Introducing the reals as a totally ordered additive group
without using the rationals
Citation for published version (APA):
Bruijn, de, N. G. (1975). Introducing the reals as a totally ordered additive group without using the rationals. (Eindhoven University of Technology : Dept of Mathematics : memorandum; Vol. 7513). Technische Hogeschool Eindhoven.
Document status and date: Published: 01/01/1975 Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne Take down policy
If you believe that this document breaches copyright please contact us at: openaccess@tue.nl
providing details and we will investigate your claim.
EINDHOVEN UNIVERSITY OF TECHNOLOGY
Department of Mathematics
Memorandum 1975-13
Issued November 1975
Introducing the reals as a totally ordered additive group without using the rationals
University of Technology Department of Mathematics PO Box 513, Eindhoven The Netherlands by N.G. de Bruijn
6qb~~
- 1
-Introducing the reals as a totally ordered additive group without using
the rationals by N.G. de Bruijn.
I. The purpose of this note is to define the reals as infinite decimal fractions (we use the word decimal also for the case where the base is not ten but any positive integer b), and to derive its additive structure
and its order without reference to rational numbers. Actually our treat-ment can be made entirely combinatorial. The arithmetic operations we use
in order to define operations on decimal fractions, are restricted to a limited range (especially if b
=
2), and can be replaced by combinatorial operations.Our point of departure is the set E of all integers. It can be
characterized as a non-empty totally ordered set without maximal or minimal element, with the property that every non-empty subset which is bounded above (below) has a maximal (minimal) element. We use z + 1 and
z - I in order to indicate successor and predecessor.
The author intends to write another version of this paper where the role of the addition algorithm is taken over by the subtraction algorithm.
(This is slightly simpler: the study of inequality between decimal
fractions is related to the subtraction algorithm). That version will have a more extensive introduction, and will say something on the prospects of multiplication.
2. Notation. Z ~s the set of all integers. b is a fixed integer· >1.
If P and Q are sets then QP is the set of all mappings of Pinto Q.
L is the set of all f E {O,I, •.• ,b-l}Z with the following property: for
every z E :l. there exists an x E Z with x > z and f(x) < b-l. Z
I f f,g E Z then f + g is defined by (f+g) (z) = fez) + g(z) (z E E).
This f + g is not the same thing as sum(f,g) (to be introduced in Section 4).
If f E Zl then (f) is the function h that satisfies, for all z E Z,
mod b
h(z) E l , h(z) - fez) _
°
(mod b),°
~ h(z) < b.If A is any rational number, then [A] denotes the largest integer ~A.
3. Carry. If k E {0,1, ••• ,2b-2}Z we define an element p of {O,I}Z which is
to be called the carry of k, denoted as p
=
carr(k). For every z E Z wedefine p(z) as follows: p(z)
°
if there exists an x E 2, with x > z, k(x) < b-l, and such that k(u) ~ b-l for all u with u E 2, z < u < x. Inall other cases we put p(z)
=
I.E
Theorem 3.1. if k E {0,1, ••• ,2b-2} , p
=
carr(k) then we have, if z E Z,p(z-l)
=
[(k(z) + p(z»/b].Proof. From the definition of carr we infer: if k(z) < b-l then p(z-I)
=
0, if k(z) > b-I then p(z-I)=
I, if k(z)=
b-I then p(z-I)=
p(z).4. Sum. If f,g E {O,I, •.. ,b-l} we have f+g Z E {0,1, ••. ,2b-2} , and we E
define sum(f,g) by
3
-Theorem 4. 1 • I f f,g E {O,l, ••• ,b-I} then sumCf,g) E {O,l, .•• ,b-I} and
z
Z
sum(f,g)=
sum(g,f).Theorem 4.2. If f,g E ~ then sum(f,g) E E.
Proof. Assume that f and g are in E but s ~s not, where s
=
sum(f,g). Sothere is some w E Z such that s(z)
=
b-l for all z > w. We put f+g=
k, carr(k) = p, whence s=
(k+p)mod b' If z > w we have either k(z) + p(z)=
b-1 or k(z) + p(z) = 2b-l. Since k(z) < 2b-l, p(z) s I, we have just three possibilities if z > w:(i) k(z) = b-I, p(z) = 0, p(z-I)
=
0;(ii) k(z) = b-2, p(z) = I , P (z-I)
=
0; (iii) k(z) = 2b-2, p(z) = I , p(z-l) = 1.(the values of p(z-I) follow from Theorem 3.1). We cannot have case (i) all the time, for if k(z) = b-l for all z > w we have pew) = 1. If z is in case (ii) or (iii), then its successor z + is in case (iii). So there is some
v > w such that we are in case (iii) for all z > v. But k(z) = 2b-2 implies fez)
=
g(z)=
b-l. Since fEE we have a contradiction.5. Associativity of sum.
