Introduction

Recall that a rational number is a number that can be written in the form where are integers with . The rational number system is all you need to accomplish most everyday tasks. For instance, to measure distances when building a house it suffices to use a tape measure with an accuracy of about of an inch. However, to do mathematical analysis the rational numbers have some very serious shortcomings; here is a an example.
If then is not a rational number.
Suppose by contradiction that there exists such that . We may write for some integers , and we can assume that and have no common factor other than (that is, and are relatively prime). Now, since then and thus is an even number. This implies that is also even. Since is even, we may write for some and therefore , from which it follows that . Hence, is even and thus is also even. Thus, both and are even, which contradicts the fact that and are relatively prime.
The previous theorem highlights that the set of rational numbers are in some sense incomplete, or that there are gaps in , and that a larger number system is needed to enlarge the set of math problems that can be analyzed and solved. Although mathematicians in the 1700s were using the real number system and resorting to limiting processes to analyze problems in physics, it was not until the late 1800s that mathematicians gave a rigorous construction of the real number system. Motivated by Theorem 2.1.1, we might be tempted to define the real numbers as the set of solutions of all polynomial equations with integer coefficients. As it turns out, however, this definition of the reals would actually leave out almost all the real numbers, including some of our favorites like and . In fact, the set of all numbers that are solutions to polynomial equations with rational coefficients is countable! There are two standard ways to construct the set of real numbers. One standard method to construct uses the notion of Cauchy sequences of rational numbers and is attributed to Georg Cantor \cite{DG:01}. We will cover Cauchy sequences in Section and therefore postpone describing some of the details of the construction until then. The second standard method to construct the reals relies on the notion of a Dedekind cut and is attributed to Richard Dedekind (1831-1916). A Dedekind cut is a partition of such that both and are non-empty and
  1. if and then , and
  2. for any there exists such that .
The set of real numbers is then defined to be the set of all Dedekind cuts. For example, let and thus . Then one can show that is a Dedekind cut of and the idea is that represents the real number such that , that is, the irrational number . Having defined as the set of Dedekind cuts we then proceed to define all the usual operations of arithmetic and arrive at the familiar model of \cite{DG:01}. Additionally, if and then we write that if and write if and . Refer to \cite{DG:01} for further details. In this book, we instead adopt the familiar viewpoint that the real numbers are in a one-to-one correspondence with the points on an infinite line:
figures/real-line.svg
The real numbers are in a one-to-one correspondence with points on an infinite line
The essential feature that we want to capture with this view of is that there are no holes in the real number system. This view of allows us to quickly start learning the properties of instead of focusing on the details of constructing a model for . Naturally, the rational numbers are a subset of and we say that a real number is irrational if it is not rational. As we saw in Thereom 2.1.1, the positive number such that is irrational.

Exercises

Let be fixed. Prove the following statements without using proof by contradiction.
  1. Prove that if then .
  2. Suppose in addition that . Prove that if then .
Prove that if then for all natural numbers . Do not assume that is rational.

Algebraic and Order Properties

We will soon see the main difference between and from an analysis point of view but in this section we will discuss one important thing that and have in common, namely, both are ordered fields. We begin then with the definition of a field.
A field is a set with two binary operations and , the former called addition and the latter multiplication, satisfying the following properties:
  1. for all
  2. for all
  3. There exists an element such that for all
  4. There exists an element such that for all
  5. For each , there exists an element such that .
  6. For each , there exists an element such that .
It is not hard to see that and are not fields. In each case, what property of a field fails to hold?
Both and are fields.
Besides being fields, both and are totally ordered sets. By totally ordered we mean that for any either , , or . This property of is referred to as the Law of Trichotomy. For , the relation means that either or . Similarly, means that . From our number line viewpoint of , if then is on the left of . We now present some very important rules of inequalities that we will use frequently in this book.
Let .
  1. If and then . (transitivity)
  2. If then .
  3. If and then .
  4. If and then .
  5. If then either or .
  6. If then .
The two inequalities and as sometimes combined as
Suppose that and , or written more compactly as . Prove that if then .
Suppose that and . Prove that . Deduce that if and then
We will encounter situations where we will need to prove that if two numbers satisfy a certain property then . Proving that is equivalent to proving that . In such situations, the following theorem will be very useful.
Let be non-negative, that is, . If for every it holds that then .
We prove the contrapositive, that is, we prove that if then there exists such that . Assume then that and let . Then and clearly .
The next few examples will give us practice with working with inequalities.
Let . Find a natural number such that
Let . Find analytically a natural number such that
Let . Find analytically a natural number such that

