In this chapter, we study another notion of convergence that is surely familiar to the reader, namely, the limit of a function at a given point. After introducing the precise definition of the limit of a function, and working through some examples, we will relate limits of functions with limits of sequences resulting in the Sequential Criterion for Limits (Theorem 4.1.11). In this chapter, when not explicitly stated, the letter A will denote a subset of R.

Limits of Functions

Before we can give the definition of the limit of a function, we need the notion of a cluster point of a set.
A number cR is called a cluster point of A if for any given δ>0 there exists at least one point xA, with xc, such that |xc|<δ.
Hence, c is a cluster point of A if there are points in A that are arbitrarily close to c. In general, a cluster point of A is not necessarily an element of A. Naturally, cluster points can be characterized using limits of sequences.
A point c is a cluster point of A if and only if there exists a sequence (xn) in A such that xnc and limnxn=c.
Let c be a cluster point of A and let δn=1n for nN. Then by definition of a cluster point, there exists xnA, xnc, such that |xnc|<δn. Since δn0 then xnc. To prove the converse, suppose that xnc, xnc and xnA for nN. Then by convergence of (xn) to c, for any δ>0 there exists KN such that |xKc|<δ. Since x=xKA, this proves that c is a cluster point of A.
Below are some examples of cluster points for a given set:
  • Consider the set A=[0,1]. Every point cA is a cluster point of A.
  • On the other hand, for A=(0,1], the point c=0 is a cluster point of A but does not belong to A.
  • For A={1n|nN}, the only cluster point of A is c=0.
  • A finite set does not have any cluster points.
  • The set A=N has no cluster points.
  • Consider the set A=Q[0,1]. By the Density theorem, every point c[0,1] is a cluster point of A.
We now give the definition of the limit of a function f:AR at a cluster point c of A.
Consider a function f:AR and let c be a cluster point of A. We say that f has a limit at c, or converges at c, if there exists a number LR such that for any given ε>0 there exists δ>0 such that if xA and 0<|xc|<δ then |f(x)L|<ε. In this case, we write that limxcf(x)=L and we say that f converges to L at c, or that f has limit L at c. If f does not converge at c then we say that f diverges at c.
Another short-hand notation to denote that converges to at is as . By definition, if , then for any there exists a such that for all not equal to it holds that .
A function can have at most one limit at .
Suppose that and as , and let . Then there exists such that and , for all satisfying . Then if then Since is arbitrary, Theorem 2.2.7 implies that .
Consider the function with domain . Prove that
We begin by analyzing the quantity : Hence, if then Thus, given we let and thus if then . Thus, by definition, .
Consider the function with domain . Prove that
We have that Let be arbitrary and let . Then if then . Hence, if then Thus, by definition, .
Consider the function with domain . Prove that for any ,
We first note that By the triangle inequality, and therefore We now need to analyze how large can become when is say within of . To be concrete, suppose that . Hence, if then Therefore, if it holds that Now suppose that is arbitrary and let . Then if then and therefore This proves, by definition, that for any .
Consider the function with domain . Prove that
We first note that is indeed a cluster point of . Now, We now obtain a bound for when is close to . Suppose then that . Then and therefore, , which implies that . Similarly, if then and therefore , which implies that . Therefore, if then Suppose now that is arbitrary and let . If then from our analysis above it follows that . Therefore, if then This proves that .
Consider the function defined as Prove that .
If then and if then Therefore, for all it holds that since . Thus, given let and thus if then This proves that .
The following important result states that limits of functions can be studied using limits of sequences.
Let be a function and let be a cluster point of . Then if and only if for every sequence in converging to (with for all ) the sequence converges to .
Suppose that . Let be a sequence in converging to , with for all . We must prove that the sequence converges to . To that end, let be arbitrary. Then, by convergence of to at , there exists such that if then . Now, since , there exists such that for all . Therefore, for we have that . This proves that . To prove the converse, we prove the contrapositive. Hence, we must show that if does not converge to then there exists a sequence in (with ) converging to but the sequence does not converge to . Assume then that does not converge to . Then, negating the definition of the limit of a function, there exists such for all there exists such that and . Then, let for . Then there exists such that and . Since then but clearly does not converge to . This ends the proof.
The following theorem follows immediately from Theorem 4.1.11.
Let be a function and let be a cluster point of and let . Then does not converge to at if and only if there exists a sequence in converging , with for all , and such that does not converge to .
Note that in Corollary 4.1.12, if the sequence diverges then by definition it does not converge to any and then does not have a limit at . When applicable, the following corollary is a useful tool to prove that a limit of a function does not exist.
Let be a function and let be a cluster point of . Suppose that and are sequences in converging to , with and for all . If and converge but then does not have a limit at .
Prove that does not exist.
Consider , which clearly converges to and for all . Then which is unbounded and thus does not converge. Thus, by Corollary 4.1.12, does not exist.
Prove that does not exist.
Let with domain . Consider the sequence . It is clear that and for all . Now and therefore does not converge. Therefore, has no limit at . In fact, for each , consider the sequence . Clearly and for all . Now, . Hence, converges to . This shows that oscillates within the interval as approaches .
The sign function, denoted by , is defined as Prove that does not exist.
Consider the sequence . Then and for all . Now, and thus does not converge. Therefore, by Corollary 4.1.12, the function has no limit at .

