In the tool box used to build analysis, if the Completeness property of the real numbers is the hammer then sequences are the nails. Almost everything that can be said in analysis can be, and is, done using sequences. For this reason, the study of sequences will occupy us for the next foreseeable future.

Limits of Sequences

A sequence of real numbers is a function X:NR. Informally, the sequence X can be written as an infinite list of real numbers as X=(x1,x2,x3,), where xn=X(n). Other notations for sequences are (xn) or {xn}n=1; we will use (xn). Some sequences can be written explicitly with a formula such as xn=1n, xn=12n, or xn=(1)ncos(n2+1), or we could be given the first few terms of the sequence, such as X=(3,3.1,3.14,3.141,3.1415,). Some sequences may be given recursively. For example, x1=1,xn+1=xnn+1,n1. Using the definition of xn+1 and the initial value x1 we can in principle find all the terms: x2=12,x3=1/23,x4=1/64, A famous sequence given recursively is the Fibonacci sequence which is defined as x1=1, x2=1, and xn+1=xn1+xn,n2. Then x3=2,x4=3,x5=5, The range of a sequence (xn) is the set {xn|nN}, that is, the usual range of a function. However, the range of a sequence is not the actual sequence (the range is a set and a sequence is a function). For example, if X=(1,2,3,1,2,3,) then the range of X is {1,2,3}. If xn=sin(nπ2) then the range of (xn) is {1,0,1}. Many concepts in analysis can be described using the long-term or limiting behavior of sequences. In calculus, you undoubtedly developed techniques to compute the limit of basic sequences (and hence show convergence) but you might have omitted the rigorous definition of the convergence of a sequence. Perhaps you were told that a given sequence (xn) converges to L if as n the values xn get closer to L. Although this is intuitively sound, we need a more precise way to describe the meaning of the convergence of a sequence. Before we give the precise definition, we will consider an example.
Consider the sequence (xn) whose nth term is given by xn=3n+2n+1. The values of (xn) for several values of n are displayed in Table 3.1.
nxn
12.50000000
22.66666667
32.75000000
42.80000000
52.83333333
502.98039216
1012.99019608
10,0002.99990001
1,000,0002.99999900
2,000,0002.99999950
Values of the sequence
The above data suggests that the values of the sequence become closer and closer to the number . For example, suppose that and consider the -neighborhood of , that is, the interval . Not all the terms of the sequence are in the -neighborhood, however, it seems that all the terms of the sequence from and onward are inside the -neighborhood. In other words, IF then , or equivalently . Suppose now that and thus the new -neighborhood is . Then it is no longer true that for all . However, it seems that all the terms of the sequence from and onward are inside the smaller -neighborhood, in other words, for all . We can extrapolate these findings and make the following hypothesis: For any given there exists a natural number such that if then .
The above example and our analysis motivates the following definition.
The sequence is said to converge if there exists a number such that for any given there exists such that for all . In this case, we say that has limit and we write If is not convergent then we say that it is divergent.
Hence, converges to if for any given (no matter how small), there exists a point in the sequence such that , , , that is, for all . We will sometimes write simply as or .
Using the definition of the limit of a sequence, prove that .
Let be arbitrary but fixed. By the Archimedean property of , there exists such that . Then, if then . Therefore, if then This proves, by definition, that .
Using the definition of the limit of a sequence, prove that .
Given an arbitrary , we want to prove that there exists such that Start by analyzing : Now, the condition that it is equivalent to Now let's write the formal proof. Let be arbitrary and let be such that . Then, . Now if then and thus if then By definition, this proves that .
Using the definition of the limit of a sequence, prove that .
Let . We want to show that for any given , there exists such that if then Start by analyzing : In this case, it is difficult to explicitly isolate for in terms of . Instead we take a different approach; we find an upper bound for : Hence, if then also by the transitivity property of inequalities. The inequality holds true if and only if . Now that we have done a detailed preliminary analysis, we can proceed with the proof. Suppose that is given and let be such that . Then , and thus . Then, if then and therefore This proves that .
Prove that for any irrational number there exists a sequence of rational numbers converging to .
Let be any sequence of positive numbers converging to zero, for example, . Now since for each , then by the Density theorem there exists a rational number such that . In other words, . Now let be arbitrary. Since converges to zero, there exists such that for all , or since , then for all . Therefore, if then . Thus, for arbitrary there exists such that if then . This proves that converges to .
Let where . Prove that .
We want to prove given any there exists such that Now, since for all we have that Thus, if then . Let be arbitrary. Let be such that . Then . Therefore, if then and therefore This proves that .
Does the sequence defined by converge?
A useful tool for proving convergence is the following.
Let be a sequence and let . Let be a sequence of positive numbers such that. Suppose that there exists such that Then .
Let be arbitrary. Since , there exists such that for all . Let . Then, if then and . Thus, if then
Suppose that . Prove that .
We first note that where and since then . Now, by Bernoulli's inequality (Example 1.2.5) it holds that for all and therefore Now since then it follows by Theorem 3.1.9 that
Consider the sequence . Prove that .
We have that Using the definition of the limit of a sequence, one can show that and therefore .
Notice that in the definition of the limit of a sequence, we wrote there exists a number . Could there be more than one number satisfying the definition of convergence of a sequence? Before we go any further, we prove that if a sequence converges then it has a unique limit.
A convergent sequence can have at most one limit.
Suppose that and that . Let be arbitrary. Then there exists such that for all and there exists such that for all . Let . Then for it holds that and also and therefore Hence, for all , and therefore by Theoreom 2.2.7 we conclude that , that is, .
The ultimate long-time behavior of a sequence will not change if we discard a finite number of terms of the sequence. To be precise, suppose that is a sequence and let , that is, is the sequence obtained from by discarding the first terms of . In this case, we will call the -tail of . The next theorem states, not surprisingly, that the convergence properties of and are the same.
Let be a sequence and let be the sequence obtained from by discarding the first terms of , in other words, . Then converges to if and only if converges to .