Exercises

Let and find analytically a natural number such that
Let and find analytically a natural number such that
Let and find analytically a natural number such that

The Absolute Value

To solve problems in calculus you need to master differentiation and integration. To solve problems in analysis you need to master inequalities. The content of this section, mostly on inequalities, is fundamental to everything else that follows in this book. Given any , we define the absolute value of as the number Clearly, if and only if , and for all . Below we record some important properties of the absolute value function.
Let and .
  1. if and only if
Statements (i) and (ii) are trivial. To prove (iii), first suppose that . If then . Hence, and since then . Hence, . If then , and thus . Since then . Thus, . Now suppose that . If then . If then from multiplying the inequality by we have and thus . To prove part (iv), notice that and thus applying (iii) we get .
If prove that .
From Theorem 2.3.1 part (i), we have that . Therefore, . Similarly, one can show that . By induction, for each it holds that .
Below is the most important inequality in this book.
For any it holds that
We have that from which it follows that and thus
By induction, one can prove the following corollary to the Triangle inequality.
For any it holds that
A compact way to write the triangle inequality using summation notation is Here is another consequence of the Triangle inequality.
For it holds that
For part (i), we have For part (ii), consider and therefore . Switching the role of and we obtain , and therefore multiplying this last inequality by yields . Therefore, which is the stated inequality.
For prove that .
Let for . Find a number such that for all .
Clearly, if then . Apply the triangle inequality and the properties of the absolute value: Therefore, if then for all .
Let . Find a number such that for all .
It is clear that if then . Using the properties of the absolute value and the triangle inequality repeatedly on the numerator: Now, for we have that and therefore and then . Therefore, Hence, we can take .
Let . Find a number such that for all .
In analysis, the absolute value is used to measure distance between points in . For any , the absolute value is the distance from to . This interpretation of the absolute value can be used to measure the difference (in magnitude) between two points. That is, given , the distance between and is . From the properties of the absolute value, this distance is also .
figures/absolute-value-distance.svg
The number is the distance between and .
We will often be concerned with how close a given number is to a fixed number . To do this, we introduce the notion of neighborhoods based at .
Let and let . The -neighborhood of of radius is the set
Notice that if then .
Let and let . From calculus, we know that . Find a natural number such that for every .
The inequality means that is within of . That is, This inequality does not hold for all , but it will eventually hold for some and for all . For example, and , and similarly for and . In fact, . However, because we know that , eventually for large enough . To find out how large, let's analyze the magnitude : Hence, if and only if which after re-arranging can be written as With we obtain Hence, if then if then .
Let and , and let . Show that and are -neighborhoods of for some appropriate value of .

Exercises

Prove that if and then . Draw a number line with points satisfying the inequalities and graphically interpret the inequality .
Let be positive real numbers and consider the polynomial Prove that for all . {Hint: For example, if say then you are asked to prove that However, prove the claim for a general polynomial with .}
Let for . Find analytically a number such that for all . Do not use calculus to find .
Let for . Find analytically a number such that for all . Do not use calculus to find .
Let for . Find analytically a number such that for all . Do not use calculus to find . (Hint: Complete the square.)
Let be distinct points. Show that there exists neighborhoods and such that .

