UNIT IV

Functions with non-zero derivatives

Theorem 1: Let f be defined on an open interval a, b and assume that for some c

in a, b , we have f c   0 or f c   . Then there is a 1-ball Bc   a, b in

which
f x   f c  if x  c and f x  f c if x  c .
Proof: Case (i): f c   0 and is finite.

Since f c  exists, by a known theorem, we have,

f x   f c   x  c  f  x   1

where f  is continuous at c and f  c   f c  . Since f c   0 , we have f  c   0 .

By the sign preserving property of continuous functions, there is a 1-ball Bc   a, b

in which f  x  has the same sign as f  c  .

Then 1 
f  x   f c 
 f   x   0 when x  c
xc
That is,
f  x   f c 
 0  2
xc
2 
f x  f c  0 if x  c  0 and f x  f c  0 if x  c  0
That is,
f x  f c if x  c and f x  f c if x  c
Case(ii): f c  

If f c   , then there is a 1-ball Bc   a, b in which

f  x   f c 
 1 whenever x  c
xc
This implies that the quotient is positive, that is f c   0 , then from case(i), we have

f x  f c if x  c and f x  f c if x  c

Definition 1: Let f be a real-valued function defined on a subset S of a metric space
M , and assume a  S . Then f is said to have a local maximum at a if there is a ball

Ba such that

f x  f a  x  Ba  S
If
f x  f a  x  Ba  S
then f is said to have a local minimum at a .

Theorem 2: Let f be defined on an open interval a, b and assume that f has a

local maximum or a local minimum at an interior point c of a, b . If f has a

derivative (finite or infinite) at c , then f c  must be zero.
Proof: Assume that f has a local maximum or a local minimum at an interior point c

of a, b . Let f has a derivative at c .

To prove: f c   0
Assume the contrary. That is suppose f c   0 , then either f c   0 or f c   0 . If

f c  0 or f c   , then by Theorem 1,

f x  f c if x  c and f x  f c if x  c
This implies f cannot have a local maximum or local minimum at c . Similarly, if

f c  0 or f c   , then again by the Theorem 1,

f x  f c if x  c and f x  f c if x  c
This implies f cannot have a local maximum or local minimum at c . Hence we
arrive at a contradiction to our assumption in both the arguments and our assumption
that f c   0 is wrong. But since f has a derivative at c , the only possibility is

f c  0 . Hence the theorem.

Note 1: It is to be noted that the converse of Theorem 2 is not true.

Rolle’s theorem: Assume f has a derivative (finite or infinite) at each point of an

open interval a, b , and assume that f is continuous at both end points a and b . If

f a   f b, there is at least one interior point c at which f c  0 .

Proof: (by contradiction)
Assume f has a derivative at each point of an open interval a, b , and f is

continuous on a, b. Let f a   f b and c  a, b . Suppose f  is never 0 in a, b .
Since f is continuous on a compact set, f attains its maximum say M and a

minimum say m somewhere in a, b . But no extreme value is attained at an interior

point as we assume that f  is never 0 in a, b . So both exist at the end points a and

b . As f a   f b, m  M and hence f is constant on a, b . A contradiction to our

assumption that f  is never 0 in a, b . Therefore f c   0 for some c  a, b .

Mean-Value theorem: Assume f has a derivative (finite or infinite) at each point of

an open interval a, b , and assume also that f is continuous at both end points a and

b . Then there is a point c  a, b such that

f b  f a   f cb  a
Proof: Assume f has a derivative at each point of an open interval a, b , and assume

also that f is continuous at a and b . Let c  a, b . Then we have,

lim f x   f a  and lim f x   f b 
x a x b

Also, since f has a derivative at each point of a, b and c  a, b , by a known
theorem,
f x  f c  f cx  c  1
When x  a , 1 becomes,

f a  f c  f ca  c  2

And when x  b , 1 becomes,

f b  f c  f cb  c  3

3  2 
f b   f c   f a   f c   f c b  c  a  c 
 f b   f a   f c b  a 
Hence the theorem.