Exercises

Use the definition of the limit of a function to prove that the following limits do indeed hold.
Let , let , and suppose that is a cluster point of . Suppose that there exists a constant such that for all . Prove that .
Consider the function Prove that .
Let be defined as follows:
  1. Prove that has a limit at .
  2. Now suppose that . Prove that has no limit at .
  3. Define by . Prove that has a limit at any .
Hint: The Density Theorem will be helpful for (b). In particular, the Density Theorem implies that for any point , there exists a sequence of rational numbers such that , and that there exists a sequence of irrational numbers such that .
Use any applicable theorem to explain why the following limits do not exist.
Recall that the function is defined as follows:

Limit Theorems

In this section, we establish basic limit theorems for limits of functions. The reader should compare the results of this section with Section 3.2 where we established limit theorems for sequences. In fact, thanks to the sequential criterion for limits of functions (Theorem 4.1.11), all of the theorems in this section can be proved using limits of sequences. To begin, we first show that if has a limit at then satisfies a local boundedness property at . Let us first define then what it means for a function to be locally bounded at a given point.
Consider a function and let be a cluster point of . We say that is bounded locally at if there exists and such that if then .
Consider a function and let be a cluster point of . If exists then is bounded locally at .
Let and let be arbitrary. Then there exists such that for all such that . Therefore, for all and we have that If then let and if then let . Then for all such that , that is, is bounded locally at .
Consider the function defined on the set . Clearly, is a cluster point of . For any and any let be such that . Then , that is, . Since was arbitrary, this proves that is unbounded at and consequently does not have a limit at .
We now state and prove some limit laws for functions. Let be functions and define the functions , , , and on as follows: where for we require that for all .
Let be functions and let be a cluster point of . Suppose that and . Then
  1. , if
The proofs are left as an exercises. (To prove the results, use the sequential criterion for limits and the limits laws for sequences).
Let be functions and let be a cluster point of . If exists for each then
If is a polynomial function then for every . If is another polynomial function and in a neighborhood of and then
Prove that .
We cannot use the Limit Laws directly since . Instead, notice that if then . Hence, the functions and are equal at every point in . It is clear that and therefore it follows that also .
Let be a function and let be a cluster point of . Suppose that has limit at . If for all then .
We prove the contrapositive. Suppose then that . Let be such that . Then since , there exists such that if then . Hence, for some . We give another proof using the sequential criterion for limits. To that end, if converges to at then for any sequence converging to , , we have that . Now and therefore from our results on limits of sequences (Theorem 3.2.7).
Let be a function and let be a cluster point of . Suppose that for all and suppose that . Then .
We have that and therefore by Theorem 4.2.8 we have that . Similarly, from we deduce that . From this we conclude that . An alternative proof: Since at , for any sequence with , we have that . Clearly, and therefore from our results on limits of sequences (Theorem 3.2.7).
The following is the Squeeze Theorem for functions.
Let be functions and let be a cluster point of . Suppose that and . If for all , , then .
Let be a sequence in converging to with for all . Then, by the sequential criterion, By assumption, it holds that for all , and therefore by the Squeeze Theorem for sequences, we have that . This holds for every such sequence and therefore .
Let Show that .
We end this section with the following theorem.
Let be a function and let be a cluster point of . Suppose that . If then there exists such that for all , .
Choose so that , take for example . Then there exists such that for all , , and thus by transitivity it follows that for all , .

Exercises

Let and suppose that is a cluster point of . Suppose that at , converges to and converges to . Prove that converges to at in two ways: (1) using the definition of the limit of a function, and (2) using the sequential criterion for limits.
Give an example of a set , a cluster point of , and two functions such that exists but does not exist.
Give an example of a function that is bounded locally at but does not have a limit at . Your answer should not be in the form of a graph.
Let be a function that is bounded locally at and suppose that converges to at . Prove that .