Skip to content

Unit 1: Limits and Continuity

Working with the Intermediate Value Theorem

Use the Intermediate Value Theorem to prove a continuous function takes a value or has a zero, check its conditions, and write the justification graders expect.

Topic 1.16 · 9 min read

At 6 a.m. the air temperature is 4 degrees Celsius, and at 2 p.m. it is 17 degrees. You can be sure that at some moment in between it was exactly 10 degrees, even with no thermometer reading at that moment, because temperature cannot get from 4 to 17 without passing through every value in between. The Intermediate Value Theorem (IVT) makes that reasoning precise for any continuous function. It is an existence theorem: it proves that a value is reached without telling you where.

What the theorem says

The picture explains it. Continuity on [a,b][a, b] means the graph from (a,f(a))(a, f(a)) to (b,f(b))(b, f(b)) is a single unbroken curve. If the horizontal line y=dy = d lies between the heights of the two endpoints, one endpoint is below the line and the other above it, and an unbroken curve cannot get from one side to the other without meeting the line. Each place it meets the line is an input cc with f(c)=df(c) = d.

A continuous curve on the interval from a to b starting at the point (a, f(a)) below the dashed horizontal line y = d and ending at the point (b, f(b)) above it. The curve crosses the line three times, at x = c1, c2 and c3, so f(c) = d has three solutions in this case; the theorem only promises at least one.

This curve meets y=dy = d three times, at c1c_1, c2c_2 and c3c_3, while the theorem promises only one; the section after next says what it leaves out. A full proof relies on a property of the real numbers (that they have no gaps) studied in college analysis, and the AP exam treats the IVT as a theorem you apply, not one you prove.

The most-used special case is d=0d = 0. If ff is continuous on [a,b][a, b] and f(a)f(a) and f(b)f(b) have opposite signs, then 0 is between them, so ff has at least one zero in (a,b)(a, b).

Continuity on the closed interval is the condition that matters

Every conclusion of the IVT depends on continuity on all of [a,b][a, b], including the endpoints. Without it, the graph can skip over the line y=dy = d.

  • A jump skips values. Let f(x)=x+1f(x) = x + 1 for x<2x < 2 and f(x)=x+3f(x) = x + 3 for x≥2x \geq 2. On [0,4][0, 4], f(0)=1f(0) = 1 and f(4)=7f(4) = 7, and 4 is between them. But the left piece only produces values from 1 up to (not including) 3, and the right piece produces values from 5 to 7. No input gives 4. The IVT does not apply, because ff has a jump at x=2x = 2.
  • An asymptote skips values. On [π4,3π4]\left[\frac{\pi}{4}, \frac{3\pi}{4}\right], tan⁡π4=1\tan\frac{\pi}{4} = 1 and tan⁡3π4=−1\tan\frac{3\pi}{4} = -1, opposite signs. Yet tan⁡x\tan x has no zero on that interval (tan⁡x=0\tan x = 0 only where sin⁡x=0\sin x = 0, at multiples of π\pi). The graph gets from positive to negative by going up to ∞\infty and coming back from −∞-\infty at the vertical asymptote x=π2x = \frac{\pi}{2}.

On the exam you confirm continuity in one of three ways:

  • the problem says so ("ff is continuous on [0,12][0, 12]");
  • the function is built from continuous pieces: polynomials, exponentials, sin⁡x\sin x and cos⁡x\cos x everywhere; rational functions, logarithms, roots and tan⁡x\tan x on intervals inside their domains; and sums, products, quotients (with nonzero denominator) and compositions of these (topic 1.12);
  • the problem says ff is differentiable, since a function that is differentiable at a point is continuous there (topic 2.4).

What the theorem does not say

The IVT is a one-way statement, and three tempting extensions of it are false.

It doesn't count. It guarantees at least one cc, never exactly one. Claiming "the solution" or "exactly one zero" needs more information, such as knowing ff is always increasing (Unit 5).

It doesn't locate. It gives an interval that contains a cc, not the value of cc. You can shrink the interval by testing more inputs (Example 1 does this), but the IVT alone never names cc.

It is silent when the value isn't bracketed. If dd is not between f(a)f(a) and f(b)f(b), the theorem makes no claim at all; in particular, if f(a)f(a) and f(b)f(b) have the same sign, it says nothing about zeros. The function might still have some. For f(x)=4x2−1f(x) = 4x^2 - 1 on [−1,1][-1, 1], both endpoint values are 3, and still f(±12)=0f\left(\pm\frac{1}{2}\right) = 0. "The IVT does not guarantee a zero" and "there is no zero" are different statements.

Equations, tables, and the written justification

Equations between two functions. To show two continuous functions are equal somewhere, apply the IVT to their difference h(x)=f(x)−g(x)h(x) = f(x) - g(x), which is continuous because ff and gg are. A sign change in hh between aa and bb means f(c)=g(c)f(c) = g(c) for some cc between them.