Generalized Mean-Value Theorem: Let f and g be two functions each having a

derivative (finite or infinite) at each point of an open interval a, b and each
continuous at the end points a and b . Assume also that there is no interior point x at
which both f x and g x  are infinite. Then for some interior point c , we have,

f cg b  g a  gc f b  f a

Proof: Let
hx  f xg b  g a  g x f b  f a  1
Then h is continuous at the end points a and b , as f and g are continuous at the

end points a and b . Also from 1 ,

ha   f a g b   g a   g a  f b   f a 
 f a g b   f a g a   g a  f b   g a  f a 
 f a g b   g a  f b 
That is,
ha  f ag b  g a f b  2
And,
hb   f b g b   g a   g b  f b   f a 
 f b g b   f b g a   g b  f b   g b  f a 
 f a g b   g a  f b 
That is,
hb  f ag b  g a f b  3
From 2  and 3 ,
ha   hb  4
Differentiating 1 with respect to x , we have,

hx  f xg b  g a  gx f b  f a  5

Then hx  is finite if both f x  & g x  are finite and hx  is infinite if any one of

f x or gx is infinite. Hence from all the above discussions, h satisfies the Rolle’s
theorem, and hence,
hc  0 for some c  a, b
Therefore from 5 , we have,
hc   f c g b   g a   g c  f b   f a   0
 f c g b   g a   g c  f b   f a 
Hence the theorem.
Theorem 3: Let f and g be two functions each having a derivative (finite or infinite)

at each point of a, b . At the end points assume that the limits f a , g a , f b 

and g b  exist as finite values. Assume further that there is no interior point x at
which both f x and g x  are infinite. Then for some interior point c , we have

f cg b   g a   gc f b   f a 

Proof: Define two new functions F and G on a, b as follows:

F x  f x and Gx  g x, if x  a, b  1

F a  f a , Ga  g a , F b  F b , Gb  g b   2
Since f and g are continuous, from 1 F and G are continuous on a, b . Then by
the Generalized Mean-value theorem, we have
F cGb  Ga  GcF b  F a  3
From 1 and 2 ,
F x   f x   F c   f c  and Gx   g x   Gc   g c 
F a   f a   f a    f a  and Ga   g a   g a    g a 
F b   f b   f b    f b  and Gb   g b   g b    g b 
Therefore 3 becomes,
f c g b   g a   g c  f b   f a 
 f c g b    g a    g c  f b    f a  
Hence the theorem.

Theorem 4: Assume f has a derivative (finite or infinite) at each point of an open

interval a, b and that f is continuous at the endpoints a and b .

a  If f  takes only positive values (finite or infinite) in a, b , then f is strictly

increasing on a, b
b  If f  takes only negative values (finite or infinite) in a, b , then f is strictly

decreasing on a, b

c If f  is zero everywhere in a, b , then f is constant on a, b

Proof: Assume f has a derivative at each point of an open interval a, b and that f

is continuous at the endpoints a and b . Let x, y  a, b , then x  y and by the
Mean-value theorem to x, y  , we have
 f  y   f x   y  x f c  1 , where c  x, y 
a  If f  takes only positive values in a, b , then f c   0 and from 1 , we have

f  y   f x    y  x  f c   0
 f  y   f x   0
 f x   f  y 
Hence x  y  f x  f  y  and hence f is strictly increasing on a, b .
b  If f  takes only negative values in a, b , then f c   0 and from 1 , we have

f  y   f x    y  x  f c   0
 f  y   f x   0
 f x   f  y 
Hence x  y  f x  f  y  and hence f is strictly decreasing on a, b .

c If f  is zero everywhere in a, b , then from 1 , we have

f  y   f x    y  x  f c   0
 f  y   f x 
This implies that f is constant on a, b . Hence the theorem.

Corollary 1: If f and g are continuous on a, b and have equal finite derivatives in
a, b , then f  g is constant on a, b.

Proof: Let f and g be continuous on a, b and f   g . Then


f   g  0  f  g   0 and by Theorem 4, f is constant on a, b .