Exercises

Write the first three terms of the recursively defined sequence , for .
Use the definition of the limit of a sequence to establish the following limits:
\mbox{}
  1. Prove that if and only if .
  2. Combining the previous result and Example 3.1.10, prove that if then .
  3. Conclude that for any real number , if then .
Let and assume that .
  1. Prove that for all .
  2. Use Theorem 3.1.9 to show that .
Note: Do not use Example 3.1.10 to show that .
Suppose that is non-empty and bounded above and let . Show that there exists a sequence such that for all and . Hint: If then clearly . Since there exists such that . Example 3.1.6 is similar.
Let be the sequence defined as Using the definition of the limit of a sequence, find .

Limit Theorems

Proving that a particular number is the limit of a given sequence is usually not easy because there is no systematic way to determine a candidate limit for a given arbitrary sequence. Instead, we are frequently interested in just knowing if a given sequence converges or not, and not so much on finding the actual limit. The theorems in this section help us do just that. We begin with a definition.
A sequence is said to be bounded if there exists such that for all .
Prove that is bounded if and only if there exists numbers and such that for all .
A convergent sequence is bounded.
Suppose that converges to . Then there exists such that for all . Let and we note that . Then for all it holds that . Indeed, if then and if then Thus, for all it holds that and this proves that is bounded.
If then .
Follows by the inequality (see Corollary 2.3.6). Indeed, for any given there exists such that for all and therefore for all .
The following theorem describes how the basic operations of arithmetic preserve convergence.
Suppose that and .
  1. Then and .
  2. Then .
  3. If and then .
(i) By the triangle inequality Let . There exists such that for and there exists such that for . Let . Then for The proof for is similar.
(ii) We have that Now, is bounded because it is convergent, and therefore for all for some . By convergence of and , there exists such that and for all . Therefore, if then (iii) It is enough to prove that and then use (ii). Now, since and then is bounded below by some positive number, say . Indeed, and . Thus, for all . Now, For , there exists such that for all . Therefore, for we have that
Suppose that . Then for any .
The next theorem states that the limit of a convergent sequence of non-negative terms is non-negative.
Suppose that . If for all then .
We prove the contrapositive, that is, we prove that if then there exists such that . Suppose then that . Let be such that . Since , there exists such that , and thus by transitivity we have .
Suppose that and are convergent and suppose that there exists such that for all . Then .
Suppose for now that , that is, for all . Consider the sequence . Then for all and is convergent since it is the difference of convergent sequences. By Theorem 3.2.7, we conclude that . But and therefore , which is the same as . If , then we can apply the theorem to the -tail of the sequences of and and the result follows.
Suppose that and . Then .
We have that . The sequence is constant and converges to . The sequence converges to . Therefore, by the previous theorem, , or .
Suppose that for all . Assume that and also . Then is convergent and .
Let be arbitrary. There exists such that for all and there exists such that for all . Let . Then if then Therefore, for we have that and thus .
Some people call the Squeeze Theorem the Sandwich Theorem; we are not those people.
Let and let . Prove that .
Let be a sequence such that for all and suppose that exists. If then converges and .
Let be such that and set . There exists such that for all . Therefore, for all we have that Thus, , and therefore , and inductively for it holds that Hence, the tail of the sequence given by satisfies Since it follows that and therefore by the Squeeze theorem. This implies that converges to also.

