The Signum Function of the Second Derivative and its application to the

determination of relative extremes of fractional functions.

Jesús Álvarez Lobo.

Asturias – Spain
J3A3L3@gmail.com

ABSTRACT ___________________________________________________________

Usually, the complexity of a fractional function increases significantly in its second

derivative, so the calculation of the second derivative can be tedious and difficult to
simplify and evaluate its value at a point, especially if the abscise isn't an integer.

However, to determine whether a point at which cancels the first derivative of a function
is a relative extremum (maximum or minimum) of it, is not necessary to know the value
of the second derivative at the point but only its sign.

Motivated by these facts, we define a signum function for the second derivative of
fractional functions in the domain of the roots of the first derivative of the function.

The method can dramatically simplify the search for maximum and minimum points in
fractional functions and can be implemented by means of a simple algorithm.

2321232

Signum Function of the Second Derivative (SF2D) ___________________________

Let

u ( x)
f ( x)  B
v( x)

Applying the rule for the derivative of a quotient in an unspecified point,

u´v  uv´
f ´ C
v2

Let x0  Dom f such that f is differentiable at x0 and f ( x0 )  0.

Then,
 u´v  uv´
f ( x0 )  0   0  u´v  uv´  0.
v2 x0 x0

And

(uv  uv  uv  uv)v  2v(u´v  uv´)

2
(uv  uv)
f ( x0 )  4
 .
v x0 v2 x0

But v2 > 0 as, by hypothesis, f is differentiable at x0 and so must be v  0 at x0  Dom f ;

therefore, the sign of f  doesn’t depend on v2 and so we can ignore v2. Therefore, the sign
of f  at the root x0 of f  is given by the simplest function

sgn f x0  sgn  uv  uv x0 D

Actually the sign function is not only a function but a class of functions, because if we define in a set of functions the relationship
"take the same sign at x0 that...," a partition is set in such set, classifying the functions in three classes of equivalence according to its
sign at that point: "+", "-", "0"". Therefore, as the sign for the second derivative at the points where it exists and annul the first
derivative we can adopt D or any other with the same sign at the same point. Taking into account this considerations, frequently, is
possible to use as signum function a simplest function that D.

Let's see through some examples how to implement the algorithm and its advantages over
the standard method.

Exemple 1_____________________________________________________________

Find all the relative extremum, minimums and maximums, in the following function:

(2 x  5) 2
f ( x)  .
x2  9
______________________________________________________________________

Proceeding in the conventional way, we have:

2(2 x  5)(5 x  18)  x  52

f ´(x)   0  
( x 2  9)2 x   5
18

2(20 x3  33x 2  540 x  99)

f ´´(x)  .
( x 2  9)3

Getting f  has been a so ungrateful work as unnecessary; but we are still waiting for the
worst: the evaluation of f  at the roots of f . This task would can be really deterrent, even
using a calculator.
Applying the SF2D :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

f   0  uv  uv  0 u  (2 x  5) 2  u´ 4(2 x  5)  u´´ 8

sgn f   sgn  u v  uv  v  x 2  9  v´ 2 x  v´´ 2

 x  52
f   0  (2 x  5)(5 x  18)  0  
x   5
18

sgn f   sgn  8( x 2  9)  2(2 x  5) 2   sgn  2(20 x  11) 

The canonical form of the SF2D (the simplest representative sign function of the class)
is sgn  20 x  11 , as the factor 2 can be neglected since does not affect the sign.

Now, the determination of the sign of f  at the roots of f  is immediate:

 sgn f   52   0  f ´´  52   0  min of f at x  52
  

sgn f x0  sgn 20 x  11 x0  
 sgn f    185   0  f ´´   185   0  max of f at x   185


 185 5
2
Exemple 2 _____________________________________________________________

Find all the relative extremum, minimums and maximums, in the following function:

3
f ( x) 
.
ln( x  3x 2  3) 4

______________________________________________________________________

Conventional way ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

 x1  0
6 x(2 x 2  3) 
f ( x)  2
 0   x2   2
6

( x  3x  3)  ln( x  3x  3) 
4 2 4

 x3  
6
2

(2 x 6  3x 4  9 x 2  9) ln( x 4  3x  3)  4 x 2 (2 x 2  3) 2
f ( x)  6  3
( x 4  3x 2  3) 2  ln( x 4  3x  3) 

As can be seen, even being a simple fractional function, its second derivative can present
monstrous appearance. In this case, although it is not necessary to use f  to determine the
sign of f , since
 (2 x6  3x 4  9 x 2  9) ln( x 4  3x  3)  4 x 2 (2 x 2  3) 2 

sgn f ( x)  sgn  ,
 ln( x 4  3x  3) 

the calculation is still complicated.

