdot q<p\cdot q<1 by (Q2) and 0<q0<q, so p′∈wp'\in w. Openness: if p∈wp\in w with witness qq, then p=(p⋅q)⋅q−1<1⋅q−1=q−1p=(p\cdot q)\cdot q^{-1}<1\cdot q^{-1}=q^{-1} by (Q2) and (Q3); take p′p' with p<p′<q−1p<p'<q^{-1} (midpoint, (Q3)); then p′⋅q<q−1⋅q=1p'\cdot q<q^{-1}\cdot q=1 by (Q2), so p′∈wp'\in w. Thus w∈Cw\in C, and w∈Pw\in P, w≠0∗w\neq0^{*} by (C3) since 0∈w0\in w.

(4b) x⊙w⊆1∗x\odot w\subseteq1^{*}. Elements t<0t<0 lie in 1∗1^{*} by (Q1). If 0≤t∈x⊙w0\le t\in x\odot w, then t=u⋅pt=u\cdot p with u∈xu\in x, p∈wp\in w, 0≤u0\le u, 0≤p0\le p, and p⋅q<1p\cdot q<1 for some q∈Q∖xq\in\mathbb{Q}\setminus x. By (C2), u<qu<q, so t=u⋅p≤q⋅p<1t=u\cdot p\le q\cdot p<1 by (Q2) and (Q1), i.e. t∈1∗t\in1^{*}.

(4c) 1∗⊆x⊙w1^{*}\subseteq x\odot w. Let t∈1∗t\in1^{*}, so t<1t<1; if t<0t<0, then t∈x⊙wt\in x\odot w. Let 0≤t0\le t. Then 0<1−t0<1-t (Q3), and d:=a⋅(1−t)d:=a\cdot(1-t) satisfies 0<d0<d by (Q2). By (C4) choose c∈xc\in x and q∈Q∖xq\in\mathbb{Q}\setminus x with q<c+dq<c+d, and let uu be the larger of cc and aa. Then u∈xu\in x, 0<a≤u0<a\le u (so 0<u0<u by (Q1)), and q<c+d≤u+dq<c+d\le u+d by (Q3) and (Q1). We claim t⋅q<ut\cdot q<u. If t=0t=0, then t⋅q=0<ut\cdot q=0<u. If 0<t0<t, then t⋅q<t⋅(u+d)=t⋅u+t⋅dt\cdot q<t\cdot(u+d)=t\cdot u+t\cdot d by (Q2); moreover t⋅a≤1⋅a=a≤ut\cdot a\le1\cdot a=a\le u by (Q2), so t⋅d=(t⋅a)⋅(1−t)≤u⋅(1−t)t\cdot d=(t\cdot a)\cdot(1-t)\le u\cdot(1-t) by (Q2), and therefore t⋅q<t⋅u+u⋅(1−t)=ut\cdot q<t\cdot u+u\cdot(1-t)=u by (Q3) and (Q1). Now let p=t⋅u−1p=t\cdot u^{-1}. Then 0≤p0\le p by (Q2) and (Q3), u⋅p=tu\cdot p=t, and p⋅q=(t⋅q)⋅u−1<u⋅u−1=1p\cdot q=(t\cdot q)\cdot u^{-1}<u\cdot u^{-1}=1 by (Q2), so p∈wp\in w with witness qq. Hence t=u⋅p∈x⊙wt=u\cdot p\in x\odot w.

By (4b) and (4c), x⊙w=1∗x\odot w=1^{*}.

Step 5 (signs). For x∈Cx\in C, ∣x∣C∈P|x|_{C}\in P by The Real Numbers §operations. Write id\mathrm{id} for the identity map of CC, and let ε,η,ζ\varepsilon,\eta,\zeta range over {id,⊖}\{\mathrm{id},\ominus\}, unary operations on CC, given by Maps and Relations Given by Formulas §map since ⊖x∈C\ominus x\in C for every x∈Cx\in C by Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §negative. Since ⊖⊖a=a\ominus\ominus a=a (G2), the composite εη\varepsilon\eta of two of them again lies in {id,⊖}\{\mathrm{id},\ominus\}, and composition is associative. Let x,y∈Cx,y\in C.

(5a) Let εx=id\varepsilon_{x}=\mathrm{id} if x∈Px\in P and εx=⊖\varepsilon_{x}=\ominus otherwise. Then x=εx(∣x∣C)x=\varepsilon_{x}(|x|_{C}): if x∉Px\notin P, then ∣x∣C=⊖x|x|_{C}=\ominus x and ⊖∣x∣C=⊖⊖x=x\ominus|x|_{C}=\ominus\ominus x=x by (G2).