The Completeness Axiom

In this section, we introduce the Completeness Axiom of . Recall that an axiom is a statement or proposition that is accepted as true without justification. In mathematics, axioms are the first principles that are accepted as truths and are used to build mathematical theories; in this case real analysis. Roughly speaking, the Completeness Axiom is a way to say that the real numbers have no gaps or no holes, contrary to the case of the rational numbers. As you will see below, the Completeness Axiom is centered around the notions of bounded sets and least upper bounds; let us begin then with some definitions.
Let be a non-empty set.
  1. We say that is bounded above if there exists such that for all . We then say that is an upper bound of .
  2. We say that is bounded below if there exists such that for all . We then say that is a lower bound of .
  3. We say that is bounded if it is both bounded above and bounded below.
  4. We say that is unbounded if it is not bounded.
For each case, determine if is bounded above, bounded below, bounded, or unbounded. If the set is bounded below, determine the set of lower bounds, and similarly if it is bounded above.
Let and be sets and suppose that .
  1. Prove that if is bounded above (below) then is bounded above (below).
  2. Give an example of sets and such that is bounded below but is not bounded below.
We now come to an important notion that will be at the root of what we do from now.
Let be non-empty.
  1. Let be bounded above. An upper bound of is said to be a least upper bound of if for any upper bound of . In this case we also say that is a supremum of and write .
  2. Let be bounded below. A lower bound of is said to be a greatest lower bound of if for any lower bound of . In this case we also say that is an infimum of and write .
It is straightforward to show that a set that is bounded above (bounded below) can have at most one supremum (infimum). At the risk of being repetitive, when it exists, is a number that is an upper bound of and is the smallest possible upper bound of . Therefore, for all , and any number less than is not an upper bound of (because is the least upper bound!). Similarly, when it exists, is a number that is a lower bound of and is the largest possible lower bound of . Therefore, for all and any number greater than is not a lower bound of (because is the greatest lower bound of !).
In some analysis texts, is written as and is written as . In other words, and .
Does every non-empty bounded set have a supremum/infimum? You might say Yes, of course!! and add that It is a self-evident principle and needs no justification!. Is not that what an axiom is?
Every non-empty subset of that is bounded above has a least upper bound (a supremum) in . Similarly, every non-empty subset of that is bounded below has a greatest lower bound (an infimum) in .
As you will see in the pages that follow, The Completeness Axiom is the key notion upon which the theory of real analysis depends on.
Determine and , if they exist.
The Completeness Axiom is sometimes called the supremum property of or the least upper bound property of . The Completeness property makes into a complete ordered field. The following example shows that does not have the completeness property.
The set of rational numbers is an ordered field but it is not complete. Consider the set . By definition, . Clearly, is bounded above, for example is an upper bound of , but the least upper bound of is which is not a rational number. Therefore does not have a supremum in and therefore does not have the Completeness property. From the point of view of analysis, this is the main distinction between and (note that both are ordered fields).
In some cases, it is obvious what and are, however, to do analysis rigorously, we need systematic ways to determine and . To start with, we first need to be a bit more rigorous about what it means to be the least upper bound, or at least have a more concrete description that we can work with, i.e., using inequalities. The following lemma does that and, as you will observe, it is simply a direct consequence of the definition of the supremum.