Tables. When a continuous function is given only by a table, apply the IVT on each pair of consecutive inputs. Each subinterval where ff passes from below dd to above it (or the reverse) contains its own solution of f(x)=df(x) = d, and subintervals that don't overlap give different solutions. Counting those subintervals gives the fewest solutions the table guarantees.

Worked examples

Worked exampleExample 1: an equation with no algebraic solution

Show that the equation x3+x=3x^3 + x = 3 has a solution between x=1x = 1 and x=2x = 2, and then find an interval of length 0.10.1 that contains a solution.

Show solution

Let h(x)=x3+xh(x) = x^3 + x, a polynomial, so hh is continuous on [1,2][1, 2]. Evaluate the endpoints:

h(1)=2,h(2)=10.h(1) = 2, \qquad h(2) = 10.

Since h(1)=2<3<10=h(2)h(1) = 2 < 3 < 10 = h(2), the Intermediate Value Theorem gives a number cc with 1<c<21 < c < 2 and h(c)=3h(c) = 3, so c3+c=3c^3 + c = 3.

To narrow it down, test inputs inside the interval: h(1.2)=1.728+1.2=2.928h(1.2) = 1.728 + 1.2 = 2.928, which is less than 3, and h(1.3)=2.197+1.3=3.497h(1.3) = 2.197 + 1.3 = 3.497, which is more than 3. Since hh is continuous on [1.2,1.3][1.2, 1.3] and h(1.2)<3<h(1.3)h(1.2) < 3 < h(1.3), the IVT puts a solution in (1.2,1.3)(1.2, 1.3). Repeating this with more test points shrinks the interval as far as you like, though it never produces cc exactly.

Worked exampleExample 2: showing at least two solutions

Show that the equation ex=4−x2e^x = 4 - x^2 has at least two real solutions.

Show solution

Move everything to one side and define h(x)=ex+x2−4h(x) = e^x + x^2 - 4. A solution of the equation is a zero of hh. The function hh is a sum of an exponential and a polynomial, so it is continuous for all real xx.

To get two solutions, find two non-overlapping intervals that each have a sign change. Try a few easy inputs:

h(−2)=e−2>0,h(0)=1+0−4=−3<0,h(2)=e2>0.h(-2) = e^{-2} > 0, \qquad h(0) = 1 + 0 - 4 = -3 < 0, \qquad h(2) = e^2 > 0.

On [−2,0][-2, 0]: hh is continuous and h(0)<0<h(−2)h(0) < 0 < h(-2), so by the IVT there is a c1c_1 in (−2,0)(-2, 0) with h(c1)=0h(c_1) = 0.

On [0,2][0, 2]: hh is continuous and h(0)<0<h(2)h(0) < 0 < h(2), so by the IVT there is a c2c_2 in (0,2)(0, 2) with h(c2)=0h(c_2) = 0.

The intervals (−2,0)(-2, 0) and (0,2)(0, 2) do not overlap, so c1≠c2c_1 \neq c_2, and the equation has at least two solutions. The middle test point x=0x = 0 is what makes this work: hh is negative there and positive at both ends, so there is one sign change on each side of it.

Worked exampleExample 3: a table in context (exam style)

Water flows into and out of a tank. The volume of water in the tank, in liters, is modeled by a differentiable function VV of time tt, in hours, for 0≤t≤120 \leq t \leq 12. Selected values of V(t)V(t) are given in the table below.

t025912V(t)4638445139\begin{array}{c|ccccc} t & 0 & 2 & 5 & 9 & 12 \\ \hline V(t) & 46 & 38 & 44 & 51 & 39 \end{array}

(a) Must there be a time tt, 0<t<120 < t < 12, at which V(t)=50V(t) = 50? Justify your answer.

(b) What is the fewest number of times in the interval 0<t<120 < t < 12 at which V(t)=40V(t) = 40? Justify your answer.

(c) Does the Intermediate Value Theorem guarantee a time at which V(t)=55V(t) = 55? Explain.

(d) A second tank holds U(t)=30+1.5tU(t) = 30 + 1.5t liters at time tt. Show that there is a time tt, 0<t<120 < t < 12, at which the two tanks hold the same volume.

Show solution

Since VV is differentiable, it is continuous on [0,12][0, 12], and so on every closed subinterval. State this once at the start; every part uses it.

(a) Yes. VV is continuous on [5,9][5, 9], and V(5)=44<50<51=V(9)V(5) = 44 < 50 < 51 = V(9). By the Intermediate Value Theorem, there is a time tt with 5<t<95 < t < 9 at which V(t)=50V(t) = 50.