Exercises

Use the Limit Theorems to prove that if converges and converges then converges. Give an example of two sequences and such that both and diverge but converges.
Is the sequence convergent? Explain.
Let and be sequences in . Suppose that and that is bounded. Prove that .
Show that if and are sequences such that and are convergent, then is convergent.
Give examples of the following:
  1. Divergent sequences and such that converges.
  2. Divergent sequences and such that diverges.
  3. A divergent sequence and a convergent sequence such that converges.
  4. A divergent sequence and a convergent sequence such that diverges.
Let and be sequences and suppose that converges to . Assume that for every there exists such that for all . Prove that also converges to .
Let be a sequence and define a sequence as for . Show that if then .
Let be a convergent sequence with limit . Let be a polynomial. Use the Limit Theorems to prove that the sequence defined by is convergent and find the limit of .
Apply the Limit Theorems to find the limits of the following sequences:
Let be a sequence such that for all . Suppose that and . Let Prove that .
Let be a sequence of positive numbers such that . Show that is not bounded and hence is not convergent.

Monotone Sequences

As we have seen, a convergent sequence is necessarily bounded, and it is straightforward to construct examples of sequences that are bounded but not convergent, for example, . In this section, we prove the Monotone Convergence Theorem which says that a bounded sequence whose terms increase (or decrease) must necessarily converge.
Let be a sequence.
  1. We say that is increasing if for all .
  2. We say that is decreasing if for all .
  3. We say that is monotone if is either increasing or decreasing.
Prove that if is increasing then is bounded below. Similarly, prove that if is decreasing then is bounded above.
If is bounded and monotone then is convergent. In particular:
  1. if is bounded above and increasing then
  2. if is bounded below and decreasing then
Suppose that is bounded above and increasing. Let and let be arbitrary. Then by the properties of the supremum, there exists such that . Since is increasing, and is an upper bound for the range of the sequence, it follows that for all . Therefore, for all . Clearly, this implies that for all . Since was arbitrary, this proves that converges to .
Suppose now that is bounded below and decreasing. Let and let be arbitrary. Then by the properties of the infimum, there exists such that . Since is decreasing, and is a lower bound for the range of the sequence, it follows that for all . Therefore, for all . Hence, for all . Since was arbitrary, this proves that converges to .
The Monotone Convergence Theorem (MCT) is an important tool in real analysis and we will use it frequently; notice that it is more-or-less a direct consequence of the Completeness Axiom. In fact, we could have taken as our starting axiom the MCT and then proved the Completeness property of .
By the MCT, a bounded sequence that is also monotone is convergent. However, it is easy to construct a convergent sequence that is not monotone. Provide such an example.
The MCT can be used to show convergence of recursively defined sequences. To see how, suppose that is defined recursively as and where is some given function. For example, say and Hence, in this case . If is bounded and increasing then by the MCT converges, but we do not know what the limit is. However, for example, if is a polynomial/rational function of then we can conclude that must satisfy the equation Indeed, if is a polynomial/rational function then by the Limit Laws we have But and therefore , which is equivalent to since is just the -tail of the sequence . Therefore, as claimed. From the equation we can solve for if possible.
Consider the sequence defined recursively as and Prove that converges and find the limit.
We prove by induction that for all , that is, is bounded below by . First of all, it is clear that . Now assume that for some . Then Hence, is bounded below by . We now prove that is decreasing. We compute that , and thus . Assume now that for some . Then Hence, by induction we have shown that is decreasing. By the MCT, is convergent. Suppose that . Then also and the sequence converges to . Therefore, and thus .
Before we embark on the next example, we recall that and if then and therefore
Consider the sequence Note that this can be defined recursively as and . Prove that converges.
We will prove by the MCT that converges. By induction, one can show that for all . Therefore, Hence, and therefore is bounded. Now, since then clearly . Thus is increasing. By the MCT, converges. You might recognize that .
Let and let for . Prove that converges and find its limit.
Clearly, . Assume by induction that for some . Then Hence, is an increasing sequence. We now prove that is bounded above. Clearly, . Assume that for some . Then . This proves that is bounded above. By the MCT, converges, say to . Moreover, since (as can be proved by induction), then . Therefore, and then . Hence, . Since then .
Consider the sequence . We will show that is bounded and increasing, and therefore by the MCT convergent. The limit of this sequence is the number . From the binomial theorem where we used that for all . This shows that is bounded. Now, for each , we have that And similarly, It is clear that for all . Hence, . Therefore, , that is, is increasing. By the MCT, converges to .