Let be non-empty and suppose that is an upper bound of . Then is the least upper bound of if and only if for any there exists such that
Suppose that is the supremum of , that is, is the least upper bound of . Since then is not an upper bound of . Thus, there exists such that . Now suppose that for any there exists such that . Now let be such that . Then there exists such that , and thus by assumption, there exists such that . Hence, is not an upper bound of and this shows that is the least upper bound of , that is, .
If , and is bounded above, prove that is bounded above and that .
Since is bounded above, exists by the Completeness property of . Let . Then and therefore . This proves that is bounded above by and therefore exists. Since is the least upper bound of we must have . For example, if say and then , while if then .
Let be non-empty and bounded above. Let and define the set Prove that if then is bounded above and . Show by example that if then need not be bounded above even if is bounded above.
Let be arbitrary. Then there exists such that . By the Completeness property, exists and . If then which is equivalent to . Since is arbitrary, this shows that is an upper bound of the set and thus is bounded above and thus exists. Because is the least upper bound of then Now, by definition, for all . Thus, for all and therefore for all . Therefore, is an upper bound of and consequently because is the least upper bound of . We have therefore proved that Combining and we conclude that
Suppose that and are non-empty and bounded above. Prove that is bounded above and that
Let . Then clearly and . We first show that is bounded above by showing that is an upper bound of . Let . Then either or (or both). If then and if then . Hence, is bounded above and is an upper bound of . Consequently, exists by the Completeness axiom and moreover , that is, Now, by definition of the supremum, for all . Now since this implies that for all and for all . In other words, is an upper bound of both and and thus and . Then clearly and combining and we have proved that .
Suppose that and are non-empty and bounded below, and suppose that is non-empty. Prove that is bounded below and that
If then and therefore , and also and thus . Therefore, both and are lower bounds of , and by definition of we have that and , and consequently .
For any set define the set Prove that if is non-empty and bounded then .
It holds that for all and therefore, for all , which is equivalent to for all . Therefore, is an upper bound of the set and therefore . Now, for all , or equivalently for all . Hence, for all . This proves that is a lower bound of and therefore , or . This proves that .
Let be non-empty and bounded below. Let and define the set Prove that is bounded below and that .
For all it holds that and therefore . This proves that is a lower bound of the set , and therefore . Now, for all and thus for all , which is the same as for all . Hence, or equivalently . This proves the claim.
Let and be non-empty subsets of , and suppose that and are bounded below. Define the set .
  1. Prove that is bounded below.
  2. Prove that . Hint: Consider two cases, when say and when .
  3. How do things change if we do not assume and are subsets of .
Give an example of a non-empty set such that .
For any two non-empty sets and of let us write that if, for each , there exists such that .
  1. Prove that if then .
  2. Show via an example that if then it is not necessarily true that .
Let and be non-empty bounded sets of positive real numbers. Define the set Assume that . Prove that
Since is bounded above, exists and for all . Since is bounded below, exists and for all . Let be an arbitrary point in . Then since and and we obtain that Since and (and then clearly ) we obtain This proves that and since was arbitrary we have proved that is an upper bound of the set . This proves that exists. Moreover, by the definition of the supremum, we also have that Now, for all and thus for all and all . If is held fixed then for all and thus . Therefore, , which holds for all . Therefore, and consequently Combining and completes the proof.

