What can we say about functions that grow beyond all bounds?

In this section, we learn:


  limxa f(x)=+∞

Consider the function f defined by the equation

f(x)=1(x1)2

The graph of this function is illustrated in Figure 1.

Figure 1: Graph of y=1(x1)2

Note that f is not defined at x=1 (division by zero is undefined), but let\textquoteright s consider the values of f when x is close to 1. Letting x approach 1 from both sides, the corresponding values of f are given in the following table.

From this table, we see that as x gets closer and closer to 1 but never quite equal to 1, f(x) becomes indefinitely large; in other words, we can make f(x) as large as we desire if we take x close enough to 1.

To express that f(x) increases without bound as x approaches 1, we write

limx11(x1)2=+

  • Instead of +, we may simply write .

 

  • The equals sign does not mean that the limit exists. Note that is not a number; it is just a symbol for indicating that f(x) indefinitely increases. 

Description 1: Let f be a function defined on both sides of a, except possibly at a itself. If we can make f(x) as large as we wish by taking x sufficiently close to a but not equal to a, then we say f(x) approaches as x approaches a (or the limit of f(x), as x approaches a, is positive infinity and write

limxaf(x)=+


Show the Precise Definition of limxa f(x)=+∞


 

limxa f(x)=-∞

Now consider the function g defined by the equation

g(x)=1(x1)2.

The graph of g(x) is shown in Figure 3. We notice that g(x)=f(x) and as x approaches 1 from either side, g(x) decreases without bound. In this case we write

limx1g(x)=.

Figure 3: Graph of y=1(x1)2

Description 2: Let g be a function defined on both sides of a, except possibly at a itself. If we can make g(x) as large negative as we wish by taking x sufficiently close to a but not equal to a, then we say g(x) approaches as x approaches a (or the limit of g(x) is as x approaches a is infinity and write

limxag(x)=.


Show the Precise Definition of limxa f(x)=-∞


 

One-Sided “Infinite” Limits

One-sided limits can be also defined accordingly. For example, consider the function h(x) defined by the equation

h(x)=1x2.

Here we can assign all values to x except 2 as the denominator becomes zero when x=2 . If x<2, h(x) is negative and if x>2, h(x) is positive. The graph of this function is represented in Figure 4.

As x approaches 2 from the left (through the values that are less than 2), h(x) is negative and decreases without bound; while as x approaches 2 from right (through values that are larger than 2), h(x) increases without bound. In this case, we write

limx2h(x)=,andlimx2+h(x)=+.

Figure 4: Graph of h(x)=1x2. As we can see here limx2+h(x)=+ and limx2h(x)=.

 

Example 1

Use the precise definition to prove that

limx1+2x+1=+.

Solution

Suppose K>0 is given. We must show that there is a δ>0 such that

2x+1>Kwhenever1<x<1+δ.

Because K>0 and x+1>0 when x>1 [Recall that x1+ means x is approaching 1 through values greater than 1], we have

x+12<1K

[Recall property 7 in the Section on Inequalities] or

x+1<2K

or

x<1+2K.

So if we have two restrictions on x: (1) x>1 that guarantees x+1>0 and (2) x<1+2/K which are equivalent to

1<x<1+2K,

then

2x+1>K.

That is, if we choose δ2/K then

1<x<1+δ2x+1>K.

Let's show that this δ works. If K>0 is given, let δ=2/K. If 1<x<1+δ then

0<x+1<δ=2K(Add 1 to each side)

1x+1>K2

[Recall property 7 in the Section on Inequalities. For example, 5>3 but 15<13]

2x+1>K.

Therefore

limx1+2x+1=+.

The graph of y=2x+1 is shown below.

Figure 5: Graph of y=1x+1. As we can see here limx1+y=+ and limx1y=.

 

Logarithmic Functions

Let’s look at the graph of y=lnx (Figure 6); we realize that

limx0+lnx=

But because lnx is not defined for x<0, we cannot talk about limx0lnx (and hence neither the two-sided limit limx0lnx).

Figure 6: Graph of y=lnx. It is clear from this graph thatlimx0+lnx=

In general,

limx0+logbx={if b>1+if b<1.

See Figure 7.

Figure 7: Graph of y=logbx. If b>1, then limx0+logbx= and if b<1, limx0+logbx=+.

 

Algebraic operations

Theorem 1: Let a and L be real numbers. If limxag(x)=0 and limxaf(x)=L0 then

limxaf(x)g(x)={+if L>0 and g(x)>0if L>0 and g(x)<0if L<0 and g(x)>0+if L<0 and g(x)<0

The theorem is also valid for the left and right-hand limits; that is, if we replace xa by xa or xa+.

  • In the above theorem, we talk about the sign of g(x) for all x close to a (not in its entire domain); it does not matter if the sign of g changes when x is not very close to a.

We may summarize the above theorem and symbolically write

L0=+ or (if L0)

Depending on the sign of L and the sign of g(x) for x close to a, x. Recall

(+)(+)=()()=(+)and()(+)=(+)()=().

Q: We have been always told that division by zero is not defined but here we simply write L/0 is plus or minus infinity. What’s going on?