Exercises

Let be an increasing sequence, let be a decreasing sequence, and assume that for all . Prove that and exist, and that . {Note: Recall that a sequence is bounded if there exist constants (independent of ) such that for all .}
Let and let for . Prove that is bounded and monotone. Find the limit of .
Let and let for . Prove that is bounded and monotone. Find the limit of . {Hint: Using induction to prove that is monotone will not work with this sequence. Instead, work with directly to prove that is monotone. }
True or false, a convergent sequence is necessarily monotone? If it is true, prove it. If it is false, give an example.

Bolzano-Weierstrass Theorem

We can gather information about a sequence by studying its subsequences. Loosely speaking, a subsequence of is a new sequence such that each term is from the original sequence and the term appears to the right of the term in the original sequence . Let us be precise about what we mean to the right.
Let be a sequence. A subsequence of is a sequence of the form where is a sequence of strictly increasing natural numbers. A subsequence of will be denoted by .
The notation of a subsequence indicates that the indexing variable is . The selection of the elements of to form a subsequence does not need to follow any particular well-defined pattern but only that . Notice that for any increasing sequence of natural numbers, we have for all .
An example of a subsequence of is the sequence . Here we have chosen the odd terms of the sequence to create . In other words, if we write that then . Another example of a subsequence of is obtained by taking the even terms to get the subsequence , so that here . In general, we can take any increasing selection , such as to form a subsequence of .
Two subsequences of and . Both of these subsequences converge to distinct limits.
We proved that is countable and hence there is a bijection . The bijection defines a sequence where . Let be arbitrary. By the density of in , there exists such that . Now consider the interval . It has infinitely many distinct rational numbers (by the Density Theorem). Therefore, there exists such that . Consider now the interval . It has infinitely many rational numbers, and therefore there exists such that . By induction, there exists a subsequence of such that for all . Therefore, . We proved the following: For any real number there exists a sequence of rational numbers that converges to .
The following theorem is a necessary condition for convergence.
If then every subsequence of converges to .
Let . Then there exists such that for all . Since and , then for all .
The contrapositive of the previous theorem is worth stating.
Let be a sequence.
  1. If has two subsequences converging to distinct limits then is divergent.
  2. If has a subsequence that diverges then diverges.
The following is a very neat result that will supply us with a very short proof of the main result of this section, namely, the Bolzano-Weierstrass Theorem.
Every sequence has a monotone subsequence.
Let be an arbitrary sequence. We will say that the term is a peak if for all . In other words, is a peak if it is an upper bound of all the terms coming after it. There are two possible cases for , either it has an infinite number of peaks or it has a finite number of peaks. Suppose that it has an infinite number of peaks, say , and we may assume that . Then, , and therefore is a decreasing sequence. Now suppose that there are only a finite number of peaks and that is the last peak. Then is not a peak and therefore there exists such that . Similarly, is not a peak and therefore there exists such that . Hence, by induction, there exists a subsequence that is increasing.
Every bounded sequence contains a convergent subsequence.
Let be an arbitrary bounded sequence. By Theorem 3.4.7, has a monotone subsequence . Since is bounded then so is . By the MCT applied to we conclude that is convergent.
We will give a second proof of the Bolzano-Weierstrass Theorem that is more hands-on.
If is a bounded sequence, then there exists such that for all . We will apply a recursive bisection algorithm to hunt down a converging subsequence of . Let be the mid-point of the interval . Then at least one of the subsets or is infinite; if it is then choose some and let and ; otherwise choose some and let , . In any case, it is clear that , that and that . Now let be the mid-point of the interval and let and let . If is infinite then choose some and let , ; otherwise choose some and let and . In any case, it is clear that , that and . By induction, there exists sequences , , and such that and , is increasing and is decreasing. It is clear that and for all . Hence, by the MCT, and are convergent. Moreover, since then and consequently . By the Squeeze theorem we conclude that .
Notice that the proofs of the Bolzano-Weierstrass Theorem rely on the Monotone Convergence Theorem and the latter relies on the Completeness Axiom. We therefore have the following chain of implications: It turns out that if we had taken as our starting axiom the Bolzano-Weierstrass theorem then we could prove the Completeness property and then of course the MCT. In other words, all three statements are equivalent:

Exercises

Prove that the following sequences are divergent.
(Hint: Theorem 3.4.6)
Suppose that for all and suppose that exists. Prove that and that also . (Hint: Consider subsequences of .)
Let be a sequence.
  1. Suppose that is increasing. Prove that if has a subsequence that is bounded above then is also bounded above.
  2. Suppose that is decreasing. Prove that if has a subsequence that is bounded below then is also bounded below.
True or false: If is bounded and diverges then has two subsequences that converge to distinct limits. Explain.
Give an example of a sequence with the following property: For each number there exists a subsequence such that . Hint: If you are spending a lot of time on this question then you have not been reading this textbook carefully.
Suppose that and satisfy and define for . Prove that and are convergent and that . (Hint: First show that for any .)
Let be a sequence and define the sequence as Prove that if then .

limsup and liminf

The behavior of a convergent sequence is easy to understand. Indeed, if then eventually the terms of will be arbitrarily close to for sufficiently large. What else is there to say? In this section, we focus on bounded sequences that do not necessarily converge. The idea is that we would like to develop a limit concept for these sequences, and in particular, a limiting upper bound. Let be an arbitrary sequence and introduce the set defined as the set of all the limits of convergent subsequences of , that is, We will call the subsequences limit set of .
If is bounded and is the subsequences limit set of explain why is non-empty.
Here are six examples of sequences and the corresponding subsequences limit set.
enumeration of
Limits of subsequences
Notice that in the cases where is bounded, the set is also bounded, which is as expected since if then for any convergent subsequence of we necessarily have and therefore .
In general, we have seen that for a general set , and are not necessarily in . This, however, is not the case for the subsequences limit set.
Let be a bounded sequence and let be its subsequences limit set. Then and . In other words, there exists a subsequence of such that and similarly there exists a subsequence of such that .
Let . If then there exists such that . Since , there exists a subsequence of that converges to . Therefore, there exists such that for all . Hence, for each , the inequality holds for infinitely many . Consider . Then there exists such that . Now take . Since holds for infinitely many , there exists such that and . By induction, for , there exists such that and . Hence, the subsequence satisfies and therefore . Therefore, .
By Lemma 3.5.3, if is a bounded sequence then there exists a convergent subsequence of whose limit is larger than any other limit of a convergent subsequence of . This leads to the following definition.
Let be a bounded sequence and let be its subsequences limit set. We define the limit superior of as and the limit inferior of as
By Lemma 3.5.3, is simply the largest limit of all convergent subsequences of while is the smallest limit of all convergent subsequences of . Notice that by definition it is clear that . The next theorem gives an alternative characterization of and . The idea is that is a sort of limiting supremum and is a sort of limiting infimum of a bounded sequence .
Let be a bounded sequence and let . The following are equivalent:
  1. If then there are at most finitely many such that and infinitely many such that .
  2. Let . Then .
(i)(ii) Let denote the subsequences limit set of . By definition, and by Lemma 3.5.3 we have that . Hence, there exists a subsequence of converging to and thus holds for infinitely many . In particular holds for infinitely many . Suppose that holds infinitely often. Now for all and some . Since the inequality holds infinitely often, there exists a sequence such that for all . We can assume that is convergent (because it is bounded and we can pass to a subsequence by the MCT) and thus . Hence we have proved that the subsequence converges to a number greater than which contradicts the definition of .
(ii)(iii) Let . Since holds for finitely many , there exists such that for all . Hence, is an upper bound of and thus . Since is decreasing, we have that for all . Now, holds infinitely often and thus for all . Hence, for all . This proves the claim.
(iii)(i) Let be a convergent subsequence. Since , by definition of , we have that . Therefore, . Hence, is an upper bound of . By definition of , there exists such that . By induction, there exists a subsequence such that . Hence, by the Squeeze Theorem, . Hence, and thus .
Let denote the th prime number, that is , , , and so on. The numbers and are called twin primes if . The Twin Prime Conjecture is that In other words, the Twin Prime Conjecture is that there are infinitely many pairs of twin primes.
We end this section with the following interesting theorem that says that if the subsequences limit set of a bounded sequence consists of a single number then the sequence also converges to .
Let be a bounded sequence and let . If every convergent subsequence of converges to then converges to .
Suppose that does not converge to . Then, there exists such that for every there exists such that . Let . Then there exists such that . Then there exists such that . By induction, there exists a subsequence of such that for all . Now is bounded and therefore by Bolzano-Weierstrass has a convergent subsequence, say , which is also a subsequence of . By assumption, converges to , which contradicts that for all .
Another way to say Theorem 3.5.7 is that if is bounded and then . The converse, by the way, has already been proved: if then every subsequence of converges to and therefore .