(b) At least 3 times. On each of these intervals VV is continuous and 40 lies between the endpoint values:

  • [0,2][0, 2]: V(2)=38<40<46=V(0)V(2) = 38 < 40 < 46 = V(0);
  • [2,5][2, 5]: V(2)=38<40<44=V(5)V(2) = 38 < 40 < 44 = V(5);
  • [9,12][9, 12]: V(12)=39<40<51=V(9)V(12) = 39 < 40 < 51 = V(9).

By the IVT, V(t)=40V(t) = 40 at some time in each of (0,2)(0, 2), (2,5)(2, 5) and (9,12)(9, 12). These intervals don't overlap, so the three times are different. On [5,9][5, 9] both values are above 40, so the table guarantees nothing there.

(c) No. The IVT gives a value dd only when dd lies between two values of VV, and every value in the table is at most 51, so no pair of table values brackets 55. That does not mean V(t)V(t) never reaches 55: the volume could rise above 51 between readings. The table and the IVT just can't show it.

(d) The tanks hold the same volume exactly when V(t)−U(t)=0V(t) - U(t) = 0, so let h(t)=V(t)−U(t)h(t) = V(t) - U(t). It is continuous on [0,12][0, 12] as the difference of two continuous functions (UU is linear). At the endpoints,

h(0)=46−30=16>0,h(12)=39−(30+18)=−9<0.h(0) = 46 - 30 = 16 > 0, \qquad h(12) = 39 - (30 + 18) = -9 < 0.

Since h(12)<0<h(0)h(12) < 0 < h(0), the IVT gives a time tt with 0<t<120 < t < 12 and h(t)=0h(t) = 0, that is, V(t)=U(t)V(t) = U(t): the two tanks hold the same volume at that time.

Where students lose points

On the exam

The exam asks about the IVT in forms like Example 3: a table of values of a function stated to be continuous or differentiable, with a question such as "must there be a value..." or "what is the fewest number of times...." The answer earns credit only with the three-part justification: continuity, the inequality bracketing the value with numbers from the table, and the theorem's name with the conclusion. The work is the same with or without a calculator.

Check yourself

  1. Question 1

    The function ff is continuous on [1,5][1, 5], with f(1)=4f(1) = 4 and f(5)=−2f(5) = -2. Which of the following must be true?

    1. A

      f(c)=0f(c) = 0 for at least one cc in (1,5)(1, 5).

    2. B

      ff has exactly one zero in (1,5)(1, 5).

    3. C

      f(3)=1f(3) = 1

    4. D

      f(x)≤4f(x) \leq 4 for every xx in [1,5][1, 5].

    Show answer

    A

    0 lies between −2-2 and 4, and ff is continuous on [1,5][1, 5], so the IVT gives at least one zero in (1,5)(1, 5). B claims a count and C a location, neither of which the theorem provides; 1 is the average of the endpoint values, but nothing forces f(3)f(3) to be that. D can fail because ff may rise above 4 between the endpoints.

  2. Question 2

    Let g(x)=1x−3g(x) = \dfrac{1}{x - 3}. Then g(2)=−1g(2) = -1 and g(4)=1g(4) = 1. Does the Intermediate Value Theorem guarantee that gg has a zero in (2,4)(2, 4)?

    Show answer

    No. gg is not continuous on [2,4][2, 4] (it is undefined at x=3x = 3), so the theorem does not apply, and in fact gg has no zero anywhere.

    The sign change comes from the vertical asymptote at x=3x = 3: gg jumps from −∞-\infty to ∞\infty there. A fraction is 0 only when its numerator is 0, and the numerator here is the constant 1.

  3. Question 3

    A continuous function ff has f(0)=2f(0) = 2, f(1)=−1f(1) = -1, f(2)=−3f(2) = -3, f(3)=1f(3) = 1 and f(4)=−2f(4) = -2. What is the fewest number of zeros ff can have on [0,4][0, 4]?

    Show answer

    3

    The sign changes between consecutive inputs on [0,1][0, 1], [2,3][2, 3] and [3,4][3, 4], and ff is continuous on each, so the IVT gives a zero in each of (0,1)(0, 1), (2,3)(2, 3) and (3,4)(3, 4). These intervals don't overlap, so the zeros are different. On [1,2][1, 2] both values are negative, so no zero is guaranteed there.

  4. Question 4

    The function ff is continuous on [−1,1][-1, 1] with f(−1)=3f(-1) = 3 and f(1)=3f(1) = 3. Does it follow that ff has no zeros in [−1,1][-1, 1]?

    Show answer

    No. The IVT makes no claim when 0 is not between f(−1)f(-1) and f(1)f(1); ff may or may not have zeros.

    For example, f(x)=4x2−1f(x) = 4x^2 - 1 is continuous, equals 3 at both endpoints, and has zeros at x=−12x = -\frac{1}{2} and x=12x = \frac{1}{2}. Endpoint values of the same sign only mean the theorem gives no guarantee.

Go deeper