(5b) ∣⊖x∣C=∣x∣C|\ominus x|_{C}=|x|_{C}, and if x≠0∗x\neq0^{*}, then ⊖x∈P\ominus x\in P if and only if x∉Px\notin P. For x=0∗x=0^{*} we have ⊖x=0∗\ominus x=0^{*} by (G2). For x≠0∗x\neq0^{*}, exactly one of xx and ⊖x\ominus x lies in PP by (G4); if x∈Px\in P, then ∣⊖x∣C=⊖⊖x=x=∣x∣C|\ominus x|_{C}=\ominus\ominus x=x=|x|_{C}, and if x∉Px\notin P, then ∣⊖x∣C=⊖x=∣x∣C|\ominus x|_{C}=\ominus x=|x|_{C}.

(5c) x⋅y=y⋅xx\cdot y=y\cdot x: the case distinction in The Real Numbers §operations is symmetric in xx and yy, and ∣x∣C⊙∣y∣C=∣y∣C⊙∣x∣C|x|_{C}\odot|y|_{C}=|y|_{C}\odot|x|_{C} by (C5).

(5d) If x=0∗x=0^{*} or y=0∗y=0^{*}, then x⋅y=0∗x\cdot y=0^{*}: since 0∗∈P0^{*}\in P, ∣0∗∣C=0∗|0^{*}|_{C}=0^{*}, so ∣x∣C⊙∣y∣C=0∗|x|_{C}\odot|y|_{C}=0^{*} by (C5), and ⊖0∗=0∗\ominus0^{*}=0^{*} by (G2).

(5e) If x,y∈Px,y\in P, then x⋅y=x⊙y∈Px\cdot y=x\odot y\in P, by The Real Numbers §operations and (C5).

(5f) (⊖x)⋅y=⊖(x⋅y)(\ominus x)\cdot y=\ominus(x\cdot y). If x=0∗x=0^{*}, both sides are 0∗0^{*} by (5d) and (G2). If x≠0∗x\neq0^{*}, then by (5b) the case of The Real Numbers §operations that applies to the pair (⊖x,y)(\ominus x,y) is the other one than for (x,y)(x,y), while ∣⊖x∣C⊙∣y∣C=∣x∣C⊙∣y∣C=:m|\ominus x|_{C}\odot|y|_{C}=|x|_{C}\odot|y|_{C}=:m. So one of x⋅yx\cdot y and (⊖x)⋅y(\ominus x)\cdot y is mm and the other is ⊖m\ominus m, and since ⊖⊖m=m\ominus\ominus m=m (G2), (⊖x)⋅y=⊖(x⋅y)(\ominus x)\cdot y=\ominus(x\cdot y) in either case. By (5c), x⋅(⊖y)=(⊖y)⋅x=⊖(y⋅x)=⊖(x⋅y)x\cdot(\ominus y)=(\ominus y)\cdot x=\ominus(y\cdot x)=\ominus(x\cdot y), and then (⊖x)⋅(⊖y)=⊖⊖(x⋅y)=x⋅y(\ominus x)\cdot(\ominus y)=\ominus\ominus(x\cdot y)=x\cdot y. Hence (εx)⋅(ηy)=εη(x⋅y)(\varepsilon x)\cdot(\eta y)=\varepsilon\eta(x\cdot y) for all ε,η\varepsilon,\eta.

Step 6 (lem:dedekind-cut-multiplication-nbg-2026a#ring). Let x,y,z∈Cx,y,z\in C, xˉ=∣x∣C\bar x=|x|_{C}, yˉ=∣y∣C\bar y=|y|_{C}, zˉ=∣z∣C\bar z=|z|_{C} (elements of PP), and ε=εx\varepsilon=\varepsilon_{x}, η=εy\eta=\varepsilon_{y}, ζ=εz\zeta=\varepsilon_{z}, so that x=εxˉx=\varepsilon\bar x, y=ηyˉy=\eta\bar y, z=ζzˉz=\zeta\bar z by (5a).

Commutativity is (5c). Unit: 1R=1∗∈P1_{\mathbb{R}}=1^{*}\in P by (C5), so by (5f), (5e) and (3b), x⋅1∗=ε(xˉ⋅1∗)=ε(xˉ⊙1∗)=εxˉ=xx\cdot1^{*}=\varepsilon(\bar x\cdot1^{*})=\varepsilon(\bar x\odot1^{*})=\varepsilon\bar x=x.