Theorem 5.1, If f,g,h E ~ then we have
sum(sum(f,g),h)
=
sum(f,sum(g,h».Proof. We put p
=
carr(f+g), s=
sum(f,g), q=
carr(s+h), r=
p+q, t=
sum(s,h), It follows thatWith some fixed z E Z we put
A
=
(f(z) + g(z) + p(z»/b, and we obtains(x)/b = A - [AJ.
From Theorem 3.1 we get
p(z-l) == DJ, q(z-l) == [A-CAJ +
h(Z)~q(Z)J,
whence
r(z-l) == [(fez) + g(z) + h(z) + r(z»/bJ. (5.2)
We next interchange the roles of f and h, and we write p* = carr(h+g),
*
**
* * **
*
s
=
sum(h,g), q=
carr(s +h), r = p +q) t = sum(s ,f). As a counter-part to (5.2) we get*
*
r (z-I) == [(fez) + g(z) + h(z) + r (z»/bJ. (5.3)
* *
Since p,q,p ,q have values 0 and 1 only, we have fr(z) - r*(z)1 ~ 2 for all z, whence, by (5.2) and (5.3), !r(z-I) - r*(z-I)1 ~ 1 for all z. Therefore Ir(z) - r*(z)1 ~ 1 for all z.
*
Let us assume that there is a z E E with r(z-l) < r (z-I). Comparing
(5.2) and (5.3), and using /r*(z) - r(z) I ~ I, we derive that r (z) - r(z)
*
== 1,and next that fez) + g(z) + h(z) + r(z) _ b-I(mod b) (for otherwise (5.2)
*
*
and (5.3) would give r(z-l) == r (z-I». Thus r(z) < r (z) and t(z)
=
b-I.Repeating the argument we get r(y) < r*(y) and t(y)
=
b-l for all y ~ z. This contradicts t Er
(cf. Theorem 4.2). The conclusion is that*
*
r(z) < r (z) can hold for no z E Z. Similarly we can show that r (z) <~r(z)
*
* *
for no z E E. Hence r = r , and from (5.1) (plus its counterpart for t ,r ), we get t = t •
*
ThLs proves the theorem..
5
-6. Zero element. By fO we denote the element of E with fO(z) = 0 for all
z E Z.
Theorem 6.1. For all fEE we have sum(f,fO) = f.
Proof. Follows from the definition of sum. Note that carr(£)
=
£0'7. Additive inverse. If fEE we define min(£) by
min(f)
=
(g + carr(g»mod b' (7. I)where g is defined by
g(z)
=
b-I - fez) (z E Z). (7.2)Theorem 7.1. If fEE then min(f) E E and sum(f,min(f»
=
f O'Proof. We define g by (7.2), and further h = carr(g), k = min(f), r
=
f+k, s=
carr(r). We distinguish three cases.(i) £
=
fO' Now g(z)
=
b-) for all z, and h(z)=
1 for all z. Hence min(f) == fO' and sum(fO' min(fO» == fO by Theorem 6.1.
(ii) There 1S an x E" with f(x) '" 0 and such that fez) == 0 for all
z > x. Now g(x) '" b-l, and g(z)
=
b-l for all z > x. Next 'we observe h(z) == 1 (z;::: x), h(z) = 0 (z<x), and therefore k(z) = b-l - f (z) (z<x) , k(x) = b-f(x), k(z) == 0 (z>x). This shows k E E. Next we note thatr(z) == b-l, b or 0 according to z<x, z=x, z>x, and s(z)
=
1,0,0 accordingto z<x, z=x, z>x. It follows that sum(f,k) = (r+s)mod b = fOe
(iii) For all y E E there is an x > y with f(x) '" O. Obviously h
=
fO' whence k = g. For all y we have an x > y with k(x) == b-l - f(x) '" b-l,whence k E L. Finally rex) = b-l, sex)
=
I for all x E Z, whence sum(f,k) = (r+s)mod b=
O.8. The additive group. It follows that at once from Theorems 4.1, 4.2, 5.1,
6. I, 7. I that (E ,sum) is an abelian group with neutral element f
O' In particular, if fEE, gEE then the unique solution h of sum(g,h)
=
f•
equals sum(f,min(g». A consequence is that min(min(g»
=
g for all gEE.9. The reals. An fEE is called positive if f
#
0 and there is an x E Z withfey)
=
0 for all y < x. We say that fEE is negative if there is an x E Z such that fey)=
b-l for all y < x. An fEE is called real if f is positive or negative or zero (i.e. fO)' These three alternatives exclude each other. The set of all real elements of E is denoted by R.
