A Note On Lagrange's Mean Value Theorem
A Note On Lagrange's Mean Value Theorem
T his chapter’s topic is called the Mean Value Theorem, or MVT. The MVT
is not something (like, say, the chain rule) that you will use daily, but it
does have some important consequences that we will address in Section 35.1.
y
The MVT is easy to visual-
y = f (x)
ize. It concerns a function
f (x) that is defined on a closed
interval [a, b], as shown on
the right. We’re doing calcu-
lus, so say that f (x) is differ- x
a b
entiable, at least on (a, b), and
continuous on [a, b]. y
Draw a line segment between y = f (x)
° ¢ ° ¢
points a, f (a) and b, f (b) . f (b)
This line segment has slope
f (a) f ( b )° f ( a )
rise f (b) ° f (a) m=
m= = . b°a
run b°a
From this picture you would a
x
b
expect that at some point the
tangent to y = f (x) has the y
same slope as the line seg- m = f 0 (c)
ment. In other words, there y = f (x)
f (b)
is a number c between a and
f ( b )° f ( a )
b for which f 0 (c) = b°a . In f (a) m=
f ( b )° f ( a )
fact, this is exactly what the b°a
mean value theorem says. x
a c b
f (b) ° f (a)
f 0 (c) = .
b°a
401
Example 35.1 Consider f (x) = x2 + 1 on the interval [a, b] = [°1, 2], which
is graphed below. Here f (bb)° f ( a)
°a
f (2)° f (°1)
= 2°(°1) = 5°3 2 = 1, so the mean value
theorem states that there is a number c between °1 and 2 for which f 0 (c) = 1.
y f (x) = x2 + 1
f (2) ° f (°1) 5 ° 2
m= = =1
2 ° (°1) 3
°1¢
m= f0 =1
2
x
°1 1 2
2
This number c for which f 0 (c) = 1 is easy to find. Because f 0 (x) = 2x, the
equation f 0 (c) = 1 yields 2c = 1, so c = 12 . The tangent line at x = c = 12 is
° ¢
shown above, and indeed its slope is f 0 12 = 2 · 12 = 1.
m = f 0 (c)
f (b)
f (b) ° f (a)
m=
f (a) b°a
a c c c b
The mean value theorem can be proved easily with familiar techniques
of finding global extrema. We will do this at the end of the chapter. But first
we’ll do one more example and then discuss some important consequences
of the MVT.
402 The Mean Value Theorem
f (t)
vel = f 0 (t)
t=0 t= 13
30 miles
Notice that f (t) is defined for times 0 ∑ t ∑ 1/3, that is, f is defined on the
interval [a, b] = [0, 1/3]. The mean value theorem says that for some time
t = c in [0, 1/3]
Thus the mean value theorem simply asserts that the average rate of change
of f on [a, b] equals exact rate of change at some c.
Consequences of the Mean Value Theorem 403
Fact 35.2 If f (x) is differentiable on an interval (a, b), and f 0 (x) = 0 for
all x in (a, b), then f is a constant function f (x) = C on (a, b).
Proof. Suppose f (x) meets the stated conditions. Fix a number d in (a, b).
Let C = f (d). We’ll show below that f (x) = C for any x in (a, b).
So take any x in (a, b). Assume first that x is to the left of the number d .
Because f is differentiable on (a, b) it is also differentiable on the interval
(x, d), and continuous on [x, d] (because differentiability implies continuity).
By the MVT, there is a number c in [x, d] with
f (d) ° f (x)
f 0 (c) = .
d°x
Because f 0 (x) = 0 on (a, b), we know f 0 (c) = 0, and the above becomes
f (d) ° f (x)
0 =
d°x
0 · (d ° x) = f (d) ° f (x)
0 = f (d) ° f (x)
f (x) = f (d) = C.
We’ve just shown that f (x) = C for any x in (a, b) that is to the left of d .
If x is to the right of d , just repeat the above argument, replacing the
interval (x, d) with (d, x). Again, we get f (x) = C . Á
In short, Fact 35.2 says: If the derivative of a function is zero, then the
function is a constant. We sometimes downplay the interval, leaving it
unspecified or regarding it simply as (a, b) = (°1, 1) = R if appropriate.
404 The Mean Value Theorem
y
y = f (x)
C y = g(x)
x
x
Thus the rise and fall of f (x) echos that of g(x); the graph of f (x) looks
just like the graph of g(x) except that it’s slightly higher. You would guess
that f (x) = g(x) + C for some constant C . That is exactly what our second
MVT consequence says.
