You are reading a page from An Elementary Treatise on Actuarial Mathematics by Harry Freeman (1932)
Part of the American Term Life Insurance History Project
Term Life Insurance
CHAPTER XII
MAXIMA AND MINIMA
if. In Chapter ix, para. 13, a definition was given of a function f (x) which is continuous at the point x = a. A property of such a function which is of frequent application in the calculus is as follows :
If f (x) is a continuous function throughout the range of values considered, then for values of x near the point x = a, f (x) has the same sign as f (a), provided that f (a) is not zero.
This proposition is almost self-evident. Since, for a continuous function, Lt f (x) is equal to f (a), it follows from the definition of
x-ka
a limit that there is a range of values over which f (x) differs from f (a) by less than E where E may be as small as we please. For any value of E numerically less than f (a) the sign of f (x) for values of x within the corresponding range will be the same as that of f (a).
Now let y = f (x) be a continuous function of x.
dy    Ay
Then    — Lt -
dx ,,,,. 0 Ax
so that Ox = dy
+ e where E is a small quantity whose limit as Ax -~ o is zero.
If dt is not zero, the sign of dx will be the same as that of 0~ provided that we take Ax small enough ; if Ax -> o, the sign of Ay will be the same as that of Ax dx.
Consequently if Ax is positive but -~ o, Ay will have the same d
sign as x
But    Ay = f (x + Ax) — f (x).
Therefore f (x + Ax) — f (x) will have the same sign as dx if Ax is positive, but — o.

MAXIMA AND MINIMA    191
If Ax is positive, x + Lx is greater than x; i.e. x is increasing: and if dx is positive f (x + Ox) is greater than f (x) ; i.e. y is in-creasing.
Therefore f (x) increases as x increases if /x is positive, and decreases as x increases if dx is negative.
Similarly, if Ox is negative, so that x is decreasing, f (x) decreases as x decreases if    is positive, and increases as x decreases if dx
is negative.
Values off (x) at which the function ceases to increase (decrease) and begins to decrease (increase) are called turning values or critical values.
2. Maxima and Minima.
At the points where the function y = f (x) ceases to increase and begins to decrease y is generally said to have a maximum value : conversely where the function ceases to decrease and begins to increase y is said to have a minimum value.
It should be noted that a maximum value need not necessarily
be the greatest numerical value of the function, nor need a minimum Y
value be the least. For example, in Fig. z6, there are maxima at the points A and C, and minima at B and D. The numerical value, however, of the ordinate at D is greater than that of the ordinate
at A, although the function assumes p    X
a minimum value at D and a maxi-    Fig. 26.
mum value at A.
The following is a more correct definition of maximum and minimum values :
The function y = f (x) has a maximum value at the point x = a, if f (a) exceeds both f (a + h) and f (a — h) for all positive values of h less than a small finite quantity E. Similarly, f (x) has a

192    DIFFERENTIAL CALCULUS
minimum value when x = a, if f (a) is less than both f (a + h) and f (a — h) for all positive values of h less than E.
If therefore f (a) is a maximum value of y = f (x), (i) as x in-creases from (a — h) to a, y increases and dx is positive ; and (ii) as x increases from a to (a + h), y decreases and dx is negative
(para. 1). That is, as x increases changes from a positive to a negative value. The criterion for a maximum value at x = a is therefore that dx changes sign from positive to negative as x passes
through a. Conversely, for a minimum value dx changes from negative to positive.
Since a continuous function cannot change sign without passing through a zero value, we have that, for a critical value, dx must be zero provided that it be continuous.
If therefore f (a) is a maximum or a minimum value of the function y = f (x), and f' (x) is continuous,

MAXIMA AND MINIMA    193 x2 — 3X + 2 = 0,
x=2 or I.
These values of x give critical values to y.
To find which of these values gives a maximum and which a minimum we must proceed further.
Let x = 2 — Ox and 2 + Lx in turn, where Ax is a small positive quantity.


(i) dx = (2 — 0x)2 3 (2 — Ox) + 2, when x = 2 — Ox, =4—44x+(0x)2—6+3AX+2=—Ox+(0x)2;
194    DIFFERENTIAL CALCULUS


The test is easy to apply. For example, using the same function as in Ex. 1, we find that
dx -~-~ 12 (x2—3x+2) = 12(2x-3).
This is positive when x = 2 and negative when x = 1. Consex = 2 gives a minimum value and x = 1 a maximum (as above).
4. The tests for maximum and minimum values are quite straight-forward and their application to simple problems presents little difficulty. The following examples are illustrative of the methods employed.
Example 2.
A window is in shape a rectangle with a semicircle covering the top. If the perimeter of the window be a fixed length p, find its maximum area.


We have first to choose an independent variable. Let BO, the radius of the semicircle, be x. Then since the perimeter of the figure is a fixed length p,
p=2BC+CD+AKB
= 2BC + 2x + vrx,
so that    BC=p—(2+7r)x
2
The area a will be [rectangle ABCD + semicircle AKB],
i.e.    a = BC. CD + ~~rx2
p—(2+~r)x
. 2x + Z7rx2 2
=xp—(2+?r)x2+21rx2.
da dx=p-2x(2+7r)+7rx.
For a maximum or minimum value dx = o, i.e.    x =    p
4+ITIt is evident that this will give a maximum value to a, for when x = o
the area is zero. We need not therefore apply the second test.

Fig. 27.