Theorem 9.1. If f is positive then min(f) 1S negative. If f is negative
then min(f) is positive.
Proof. (i) Let f be positive. Then x exists such that f(x)
#
0 and such that fey)=
0 for all y < x. Now with g of (7.2) we have g(y) = b-) for all y < x, and carr(g) has just zeros to the left of x (note that g(x) < b-l). Hence fey)=
b-) (y<x).(ii) Let f be negative. Since fEE we do not have fez)
=
b-I for all Z E Z. Let x be the least integer with f(x) # b-l. Now fey)=
b-l forall y < x, whence both g and carr(g) have just zeros to the left of x, and min(f) appears to be positive.
Theorem 9.2. If f and g are real then sum(f,g) is real. If f and g are
positive then surn(f,g) is positive. If f and g are negative then sum(f,g) is negative.
Proof. Let f,g E R. Put k
=
f+g, p=
carr(k), s=
sum(f,g). Let x be suchthat fey)
=
y and g(y)=
0 for all y S x, where both y and 0 are eithero
or b-I. If Y=
0=
b-I then p(y) for y < x, whence s(y)=
b-) for y < x, and s is negative. If y=
0o
then p(y)=
0 for y < x, whence7
-S(y)
=
a for y < x, and s is positive or zero. If f and g are positive thecase that s is zero is excluded: it would imply f
=
min(g), which contradicts Theorem 9.1. Finally if y # 0 we have key)=
b-I (y~x) and Theorem 3.1 showsthat p(y-I)
=
p(y) for y ~ x. It follows that either p(y) a 0, s(y)=
b-l for.all y ~ x (i.e. s is negative) or p(y) = I, s(y)
=
°
for all y ~ x (i.e. s is positive or zero).Theorem 9.3. OR,sum) is a subgroup of (L,sum).
Proof. Follows from Theorems 9.1 and 9.2.
10. Order. If f,g E
m
then we say that f > g if sum(f,min(g» is positive.Theorem 10.1. If f,g,h Em, we have
(i) Exactly one of the relations f > g, f
=
g, g > f holds. (ii) If f > g, g > h then f > h.(iii) If f > g then sum(f,h) > sum(g,h).
(iv) f is positive if and only if f > fa, and negative if and only if
fa > f.
Proof. (i) From the group properties we derive that if k
=
sum(f,min(g»,~
=
sum(g,min(f», then k=
min(~). Since k is either positive or zero or negative, application of Theorem 9.1 is sufficient.(ii) This follows from the group properties if we use (Theorem 9.2)
• that two positive elements have a positive sum.
(iii) From the group properties we derive
sum(sum(f,h), min(sum(g,h»)
=
sum(f,min(g». (iv) Note that sum(f ,min(fO» = f, sum(fO' min(£) = min(f), and apply Theorem 9. 1 •
Theorem 10.2. (i) If f is positive and g negative then f > g. (ii) If f and g have the same sign (i.e. both positive or both negative), we have
f > g if and only if there is an x E E with f(z) > g(x) and such that
f(y)
=
g(y) for all y < x.Proof. (i) Let f be postive and g negative. Then the h with f
=
sum(g,h) cannot be negative or zero since Theorem 9.1 would tell us that f is negative. Hence k is positive, whence f > g.(ii) Let f and g have the same sign. Let h be the element with
f sum(g,h). We put g+h
=
k, carry(k)=
p, whence (k+p)mod b=
f. Let x be the last integer such that f(x)#
g(x). We first assume f(x) > g(x). Since g(x) + h(x) + p(x) _ f(x) (mod b), 0 s h(x) < b,o
s p(x) s we infer 0 s h(x) + p(x) s b, g(x) + hex) + p(x)=
f(x), whence, by Theorem 3.1, p(x-l)=
O. Since f(x-l)=
g(x-I),o
s h(x-I) + p(x-I) < b, the same argument leads to p(x-2)=
0, and by induction we get p(y) = 0 for all y < x. Since g(y) = f(y) for ally < x it follows that hey)
=
0 for all y < x. So h is positive, whencef > g.
In the case where g(x) > f(x) we interchange the roles of f and g and we get g > f.