Associativity: xˉ⊙yˉ\bar x\odot\bar y and yˉ⊙zˉ\bar y\odot\bar z lie in PP by (C5), so by (5f) and (5e),

(x⋅y)⋅z=(εη(xˉ⊙yˉ))⋅(ζzˉ)=εηζ((xˉ⊙yˉ)⊙zˉ),x⋅(y⋅z)=(εxˉ)⋅(ηζ(yˉ⊙zˉ))=εηζ(xˉ⊙(yˉ⊙zˉ)),(x\cdot y)\cdot z=(\varepsilon\eta(\bar x\odot\bar y))\cdot(\zeta\bar z)=\varepsilon\eta\zeta((\bar x\odot\bar y)\odot\bar z),\qquad x\cdot(y\cdot z)=(\varepsilon\bar x)\cdot(\eta\zeta(\bar y\odot\bar z))=\varepsilon\eta\zeta(\bar x\odot(\bar y\odot\bar z)),

and these agree by (3a).

Distributivity: by (5f), x⋅(y+z)=ε(xˉ⋅(y+z))x\cdot(y+z)=\varepsilon(\bar x\cdot(y+z)) and x⋅y+x⋅z=ε(xˉ⋅y)+ε(xˉ⋅z)x\cdot y+x\cdot z=\varepsilon(\bar x\cdot y)+\varepsilon(\bar x\cdot z); when ε=⊖\varepsilon=\ominus, ⊖(xˉ⋅y)+⊖(xˉ⋅z)=⊖(xˉ⋅y+xˉ⋅z)\ominus(\bar x\cdot y)+\ominus(\bar x\cdot z)=\ominus(\bar x\cdot y+\bar x\cdot z) by (G2). So it suffices to prove, for all y,z∈Cy,z\in C, the identity D(y,z)D(y,z): xˉ⋅(y+z)=xˉ⋅y+xˉ⋅z\bar x\cdot(y+z)=\bar x\cdot y+\bar x\cdot z. By commutativity of ++, D(y,z)D(y,z) holds if and only if D(z,y)D(z,y) does.

(6i) D(y,z)D(y,z) implies D(⊖y,⊖z)D(\ominus y,\ominus z): by (G2), (5f), D(y,z)D(y,z), (G2) and (5f) in turn, xˉ⋅(⊖y+⊖z)=xˉ⋅⊖(y+z)=⊖(xˉ⋅(y+z))=⊖(xˉ⋅y+xˉ⋅z)=⊖(xˉ⋅y)+⊖(xˉ⋅z)=xˉ⋅(⊖y)+xˉ⋅(⊖z)\bar x\cdot(\ominus y+\ominus z)=\bar x\cdot\ominus(y+z)=\ominus(\bar x\cdot(y+z))=\ominus(\bar x\cdot y+\bar x\cdot z)=\ominus(\bar x\cdot y)+\ominus(\bar x\cdot z)=\bar x\cdot(\ominus y)+\bar x\cdot(\ominus z).

(6ii) If y,z∈Py,z\in P, then D(y,z)D(y,z): y+z∈Py+z\in P by (G3), so by (5e) and (3c), xˉ⋅(y+z)=xˉ⊙(y+z)=xˉ⊙y+xˉ⊙z=xˉ⋅y+xˉ⋅z\bar x\cdot(y+z)=\bar x\odot(y+z)=\bar x\odot y+\bar x\odot z=\bar x\cdot y+\bar x\cdot z.

(6iii) If y∈Py\in P, z∉Pz\notin P and y+z∈Py+z\in P, then D(y,z)D(y,z). Indeed ⊖z∈P\ominus z\in P by Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §absolute, and y=(y+z)+⊖zy=(y+z)+\ominus z by (G2). By (6ii) for the pair (y+z,⊖z)(y+z,\ominus z) and by (5f), xˉ⋅y=xˉ⋅(y+z)+xˉ⋅(⊖z)=xˉ⋅(y+z)+⊖(xˉ⋅z)\bar x\cdot y=\bar x\cdot(y+z)+\bar x\cdot(\ominus z)=\bar x\cdot(y+z)+\ominus(\bar x\cdot z). Adding xˉ⋅z\bar x\cdot z and using (G2), xˉ⋅y+xˉ⋅z=(xˉ⋅(y+z)+⊖(xˉ⋅z))+xˉ⋅z=xˉ⋅(y+z)\bar x\cdot y+\bar x\cdot z=(\bar x\cdot(y+z)+\ominus(\bar x\cdot z))+\bar x\cdot z=\bar x\cdot(y+z).

