Arithmetic with real numbers
How to add, subtract, multiply and divide sections, and why the result is always a well-defined real number.
Why this matters
A number system is useless unless we can calculate in it. If and are sections, what is the section ? In §§10–11 Hardy defines the four operations directly in terms of the classes and shows the one fact that makes the definitions work: the new classes squeeze together so tightly that at most one rational can slip between them. After that, all the familiar laws of algebra can be checked, and real numbers can be used exactly like fractions.
The idea
Suppose has lower class and upper class , and has and . Whatever is, it ought to be larger than every sum (each term is below the corresponding number) and smaller than every sum . So form two new classes:
Every is less than every . The question is whether these two classes leave room for more than one number between them. They do not, because and can both be made as small as we like, and .
Try this. With and , increase the number of decimal places and watch the gap between the classes and . Then set . Finally try , .
- sums a + b
- sums A + B
- α + β
| lower member | upper member | |
|---|---|---|
| a = 1.4 | A = 1.5 | |
| b = 1.7 | B = 1.8 | |
| c = 3.1 | C = 3.3 |
What to notice
Using -place decimals for gives a gap of exactly , so the sum is trapped more tightly at every step. For the rational stays inside every gap: it is the single rational that belongs to neither class, and the sum is . For nothing escapes, because is itself one of the sums .
So there are two possibilities. Either every rational falls into or , and these classes are a section; or exactly one rational is left over, and we put it into (where it becomes the least member, so the sum is that rational). In both cases we get a real number, and that is the definition of .
Precise statements
Let be the class of all sums and the class of all sums . If every rational lies in or , then is the section . Otherwise exactly one rational lies in neither; add it to , and is the resulting section. The difference is defined by .
For any real numbers and , the classes and leave at most one rational unclassified, and the construction above gives a real number. If and are rational, it is their ordinary sum.
For positive and , use sections of the positive rationals: the lower class of is formed by all products and the upper class by all products , with at most one leftover rational added to the upper class. For other signs,
For positive , the reciprocal has lower class all and upper class all ; for negative numbers ; and for .
With these definitions every law of elementary algebra (commutative, associative and distributive laws, the rules for signs and inequalities) holds for real numbers. Hardy proves none of them in detail, and leaves a selection as Examples V and VI; a few are below.
Proofs
The proof uses one fact from the previous lessons: in any section, a lower and an upper member can be found as close as we please. (Take any , , fill the gap with a chain of rationals with small steps, and take the last term in the lower class and the next one.)
Show that lies below , that and have the shape of the two classes of a section, and that the gap between them is too narrow for two rationals. Then add the leftover rational, if any, to the upper class.
Every is less than every
If and then and , so .
is closed downwards and has no greatest member
If then with , so is in (a lower class contains every rational below one of its members) and is in . If , pick in (no greatest member); then is a larger member of . Similarly is closed upwards.
At most one rational lies in neither class
Suppose were both in neither class. Then for all and , so always. But , and we may choose and . Contradiction.
Form the section
If no rational is left over, , is a section whose lower class has no greatest member: a real number. If one rational is left over, every is below and every above it (otherwise, by the closure property, would be in a class), so adding to gives a section in which is the least upper member: the rational number .
Agreement for rational ,
If and are rational, they are the least members of and , so is a sum and every other is larger: it is the least member of . So the new sum is the old one.
For products of positive numbers the argument is the same, with one change in the third step. The gap is now
Fix some upper members , and only use , with and , with . Then , which again can be made smaller than any given positive number.
Try this. Choose and and step up. Compare the gap with the gap for the sum above. Then choose .
- products ab
- products AB
- αβ
| lower member | upper member | |
|---|---|---|
| a = 1.4 | A = 1.5 | |
| b = 1.7 | B = 1.8 | |
| c = 2.38 | C = 2.70 |
What to notice
The product gap is , a little more than here: bigger than the gap for the sum, but still shrinking tenfold at each step. For the rational stays inside every gap; the next lesson is about exactly this.
Worked examples and exercises
Prove that for every real number . Take irrational first.
Hint
Hint
Solution
Let be irrational. The lower class of the sum is made of all , which are negative since ; the upper class of all , which are positive. So is in neither class. By the theorem at most one rational is unclassified, so every other rational is in one of the classes; the negatives cannot be in the upper class (all positive), so all negative rationals are in the lower class and all positive rationals in the upper. Adding the leftover to the upper class gives the section whose lower class is exactly the negatives: the number .
If is rational, the lower class of is all with , and the upper class is all together with . The lower sums are still negative and the upper sums (all , and , with among them) are , with the least. The same reasoning gives the section .
Prove that and .
Hint
Solution
Commutativity. The classes and are the same set of rationals, and so are and ; so the construction produces the same section.
Associativity. By the proof above, the lower class of a sum is exactly the set of all , and this equals the set of rationals lying below some (it is closed downwards). Apply this twice: the lower class of is the set of all with below some , which is the set of rationals below some . The same description holds for , since rational addition is associative. Two real numbers with the same lower class are equal.
Define directly by the classes and , and check that the gap between them shrinks to .
Hint
Solution
Since and , we have : every is below every . The gap is
which can be made as small as we please, so as before at most one rational is unclassified and we obtain a real number. It is the same as : for irrational the lower class of consists of the numbers , so the sums are exactly the ‘s here (and similarly for the upper class).
For positive , prove that .
Hint
Solution
The lower class of the product consists of products , all less than since . The upper class consists of products , all greater than . So is in neither, and since at most one rational escapes, it is the one: every other positive rational lies in one of the classes. Putting in the upper class gives the section . (If is rational, the lower class of is all with and the upper class contains as well, so the products include as their least member; the conclusion is the same.)
Check yourself
For and , take and to be the -place decimals just below and just above each number. What is ?
. Concretely , , , , so and .
The sum of two irrational numbers is always irrational.
False: . In the section picture, is the one rational that escapes both classes, and the definition puts it in the upper class. A leftover rational happens exactly when the sum is rational but and are not.
Why does Hardy define products first for positive numbers only?
Multiplying by a negative number reverses inequalities. For , the lower members give , while upper members give : the “lower” products are not below the “upper” ones. Restricting to positive members avoids this, and the sign rules handle the rest.
For which pair is there a rational number that belongs to neither nor ?
A rational escapes exactly when the sum is a rational number that is not itself a sum . With and the sum is , and since neither upper class has a least member, every exceeds and every is below it. For the sum is itself with . The other two sums are irrational, so nothing escapes.
Hardy’s language
We are then in a position to apply to all real numbers, rational or irrational, the whole of the ideas and methods of elementary algebra.
Summary
- is the section with lower class all and upper class all ; if one rational is left over it goes into the upper class.
- The key fact is that the gap can be made as small as we please, so at most one rational is left over.
- Products of positive numbers use and , with gap ; signs are handled by the usual rules, and uses and .
- With these definitions all the laws of elementary algebra hold for real numbers.