Exercises

Determine the and for each case:
Let and be bounded sequences. Let be the sequence . Show that In other words, prove that
Let and be bounded sequences. Show that if for all then

Cauchy Sequences

Up until now, the Monotone Convergence theorem is our main tool for determining that a sequence converges without actually knowing what the the limit is. It is a general sufficient condition for convergence. In this section, we prove another groundbreaking general sufficient condition for convergence known as the Cauchy criterion. Roughly speaking, the idea is that if the terms of a sequence become closer and closer to one another as then the sequence ought to converge. A sequence whose terms become closer and closer to one another is called a Cauchy sequence.
A sequence is said to be a Cauchy sequence if for every there exists a natural number such that if then .
In other words, is a Cauchy sequence if the difference is arbitrarily small provided that both and are sufficiently large.
Prove directly using the definition of a Cauchy sequence that if and are Cauchy sequences then the sequence is a Cauchy sequence.
Not surprisingly, a convergent sequence is indeed a Cauchy sequence.
If is convergent then it is a Cauchy sequence.
Suppose that . Let and let be sufficiently large so that for all . If then by the triangle inequality, This proves that is Cauchy.
A Cauchy sequence is bounded.
If is a Cauchy sequence then is bounded.
The proof is similar to the proof that a convergent sequence is bounded.
The sequence is convergent if and only if is a Cauchy sequence.
In Lemma 3.6.3 we already showed that if converges then it is a Cauchy sequence. To prove the converse, suppose that is a Cauchy sequence. By Lemma 3.6.4, is bounded. Therefore, by the Bolzano-Weierstrass theorem there is a subsequence of that converges, say it converges to . We will prove that also converges to . Let be arbitrary. Since is Cauchy there exists such that if then . On the other hand, since there exists such that . Therefore, if then This proves that converges to .
Let and suppose that for all . Using the fact that for all prove that if then . Deduce that is a Cauchy sequence.
When the MCT is not applicable, the Cauchy criterion is another possible tool to show convergence of a sequence.
Consider the sequence defined by , , and for . One can show that is not monotone and therefore the MCT is not applicable.
  1. Prove that for all .
  2. Prove that for all .
  3. Prove that if then Hint: Use part (b) and the Triangle inequality.
  4. Deduce that is a Cauchy sequence and thus convergent.
  5. Show by induction that and deduce that .
Notice that the main result used in the Cauchy Criterion is the Bolzano--Weierstrass (B--W) theorem. We therefore have the following chain of implications: A close inspection of the Cauchy Criterion reveals that it is really a statement about the real numbers not having any gaps or holes. In fact, the same can be said about the MCT and the Bolzano-Weierstrass theorem. Regarding the Cauchy Criterion, if is a Cauchy sequence then the terms of are clustering around a number and that number must be in if has no holes. It is natural to ask then if we could have used the Cauchy Criterion as our starting axiom (instead of the Completeness Axiom) and then prove the Completeness property, and then the MCT and the Bolzano-Weierstrass theorem. Unfortunately, the Cauchy Criterion is not enough and we also need to take as an axiom the Archimedean Property.
Suppose that every Cauchy sequence in converges to a number in and the Archimedean Property holds in . Then satisfies the Completeness property, that is, every non-empty bounded above subset of has a least upper bound in .
Let be a non-empty set that is bounded above. If is an upper bound of and then is the least upper bound of and since there is nothing to prove. Suppose then that no upper bound of is an element of . Let be arbitrary and let be an upper bound of . Then , and we set . Consider the mid-point of the interval . If is an upper bound of then set and set , otherwise set and . In any case, we have , , and . Now consider the mid-point of the interval . If is an upper bound of then set and , otherwise set and . In any case, we have , , and . By induction, there exists a sequence such that is not an upper bound of and a sequence such that is an upper bound of , and , , and . By Exercise 3.6.6, and using the fact that if (this is where the Archimedean property is needed), it follows that and are Cauchy sequences and therefore by assumption both and are convergent. Since it follows that . We claim that is the least upper bound of . First of all, for fixed we have that for all and therefore , that is, is an upper bound of . Since is not an upper bound of , there exists such that and therefore by the Squeeze theorem we have . Given an arbitrary then there exists such that and thus is not an upper bound of . This proves that is the least upper bound of .