Exercises

Let be a non-empty set and suppose that is an upper bound of . Prove that if then necessarily .
Let and be non-empty subsets of , and suppose that . Prove that if is bounded below then .
If and are non-empty subsets of such that and does it follow that ? Support your answer with either a proof or a counterexample.
Let be a bounded set, let be fixed, and define the set Prove that .
Let be bounded above. Let Prove that is bounded above and that .
Let denote the set of all positive real numbers and let be bounded. Assume that . Define the set Prove that

Applications of the Supremum

Having defined the notion of boundedness for a set, we can define a notion of boundedness for a function.
Let be a non-empty set and let be a function.
  1. is bounded below if the range is bounded below.
  2. is bounded above if the range is bounded above.
  3. is bounded if the range is bounded.
Hence, boundedness of a function is boundedness of its range.
Suppose are bounded functions and for all . Show that .
Since is bounded, exists and is by definition an upper bound for the set , that is, for all . Now, by assumption, for all it holds that and therefore . This shows that is an upper bound of the set , and therefore by definition of the supremum, .
Let be bounded functions and suppose that for all . Show that .
Fix . Then, for all . Therefore, is a lower bound of , and thus by definition of the infimum. Since was arbitrary, we have proved that for all . Hence, is an upper bound of and thus by definition of the supremum we have that .
The following sequence of results will be used to prove an important property of the rational numbers as seen from within .
If then there exists such that .
Suppose not. Hence, for all , and thus is an upper bound for , and therefore is bounded. Let . By definition of , is not an upper bound of and therefore there exists such that . But then and this contradicts the definition of .
If then .
Since for all then is a lower bound of . Now suppose that . By the Archimedean Property, there exists such that and thus . Hence, is not a lower bound of . Therefore is the greatest lower bound of , that is, .
For any there exists such that .
Given there exists such that .
Let . By the Archimedean property, is non-empty. By the Well-Ordering Principle of , has a least element, say that it is . Hence, and thus .
We now come to an important result that we will use frequently.
If and then there exists such that .
We first prove the claim for the case that . Suppose that and thus . There exists such that and thus . Therefore, and thus and we may take . In general, if then there exists such that and thus . Since there exists such that and thus and dividing by yields and thus we may take . This proves the claim when both and are positive. If then take and if then apply the previous arguments to .
Hence, between any two distinct real numbers there is a rational number. This implies that any irrational number can be approximated by a rational number to within any degree of accuracy.
Let be an irrational number and let be arbitrary. Prove that there exists a rational number such that , that is, is in the -neighborhood of .
If and then there exists such that .
We have that . By the Density Theorem, for some . Then .

Exercises

Let be non-empty and let be functions. Let denote the function defined by for any . If and are bounded above prove that is also bounded above and that
If prove that there exists such that . (Note: If you begin with and solve for then you are assuming that such an exists. You are not asked to find an , you are asked to prove that such an exists.)

Nested Interval Theorem

If and define The set is called an open interval from to . Define also which we call the closed interval from to . The following are called half-open (or half-closed) intervals: If then and . Infinite intervals are Below is a characterization of intervals, we will omit the proof.
Let contain at least two points. Suppose that if and then . Then is an interval.
A sequence of intervals is nested if As an example, consider where . Then is nested: Notice that since for each then Is there another point in ? Suppose that and . Then necessarily . Then there exists such that . Thus, and therefore . Therefore, In general, we have the following.
Let be a sequence of nested closed bounded intervals. Then there exists such that for all , that is, . In particular, if for , and and then
Since is a closed interval, we can write for some and for all . The nested property can be written as Since for all then for all . Therefore, the set is bounded above. Let and thus for all . We will show that for all also. Let be arbitrary. If then and therefore . On the other hand, if then and therefore . In any case, for all . Hence, is an upper bound of , and thus . Since was arbitrary, we have that for all . Therefore, for all , that is . The proof that is similar.
The following theorem gives a condition when contains a single point.
Let be a sequence of nested closed bounded intervals. If then is a singleton set.
Let for .
  1. Prove that is a sequence of nested intervals.
  2. Find .
Using the Nested Interval property of , we give a proof that is uncountable
The real numbers are uncountable.
We will prove that the interval is uncountable. Suppose by contradiction that is countable, and let be an enumeration of (formally this means we have a bijection and ). Since , there exists a closed and bounded interval such that . Next, consider . There exists a closed and bounded interval such that . Next, consider . There exists a closed and bounded interval such that . By induction, there exists a sequence of closed and bounded intervals such that for all . Moreover, by construction the sequence is nested and therefore is non-empty, say it contains . Clearly, since for all then . Now, since for each then . Therefore, for all and thus , which is a contradiction since . Therefore, is uncountable and this implies that is also uncountable.
We now give an alternative proof that is uncountable. To that end, consider the following subset : In other words, consists of numbers whose decimal expansion consists of only 0's and 1's. For example, some elements of are If we can construct a bijection , then since is uncountable then by Example 1.4.7 this would show that is uncountable. Since then this would show that is uncountable (by Theorem 1.4.8). To construct , given in define as In other words, consists of the decimal places in the decimal expansion of that have a value of . For example, It is left as an exercise to show that is a bijection (see Exercise 1.4.4).

Exercises

Let for . Prove that .