Fact 35.3 Suppose two functions f (x) and g(x) are differentiable on an
interval (a, b). If f 0 (x) = g0 (x) for all x in (a, b), then f (x) = g(x) + C for some
constant C .
Proof. Suppose f 0 (x) = g0 (x) for all x in (a, b). Let h(x) = f (x) ° g(x). Then
h0 (x) = f 0 (x) ° g0 (x) on (a, b). Because f 0 (x) = g0 (x), this becomes h0 (x) = 0
on (a, b). Then by Fact 35.2, h(x) = C for some constant C . This means
f (x) ° g(x) = C , so f (x) = g(x) + C . Á
In short, Fact 35.3 says: Two functions with the same derivative differ
by a constant. Again, we sometimes downplay the interval, regarding as
unspecified or simply as (a, b) = (°1, 1) = R. But technically the interval
has significance, because not all functions have domain R.
It is also important that f and g be defined on a single interval (a, b).
If they are defined on more than one interval, then Fact 35.3 may break
down. (See this chapter’s Exercise 6.)
Likewise if f is not defined on a single interval (a, b), then Fact 35.2
may not hold. Exercise 5 asks for an example of function f with domain
(°1, 0) [ (0, 1), such that f is non-constant but f 0 (x) = 0.
Proof of the Mean Value Theorem 405
g0 (c) = 0
y = g(x)
g(a) = g(b)
a c b
Proof. Let g be as stated, and suppose g(a) = g(b). By the extreme value
theorem (Fact 33.1), g has both a global maximum and minimum on [a, b].
If a global maximum or minimum occurs at a point c in (a, b), then c
must be a critical point of g. This means that either g0 (c) is undefined,
or g0 (c) = 0. But g is differentiable on (a, b), so g0 (c) cannot be undefined.
Therefore g0 (c) = 0.
On the other hand, if neither the global maximum nor minimum occurs
inside (a, b), then the maximum occurs at one endpoint of [a, b] and the min-
imum at the other. But since g(a) = g(b), this means the absolute maximum
and absolute minimum are equal, so g must be a constant function on [a, b].
Then f 0 (c) = 0 for any c in (a, b). Á
The mean value theorem follows quickly from Rolle’s theorem. For
convenience re restate it below.
406 The Mean Value Theorem
Proof. Say f (x) is continuous on [a, b] and differentiable on (a, b). We now
explain why the stated number c must exist.
y
y = f (x)
y = L(x)
f (b)
f (a) f ( b ) ° f ( a)
m=
b°a
x
a b
° ¢ ° ¢
Consider the straight line passing through the points a, f (a) and b, f (b) .
(See the diagram above.) This line is the graph of a linear function L(x).
The derivative of L(x) equals the slope of the line, that is,
f (b) ° f (a)
L0 (x) = (§)
b°a
g0 (c) = 0.
f 0 (c) ° L0 (c) = 0
f (b) ° f (a)
f 0 (c) ° = 0 (by (§))
b°a
f (b) ° f (a)
f 0 (c) = .
b°a
2. Consider the function ln( x) on the interval [1, 5]. Find the number c in (1, 5)
guaranteed by the mean value theorem.
3. Consider the function f ( x) = x3 ° x on the interval [°2, 3]. Find all numbers c in
(°2, 3) guaranteed by the mean value theorem.
4. The record for weight loss in a human is a drop from 487 pounds to 130 pounds
over an eight month period. Use the mean value theorem to show that the rate
of weight loss exceeded 44 pounds per month at some time during the eight
months.
5. Find an example of a function f ( x), with domain (°1, 0) [ (0, 1), for which f 0 ( x) = 0
but f ( x) is not a constant function. (This example shows that Fact 35.2 can fail
if f is not defined on a single interval (a, b).)
6. Find two functions f ( x) and g( x), each with domain (°1, 0) [ (0, 1), for which
f 0 ( x) = g0 ( x), but f ( x) 6= g( x) + C . (This example shows that Fact 35.3 can fail if f
and g are not defined on a single interval (a, b).)
5. Find an example of a function f ( x), with domain (°1, 0) [ (0, 1), for which f 0 ( x) = 0
but f ( x) is not a constant function.
(
5 if x < 0
Solution Let f ( x) =
3 if x > 0.
This has domain (°1, 0) [ (0, 1). It is not a constant function, but f 0 ( x) = 0.