Exercises

Show that if is a Cauchy sequence then is bounded. (Note: Do not use the fact that a Cauchy sequence converges but show directly that if is Cauchy then is bounded.)
Show that if converges then it is a Cauchy sequence.
Show by definition that is a Cauchy sequence.
Show by definition that is a Cauchy sequence.
Suppose that and are sequences such that for all . Show that if the sequence is convergent then so is the sequence .
Suppose that . Show that if the sequence satisfies for all then is a Cauchy sequence and therefore convergent. Hint: If then Also, if then .

Infinite Series

Informally speaking, a series is an infinite sum: Using summation notation: The series can be thought of as the sum of the sequence . It is of course not possible to actually sum an infinite number of terms and so we need a precise way to talk about what it means for a series to have a finite value. Take for instance so that the sequence being summed is . Let's compute the first 10 terms of the sequence of partial sums defined as follows: With the help of a computer, one can compute It seems as though the sequence is converging to , that is, . It is then reasonable to say that the infinite series sums or converges to We now introduce some definitions to formalize our example.
Let be a sequence. The infinite series generated by is the sequence defined by or recursively, The sequence is also called the sequence of partials sums generated by . The th term of the sequence of partial sums can instead be written using summation notation:
Let be the sequence . The first few terms of the sequence of partials is
In both examples above, we make the following important observation: if is a sequence of non-negative terms then the sequence of partials sums is increasing. Indeed, if then and thus .
Find the sequence of partial sums generated by .
We compute: Hence, .
The limit of the sequence , if it exists, makes precise what it means for an infinite series to converge.
Let be a sequence and let be the sequence of partial sums generated by . If exists and equals then we say that the series generated by converges to and we write that
The notation is therefore a compact way of writing and the question of whether the series converges is really about whether the limit exists. Often we will write a series such as simply as when either the initial value of is understood or is unimportant. Sometimes, the initial value may be , , or some other .
The geometric series is perhaps the most important series we will encounter. Let where is a constant. The generated series is and is called the geometric series. The th term of the sequence of partial sums is If then does not exist (why>, so suppose that . Using the fact that we can write Now if then , while if then does not exist. Therefore, if then Therefore, In summary: The series is called the geometric series and converges if and only if and in this case converges to .
Consider the series . The series can be written as and thus it is a geometric series with . Since , the series converges and it converges to
Use the geometric series to show that
We can write
Consider . Using partial fraction decomposition Therefore, the th term of the sequence of partial sums is This is a telescoping sum because all terms in the middle cancel and only the first and last remain. For example: By induction one can show that and therefore . Therefore, the given series converges and it converges to .
Consider the series . We are going to analyze a subsequence of the sequence of partial sums . We will show that is unbounded and thus is unbounded and therefore divergent. Consequently, the series is divergent. Consider Now consider Lastly consider, In general, one an show by induction that for we have Therefore, the subsequence is unbounded and thus the series is divergent.
We now present some basic theorems on the convergence of series; most of them are a direct consequence of results from limit theorems for sequences. The first theorem we present can be used to show that a series diverges.
If converges then . Equivalently, if then diverges.
By definition, if converges then the sequence of partial sums is convergent. Suppose then that . Recall that has the recursive definition . The sequences and both converge to and thus
The series is divergent because .
The Series Divergence Test can only be used to show that a series diverges. For example, consider the harmonic series . Clearly . However, we already know that is a divergent series. Hence, in general, the condition is not sufficient to establish convergence of the series .
A certain series has sequence of partial sums whose general th term is .
  1. What is ?
  2. Does the series converge? If yes, what does it converge to? Explain.
  3. Does the sequence converge? If yes, what does it converge to? Explain.
The next theorem is just an application of the Cauchy criterion for convergence of sequences to the sequence of partial sums .
The series converges if and only if for every there exists such that if then
The following theorem is very useful and is a direct application of the Monotone convergence theorem.