(6iv) Let y,z∈Cy,z\in C be arbitrary. If y,z∈Py,z\in P, (6ii) applies. If ⊖y,⊖z∈P\ominus y,\ominus z\in P, then (6ii) gives D(⊖y,⊖z)D(\ominus y,\ominus z) and (6i) gives D(⊖⊖y,⊖⊖z)D(\ominus\ominus y,\ominus\ominus z), which is D(y,z)D(y,z) by (G2). Otherwise, since at least one of y,⊖yy,\ominus y and at least one of z,⊖zz,\ominus z lies in PP (Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §absolute), exactly one of y,zy,z lies in PP and exactly one of ⊖y,⊖z\ominus y,\ominus z lies in PP (if y∈Py\in P then z∉Pz\notin P, so ⊖z∈P\ominus z\in P and ⊖y∉P\ominus y\notin P; if y∉Py\notin P then ⊖y∈P\ominus y\in P, so ⊖z∉P\ominus z\notin P and z∈Pz\in P). By Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §absolute applied to y+zy+z, and (G2), y+z∈Py+z\in P or ⊖y+⊖z=⊖(y+z)∈P\ominus y+\ominus z=\ominus(y+z)\in P. In the first case (6iii), applied to (y,z)(y,z) or to (z,y)(z,y), gives D(y,z)D(y,z). In the second, (6iii) likewise gives D(⊖y,⊖z)D(\ominus y,\ominus z), and (6i) with (G2) gives D(y,z)D(y,z).

Step 7 (lem:dedekind-cut-multiplication-nbg-2026a#nontrivial, lem:dedekind-cut-multiplication-nbg-2026a#reciprocal, lem:dedekind-cut-multiplication-nbg-2026a#nonnegative). Nontriviality: 0∈1∗0\in1^{*} since 0<10<1, and 0∉0∗0\notin0^{*} by (Q1); so 0R=0∗≠1∗=1R0_{\mathbb{R}}=0^{*}\neq1^{*}=1_{\mathbb{R}}.

Reciprocals: let x∈Cx\in C with x≠0∗x\neq0^{*}. If x∈Px\in P, Step 4 gives w∈Pw\in P with x⊙w=1∗x\odot w=1^{*}, and x⋅w=x⊙w=1Rx\cdot w=x\odot w=1_{\mathbb{R}} by (5e). If x∉Px\notin P, then a:=⊖x∈Pa:=\ominus x\in P, and a≠0∗a\neq0^{*} since otherwise x=⊖⊖x=⊖0∗=0∗x=\ominus\ominus x=\ominus0^{*}=0^{*} by (G2). Step 4 gives w′∈Pw'\in P with a⊙w′=1∗a\odot w'=1^{*}; let w=⊖w′∈Cw=\ominus w'\in C. Then x=⊖ax=\ominus a by (G2), and by (5f) and (5e), x⋅w=(⊖a)⋅(⊖w′)=a⋅w′=a⊙w′=1Rx\cdot w=(\ominus a)\cdot(\ominus w')=a\cdot w'=a\odot w'=1_{\mathbb{R}}.

Nonnegative products: 0R≤Cx0_{\mathbb{R}}\le_{C}x and 0R≤Cy0_{\mathbb{R}}\le_{C}y mean 0∗⊆x0^{*}\subseteq x and 0∗⊆y0^{*}\subseteq y, i.e. x,y∈Px,y\in P. By (5e), x⋅y=x⊙y∈Px\cdot y=x\odot y\in P, i.e. 0R≤Cx⋅y0_{\mathbb{R}}\le_{C}x\cdot y.

Step 8 (lem:dedekind-cut-multiplication-nbg-2026a#rational).