A: This is just a symbol for memorizing the above theorem. Also, here the denominator is not exactly zero! The denominator gets closer and closer to 0 as x gets closer and closer to a, but not equal to a. If g(x)=0 for all x close to a, then the function h(x)=f(x)/g(x) will not be defined near a and so we cannot talk about the limit of f(x)/g(x) when x approaches a.


Show the Proof of Theorem 1


It follows from the above theorem that

limx0+1xn=+

limx01xn={+if n is evenif n is odd

These limits are also clear from the graph of y=1xn (see the following figures).

(a) n is odd. limx01xn= and limx0+1xn=+. (b)n is even. limx01xn=+

Figure 8

Because tanx=sinx/cosx,

limxπ/2sinx=1,limxπ/2+cosx=0

and cosx>0 when π/2<x<π/2, we have

limxπ2tanx=10+=++

and because cosx<0 when π/2<x<3π/2, we have

limxπ2+tanx=10=.

These results are also clear from the graph of y=tanx. As we can see from Figure 9:

limxπ2tanx=+,limxπ2+tanx=

Figure 9: In general limxatanx=+ and limxa+tanx= for a=±π2,±3π2,±5π2,.

Similarly, because secx=1cosx, as x approaches a zero of cosine from one side, the secant goes to plus infinity or minus infinity (See Figure 10).

Figure 10: As x approaches where the cosine function is zero (from one side), the secant function goes to + or . The zeros of cosine occur at ±π2,±3π2,±5π2,.

Example 2

Find 

(a)limx2x22x1x23x+2

(b)limx2+x22x1x23x+2

Solution

(a) The limit of the numerator is 1

limx2(x22x1)=222×21=1

and the limit of the denominator is 0

limx2(x23x+2)=223×2+2=0.

So by Theorem 1, we know the result can be + or depending on the sign of numerator (which is negative 1<0) and the sign of the denominator (which we need to find) as x2.

To determine the sign of the denominator, we construct a sign table:

x23x+2=(x2)(x1)

so there are two roots x=2 and x=1

 

From the above table, we see that the denominator is negative as x approaches 2 from the left. Because the limit of the numerator is 1<0, we obtain

limx2x22x1x23x+2=+.

“Symbolically” we may write

10=+.

As in part (a), the limit of the numerator is 1<0 and the limit of the denominator is 0 but from the above sign table we see that the denominator is approaching zero through positive values. Therefore

limx2+x22x1x23x+2=.

“Symbolically” we may write

10+=.

The graph of y=x22x1x23x+2 is illustrated in the following figure.

Figure 11: Graph of y=x22x1x23x+2.

Example 3

Find 

(a) limx2x2x2

(b) limx2+x2x2

where y=x is the greatest integer function (also called the floor function). The other notation for x is [[x]].

Solution

(a) When 1x<2, x=1. Therefore limx2x=1 and hence limx2(x2)=1. Furthermore, limx2(x2)=22=0 and x2 is approaching 0 through negative numbers

x2<0for x<2.

Therefore, by Theorem 1, we have

limx2x2x2=+.

“Symbolically” we may write

10=+.

(b) When 2x<3, x=2. Therefore as x is close to 2 but larger than 2, x2 is exactly zero but x2 is a small quantity close to zero. If we divide the number zero by any small number (not equal to zero), we get exactly zero. To understand this reasoning better, consider the following table.

x2.0012.012.12.5x20000x20.0010.010.10.5x2x20000 

The graph of y=x2x2 is shown below.

Figure 12: Graph of y=x2x2. From this graph we can see that limx2x2x2= and limx2+x2x2=0.

Example 4

Find
(a) limx3+x29x3

(b) limx39x2x3

Solution

(a) Substituting x=3 into the given expression results in the indeterminate form 0/0. As x approaches 3 through values greater than 3, x3>0. Therefore, we can write x3=(x3)2=x3x3 [Recall that t2=|t|. Here (x3)2=|x3| but x3>0. So |x3|=x3]. Also x29=(x3)(x+3) [Recall A2B2=(AB)(A+B)]. Because here x3>0 and x+3>0, we have (x3)(x+3)=x3x+3. Therefore limx3+x29x3=limx3+x3x+3x3x3=limx3+x+3x3 By direct substitution, the limit of the numerator is 6 and the limit of the denominator is 0. The denominator approaches zero through positive numbers [the square root is always positive]. Therefore by Theorem 1, limx3+x+3x3=[60+]+. The graph of y=x29/(x3) is shown in Figure 12.

Figure 13: Graph of y=x29x3.

(b) Similar to part (a), direct subsitution yields the indeterminate form 0/0. As x3, x3<0. Thus (3x)2=|3x|=3x or x3=3x3x. Therefore limx39x2x3=limx3(3x)(3+x)3x3x=limx33x3+x3x3x=limx33+x3x. By direct subsitution the limit of the numerator is 6, and the limit of the denomiator is zero. The denominator approaches zero through negative values. Therefore, by Theorem 1limx33+x3x=[60]. The graph of y=9x2/(x3) is shown in Figure 13.

Figure 14: Graph of y=9x2x3.