Suppose that is a sequence of non-negative terms, that is, for all . Then converges if and only if is bounded. In this case,
Clearly, if exists then is bounded. Now suppose that is bounded. Since for all then and thus shows that is an increasing sequence. By the Monotone convergence theorem, converges and .
Consider the series . Since , to prove that the series converges it is enough to show that the sequence of partial sums is bounded. We will consider a subsequence of , namely, where for . We have and By induction, one can show that and therefore using the geometric series with we have This shows that the subsequence is bounded. In general, the existence of a bounded subsequence does not imply that the original sequence is bounded but if the original sequence is increasing then it does. In this case, is indeed increasing, and thus since is a bounded subsequence then is also bounded. Therefore, converges, that is, is a convergent series.
Suppose that and are convergent series.
  1. Then and are also convergent and
  2. For any constant , is also convergent and .
Using the limit laws: Hence, If is a constant then by the Limit Laws, Therefore,
Once establishing the convergence/divergence of some sequences, we can use comparison tests to the determine convergence/divergence properties of new sequences.
Let and be non-negative sequences and suppose that for all .
  1. If converges then converges.
  2. If diverges then diverges.
Let and be the sequences of partial sums. Since then . To prove (a), if converges then is bounded. Thus, is also bounded. Since is increasing and bounded it is convergent by the MCT. To prove (b), if diverges then is necessarily unbounded and thus is also unbounded and therefore is divergent.
Let be an integer and let be a sequence of integers such that . Use the comparison test to show that the series converges and converges to a point in .
Determine whether the given series converge.
  1. : First compute . Therefore, by the divergence test, the series diverges.
  2. where : We know that is convergent. Now, if then . Therefore by the Comparison test, is also convergent. This is called a -series and actually converges for any .
  3. : We will use the comparison test. We have The series converges and thus the original series also converges.
Suppose that for all . If converges prove that also converges. Does the claim hold if we only assume that ?
Since then . Since converges then by the Comparison test then also converges. More generally, suppose that and converges. Then converges to zero and thus there exists such that for all . Then since the series converges it follows that converges and consequently converges.
The following two tests can be used for series whose terms are not necessarily non-negative.
If the series converges then converges.
From we obtain that If converges then so does . By the Comparison test 3.7.18, the series converges also. Then the following series is a difference of two converging series and therefore converges: and the proof is complete.
In the case that converges we say that converges absolutely. We end the section with the Ratio test for series.
Consider the series and let . If then the series converges absolutely and if then the series diverges. If or the limit does not exist then the test is inconclusive.
Suppose that . Let be such that . There exists such that for all and thus for all . By induction, it follows that for all . Since is a geometric series with then, by the comparison test, the series converges. Therefore, the series converges and by the Absolute convergence criterion we conclude that converges absolutely. If then a similar argument shows that diverges. The case follows from the fact that some series converge and some diverge when .

Exercises

Suppose that is a convergent series. Is it true that if is divergent then is divergent? If it is true, prove it, otherwise give an example to show that it is not true.
Suppose that for all . Prove that if converges then also converges. Is the converse true? That is, if converges then does it necessarily follow that converges?
Using only the tests derived in this section, determine whether the given series converge or diverge:
Suppose that and are non-negative sequences. Prove that if and are convergent then is convergent. (Hint: Recall that if then converges iff is bounded, where is the sequence of partial sums. Alternatively, use the identity .)
Suppose that for all and converges. You will prove that converges.
  1. Let . Prove that converges.
  2. Using the fact that , prove that if . Deduce that
  3. Deduce that converges.
Any number of the form can be written as Using this fact, and a geometric series, prove that
Show that the following series converge and find their sum.
Using the fact that for all , prove that the series converges.
Let be a decreasing sequence of non-negative terms. Let be the sequence of partial sums of the series , let for , and consider the subsequence .
  1. Show by induction that for all .
  2. Conclude that if converges then .
(Note: This is a generalization of Example 3.7.16.)