Applying the SF2D ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

 u  3  u  0  u  0
u 
f ( x)    4 x3  6 x 2 x 6  3x 4  9 x 2  9
v  v  ln( x 4
 3 x 2
 3)  v    v   2 
 x 4  3x 2  3 ( x 4  3x 2  3) 2

 x1  0
4 x3  6 x 
f   0  uv  uv  0  3  4  0  x(2 x 2  3)  0   x2   6

x  3x  3
2

2

 x3  
6
2

 2 x 6  3x 4  9 x 2  9 
sgn  uv  uv x0  sgn  6    sgn  2 x 6  3x 4  9 x 2  9 x0
 ( x  3x  3)  x0
4 2 2
sgn f (0)  sgn  2 x 6  3x 4  9 x 2  9 x 0  sgn  9   0  max at x  0

  9    62   
 
sgn f  
2
6
 sgn  x 4  2 x 2  3  2   9 
  x   x  6
 sgn  4
2


2 
6
22
 3  9 
22
6
  9
 
2

9    3  6
 sgn   3  3  6   9   sgn  9   1    0  min at x  
4    2  2

Exemple 3 [1] __________________________________________________________

Find all the local extremum, minimums and maximums, in the following function:

16
f ( x) 
.
x(4  x 2 )
______________________________________________________________________

Conventional way ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

 x1  
 2
16(3x 2  4)
f ( x)  2 2  
3
0 
x ( x  4)2  x2  
2
 3

3x 4  6 x 2  8
f ( x)  64 
x3 (4  x 2 )3
  3x 4  6 x 2  8 
sgn f    2
  sgn  x3 (4  x 2 )  .
  x  23
3

 3x 4  6 x 2  8 
sgn f    2
  sgn  x3 (4  x 2 )  .
  x . 23
3

whose calculation is complicated.

Applying the SF2D ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

u 
 u  16  u  0  u  0
f ( x)  
 v  4 x  x  v  4  3x  v  6 x
v 
3 2

 x1  
 2

f   0  uv  uv  0  4  3x  0   2 3

 x2  
2
 3

 sgn f ( x1 )    min at x   2


sgn f   sgn  uv  uv   sgn  x   
3

 sgn f ( x2 )    max at x   2

 3

Therefore, in this case the signum of f  is the same as that of the roots of f  , so it is not
necessary to perform any computations to know whether there is a maximum or a
minimum at each critical points.
Example 4 [2] __________________________________________________________

A generator of E volts is connected to an inductor of L henrys, a resistor of R ohms, and

a second resistor of x ohms. Heat is dissipated from the second resistor, the power P being
given
E2x
P
 2 L 2   x  R 2

(a) Find the resistance value x0 which makes the power as large as possible. Justify with
the second derivative test.

(b) Find the maximum power which can be achieved by adjustment of the resistance x.

______________________________________________________________________

Conventional way ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

a)
dP  2 L 2   x  R 2  2 x  x  R  x 2   2 L   R 2
2

E 
2
2
 E 
2
2
.
dx   2 L 2   x  R 2    2 L 2   x  R 2 

Therefore, if E  0,

dP
 0  x 2   2 L   R 2  0 ,
2

dx

whose only solution in the context of the problem is

x0   2 L   R 2
2

2
d2y 
2 2 x  2 L
2   x  R 2   4  x  R   x 2   2 L 2  R 2    2 L 2   x  R 2 
 E  4
dx 2   2 L 2   x  R 2 

2 2 x  2 L
2   x  R 2   4  x  R   x 2   2 L 2  R 2 
 E  3
  2 L 2   x  R 2 
2 E 2  x3  3   2 L   R 2  x  2 R   2 L   R 2 
2 2

 3
.
  2 L 2   x  R 2 

The second derivative of P not only involves a considerable time of calculation and
simplification, with a high risk of committing some error, but the evaluation of it for the
root x0 of the first derivative is little less than dissuasive.
   2 L 2  R 2   3   2 L 2  R 2   2 L 2  R 2  2 R   2 L 2  R 2 
3

P   2 L   R
2 2
  2E 2
 .
  2 L 2    2 L 2  R 2  R 2 
3

The sign of this expression is the same as

3
  2 L 2  R 2   3   2 L 2  R 2   2 L 2  R 2  2R   2 L 2  R 2 
     

since 2E 2 and the denominator are positives.

… and after this hard work, we have not yet gotten to know what sign takes the second
derivative.

Applying the SF2D ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

a)
 u  E 2 x  u  E 2  u  0
u 
P( x)   
 v   2 L    x  R   v  2  x  R   v  2
v 
2 2

P´ 0  uv  uv  0  E 2   2 L    x  R    E 2 x  2  x  R   x0    2 L   R 2

2 2 2

sgn P  sgn  uv  uv   sgn  2E 2 x    sgn  x   sgn P  
 2 L 2  R 2  

Therefore, P reaches a maximum at x0   2 L   R 2 .

2

b)

The value of the maximum of P is

P  E 2  2 L   R 2
2
 2 L 2  R 2  2
.
 2 L     2 L 2  R 2  R 
2
Example 5 _____________________________________________________________

3  x  2 x2
Determine the relative maximum and minimum of the function f ( x)  .
3  2 x2

Applying the SF2D ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

u 
 u  3  x  2 x  u  1  4 x  u  4
2

f ( x)  
 v  3  2 x  v  4 x  v  4
v  2

f   0  uv  uv  0  1  4 x   3  2 x 2    3  x  2 x 2   4 x  0

 x   2 6
5 6
 2 x 2  24 x  3  0   1
 x2   2  6

5 6

 sgn f ( x1 )  
sgn f   sgn  uv  uv    sgn  x  6    sgn  5 6
2
  


 sgn f ( x2 )  

Hence, f ( x) has a minimum at x   5 26  6 of value f   5 26  6    5126 and a

maximum at x  5 26  6 of value f  5 6
2  6   5126 .
References _____________________________________________________________

[1] Demidovich (Editor). Problems in Mathematical Analysis.

MIR Publishers – Moscow. Problem 835.

[2] Jerrold Marsden, Alan Weinstein. Calculus I. 2 nd Edition.

Chapter 3: Graphing and Maximum-Minimum Problems.
Exercises for Section 3.3: Page 162. Exercise 40.