Because + and are not numbers, the Limit Laws do not apply for infinite limits. If we have infinite limits we can use the following theorem instead.

Theorem 2: (a) If limxaf(x)=L and limxag(x)=+, then limxa[f(x)+g(x)]=+ and limxa[f(x)g(x)]={+if L>0if L<0 (b) If limxaf(x)=L and limxag(x)=, then limxa[f(x)+g(x)]= and limxa[f(x)g(x)]={if L>0+if L<0 (c) If limxaf(x)=+ and limxag(x)=+, then limxa[f(x)+g(x)]=+ and limxa[f(x)g(x)]=+ (d) If limxaf(x)= and limxag(x)=, then limxa[f(x)+g(x)]= and limxa[f(x)g(x)]=+

The theorem is valid if we replace xa by xa+ or xa.

We may summarize the above theorem and say

(1) L+(±)=± (2)

L(0)×(±)={±if L>0if L<0 (3) (±)(±)=+,(±)()= (4) +=+,=.

For example, because limx03=3 and limx01x2=+, it follows from the above theorem that limx0(3+1x2)=3+=+ and limx0(3x2)=3(+)=.

Indeterminate Forms 0·∞ and ∞ – ∞

  • Note that Theorem 2 does not say anything about 0(±) or : 0(±)=?or=? In fact, we say the cases 0(±) and are “indeterminate limits.” Let’s look at them this way: We know L0=+ or  So this implies that 0(±) can be any number L, and we need to evaluate each limit case by case. On the other hand because L+(±)=± we may conclude that ±(±) can have any value.

  • Examples showing that an indeterminate limit may take any value

  • Because limits of the form 0(±) or may take any value that we cannot predict in advance, we call them indeterminate.

Because L/0=+ or , you may think that L/(±)=0. Although L/0=± is just a symbol and hence does not imply L/(±)=0, the following theorem tells us this conclusion is valid.

Theorem 3: If limxaf(x)=L and limxag(x)=+ or limxag(x)= then limxaf(x)g(x)=0.

Symbolically we may express the above theorem as 

L±=0.

Any number divided by a very large (positive or negative) number becomes approximately zero. For example, 2 divided by 1,000,000 is 0.0000020 and 7 divided by 100,000 is 0.000070.

  • Note that is not a number, and L/=0 is just a symbol to memorize the above theorem. Also the following expression helps us remember the two above theorems. L0=+ or L±=0.

  • On the left expression, L must be nonzero because 0/0 is an indeterminate form (see Section [sec:Ch4-Indeterminate-0/0]) but on the right expression L can be zero. Limits of the form 0/ are not indeterminate; roughly speaking 0=01=00=0.

  • On the right, we may replace L with + or and the result is still value, because roughly speacking (±)0=(±)10 1/0 is + or , and infinity multiplied by infinity is infinity. On the other hand, on the right hand side, we cannot replace L with infinity.

Indeterminate Form ∞ / ∞

Theorem 3 does not tell us anything about ±/±. This is another indeterminate form ±±=? In fact, we saw L(±)=+ or  so (±)/(±) can be any number L.

  • We will learn how to deal with limits of the indeterminate forms 0(±),, and (±)/(±) in Section 4.10. Finding limits of the indeterminate form 0/0 was discussed in Section 4.5.

Example 5

Find
(a) limx2+ln(x2)2x

(b) limx2ln(x2)2x

(c) limx2+2xln(x2)

(d) limx22xln(x2).

Solution

(a) Recall the graph of y=lnx (Figure [6]). Now if we shift it 2 units to the right, we obtain the graph of y=ln(x2). As x approaches 2 through values greater than 2 (x2+), ln(x2) is large and negative. That is, limx2+ln(x2)=. The function y=2x or y=x+2 is a polynomial and thus we can simply substitute 2 for x in its expression limx2+(2x)=22=0. However, when x is greater than 2, 2x approaches 0 through negative values x>22x<0. Therefore limx2+12x=[10] Finally limx2+ln(x2)2x=(limx2+lnx)(limx2+12x)=()()=+. The graph of y=ln(x2)/(2x) is shown in Figure 15.

Figure 15 

(b) Because ln(x2) is not defined for x<2, x cannot approach 2 through values less than 2 and the left-hand limit does not exist.

(c) In part of (a), we saw limx2+f(x)=+ where f(x)=ln(x2)2x. By Theorem 3, limx2+1f(x)=limx2+2xln(x2)=0. The graph of y=(2x)/ln(x2) is shown in Figure 16. As we can in this figure, the function approaches 0 as x2+.

You may wonder why the function approaches +, as x approaches to 3 from the left, and the function approaches as x approaches to 3 from the right. The reason is that the denominator ln(x2) is zero at x=3, limx3(2x)=1, and ln(x2) is negative when x is less than 3 (for example if x=2.99, ln(x2)|x=2.99=ln0.99<0) Thus by Theorem 1 limx32xln(x2)=[10]+. A similar argument, shows us that limx3+2xln(x2)=[10+].

Figure 16

(d) Again because ln(x2) is not defined for x<2, the left-hand limit does not exist.