(8a) If u,v∈Qu,v\in\mathbb{Q} with 0≤u0\le u and 0≤v0\le v, then u∗,v∗∈Pu^{*},v^{*}\in P and u∗⋅v∗=(u⋅v)∗u^{*}\cdot v^{*}=(u\cdot v)^{*}. Indeed u∗∈Cu^{*}\in C by Construction of the Dedekind Cuts: They Form a Set Totally Ordered by Inclusion, Closed under Sums, Negatives and Products of Nonnegative Cuts §rational and 0∗≤Cu∗0^{*}\le_{C}u^{*} by Addition and Order of Dedekind Cuts: an Ordered Abelian Group in Which Every Nonempty Set Bounded Above Has a Supremum §rational-order, so u∗∈Pu^{*}\in P, and likewise v∗∈Pv^{*}\in P; by (5e), u∗⋅v∗=u∗⊙v∗u^{*}\cdot v^{*}=u^{*}\odot v^{*}. Both u∗⊙v∗u^{*}\odot v^{*} and (u⋅v)∗(u\cdot v)^{*} contain 0∗0^{*} (the latter since 0≤u⋅v0\le u\cdot v by (Q2), using (Q1)), so only elements ww with 0≤w0\le w need checking. If 0≤w∈u∗⊙v∗0\le w\in u^{*}\odot v^{*}, then w=e⋅fw=e\cdot f with 0≤e<u0\le e<u and 0≤f<v0\le f<v; so 0<u0<u by (Q1), and w=e⋅f≤u⋅f<u⋅vw=e\cdot f\le u\cdot f<u\cdot v by (Q2) and (Q1), i.e. w∈(u⋅v)∗w\in(u\cdot v)^{*}. Conversely let 0≤w<u⋅v0\le w<u\cdot v. Then u⋅v≠0u\cdot v\neq0 by (Q1), so u≠0u\neq0 by Rules of Arithmetic and Order in an Ordered Field §zero and 0<u0<u. By (Q2) and (Q3), 0≤w⋅u−1<u⋅v⋅u−1=v0\le w\cdot u^{-1}<u\cdot v\cdot u^{-1}=v; take ff with w⋅u−1<f<vw\cdot u^{-1}<f<v (midpoint), so 0<f0<f by (Q1), and let e=w⋅f−1e=w\cdot f^{-1}. Then 0≤e0\le e by (Q2) and (Q3), e⋅f=we\cdot f=w, and w=(w⋅u−1)⋅u<f⋅uw=(w\cdot u^{-1})\cdot u<f\cdot u by (Q2), whence e=w⋅f−1<u⋅f⋅f−1=ue=w\cdot f^{-1}<u\cdot f\cdot f^{-1}=u by (Q2) and (Q3). So e∈u∗e\in u^{*}, f∈v∗f\in v^{*} and w=e⋅f∈u∗⊙v∗w=e\cdot f\in u^{*}\odot v^{*}.

(8b) For u∈Qu\in\mathbb{Q}, (−u)∗=⊖(u∗)(-u)^{*}=\ominus(u^{*}): by Addition and Order of Dedekind Cuts: an Ordered Abelian Group in Which Every Nonempty Set Bounded Above Has a Supremum §rational-sum and The Rational Numbers Form an Archimedean Ordered Field Containing the Integers §ordered-field, u∗+(−u)∗=(u+(−u))∗=0∗u^{*}+(-u)^{*}=(u+(-u))^{*}=0^{*}, while u∗+⊖(u∗)=0∗u^{*}+\ominus(u^{*})=0^{*} by Addition and Order of Dedekind Cuts: an Ordered Abelian Group in Which Every Nonempty Set Bounded Above Has a Supremum §group; apply (G1).

(8c) Let u,v∈Qu,v\in\mathbb{Q}, and let s=∣u∣s=|u| and t=∣v∣t=|v|, so 0≤s0\le s and 0≤t0\le t by Rules of Arithmetic and Order in an Ordered Field §absolute-value. By Absolute Value in an Ordered Field §absolute-value, if 0≤u0\le u then u=su=s, and we set ε=id\varepsilon=\mathrm{id}; otherwise s=−us=-u, so u=−su=-s by Rules of Arithmetic and Order in an Ordered Field §signs, and we set ε=⊖\varepsilon=\ominus. In both cases u∗=ε(s∗)u^{*}=\varepsilon(s^{*}), by (8b) in the second. Define η\eta from vv and tt in the same way, so v∗=η(t∗)v^{*}=\eta(t^{*}). By Rules of Arithmetic and Order in an Ordered Field §signs and commutativity, u⋅v=s⋅tu\cdot v=s\cdot t if ε=η\varepsilon=\eta, and u⋅v=−(s⋅t)u\cdot v=-(s\cdot t) if ε≠η\varepsilon\neq\eta; in the first case εη=id\varepsilon\eta=\mathrm{id} by (G2), in the second εη=⊖\varepsilon\eta=\ominus, so in both cases (u⋅v)∗=εη((s⋅t)∗)(u\cdot v)^{*}=\varepsilon\eta((s\cdot t)^{*}), using (8b) in the second. On the other hand, by (5f) and (8a), u∗⋅v∗=(ε(s∗))⋅(η(t∗))=εη(s∗⋅t∗)=εη((s⋅t)∗)u^{*}\cdot v^{*}=(\varepsilon(s^{*}))\cdot(\eta(t^{*}))=\varepsilon\eta(s^{*}\cdot t^{*})=\varepsilon\eta((s\cdot t)^{*}). Hence (u⋅v)∗=u∗⋅v∗(u\cdot v)^{*}=u^{*}\cdot v^{*}.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…