ILLUSTRATIVE EXAMPLES    195 Giving x the above value,
a=xp — (2+7r)x2+ 17rx2
p2 _ 4 + ,T    p2    p2
4+7r    2    (4+-ff)2 2(4+7r). Example 3.
Given u_I = — 5 ; ul = — I ; u2 = 4; U5 = 175 ; find the maximum and minimum values of ux.
Since four values of ux are given we must assume that the function is a rational integral function of the third degree in x.
Let y=ux=a+bx+cx2+dx3.
Then -5=u_1=a—b+c—d,
—I=u1=a+b+c+d,
 
4=u2=a+2b+4c+8d,

15 = u5 = a + 5b + 25c + 125d. Solving these equations we obtain easily that a=o; b=o; c=—3; d=2.
y=—3x2+2x3.
For critical values d= o,
x
i.e.    — 6x + 6x2 = o,
giving    x = o or 1.
Also    d y
= — 6 + 12x.
When x = o, d a is negative, giving a maximum value;
x = 1, d z is positive, giving a minimum value.
Therefore maximum value of y is o ;
minimum value of y is — I.
Example 4.
A ladder is to he carried in a horizontal position round a corner formed by two streets a feet and b feet wide meeting at right angles. Prove that the length of the longest ladder that will pass round the corner without jamming is (a* + b*)* feet.
In this example it is advisable to take as the variable the angle that the ladder makes with the wall of the street. Call this angle a. Let x be
13-2

196    DIFFERENTIAL CALCULUS
the distance across the corner. The longest ladder will of course be the shortest distance across the corner, and the problem reduces to one of finding the minimum value of x. Then
x=AB=AC+CB
= a sec a + b cosec a,


dx_asecatana — bcosecacota; da
i.e. for a maximum or minimum value, a sec a tan a — bcosecacota=o, sin a    cos a _
Or    a cost a b sing a — 0'

from which tan a = . a
B Fig. z8.
This evidently gives a minimum value to a sec a + b cosec a, the maxi-mum value being oo .
b*    al
sina=(al+bl)I; Cosa=(al+b)I,
a    b
and    x =    +
cos a sin a
= (al + b1)I
on simplifying.
5. Points of inflexion.
It has been seen above that, for a critical value at x = a, f' (a) = o and that in order to ascertain whether this value is a maximum or
a minimum, recourse must be had to the change of sign of f" (a), provided that f" (a) is not zero. The question of what happens if f" (a) is in fact zero now arises. Now f" (a) will be zero if there is no rate of change off' (a). In that event the first differential coeffiwill not increase (decrease)    and then decrease (increase) as x passes through a, although
f'(x) = o for the value x = a. f'(x) will have the same sign for

X Fig. zg.

POINTS OF INFLEXION    197
the value x = a + h as for x = a — h where h is small. There is therefore, as a rule, no maximum or minimum value at the point A, and A is said to be a point of inflexion on the curve.
In general there is a point of inflexion at x = a if f" (a) is zero : the exceptions depend upon the values of the higher derivatives.
Example 5.
Find the points of inflexion on the curve y (1 + x3) = 1. =1+x3'
dv     3x2
.. dx = (I + x3)2.
Therefore for a critical value, x = o. But    d2y    2X
6x4
doe. - _ - 3 `(I + x3)2 — (1 + x3)3
which is zero when x = o or x = ,Z/1,7. There are therefore points of inflexion where x = o and 4 .
6. We have illustrated the critical values of the function y = f (x) by reference to the geometry of the curve. The problem may also be considered analytically.
By the extension of the Mean Value Theorem (p. 177)
f (a + h) = f (a) + hf' (a) + Zh2f" (a + Ojh), and    f (a — h) = f (a) — hf' (a) + 1--h2f" (a — 02h),
where 01, B2 are positive proper fractions not necessarily equal. Now for critical values f' (a) = o.
Therefore f (a + h) — f (a) = +h2f" (a + 0,h), and    f (a — h) — f (a) = 2h2 f" (a — 02h).
If h be made sufficiently small the right-hand side will have the same sign in both expressions. For a maximum value
f (a + h) — .f (a) and f (a — h) — f (a)
will both be negative. Therefore, since 2h2 is positive, the second differential coefficient must be negative. Similarly for a minimum

198    DIFFERENTIAL CALCULUS
value the second differential coefficient must be positive. If, however, f" (a) is zero we must consider a further term : thus f (a + h) = f (a) + hf' (a)+2Zi f (a)+-i fm (a + 0,h),
h'    h'
.f (a — h) = f (a) — hf' (a) + 2 I f" (a) -i f'" (a — B4h),
which reduce to


.f (a + h) — f (a) = 3 f,,, (a + 03h), 3
and    f(a—h)—f(a)=—3f"' (a—04h).
It will be seen that here there can be no maximum or minimum value if f"' (a) is not zero, for since the sign of the right-hand side
h'
can be made to depend upon the sign of 3 ~ f"' (a), the signs of


f (a + h) — f (a) and f (a — h) — f (a) will be different and there will be a point of inflexion.
We may carry this proof further. If f"' (a) is zero, the condition for maxima and minima will depend upon the sign of f iv (a)—provided that f iv (a) is not zero—and so on. In general, therefore, we may say that
For a maximum or minimum value the first derivative that does not vanish must be of an even order : if in that event the derivative is negative the critical value is a maximum and if positive a mini-mum. Otherwise there will be a point of inflexion.
It is worthy of note that if f (x) is a continuous function, maximum and minimum values (if any) occur alternately.


Example 6.
Examine the critical values of y = x2 3x3 + 3x4 — x5. y=x2—3x3+3x4— x5.
- = 2X — 9x2 + 12x3 5X4.
Equating this to zero we obtain the four values x = o, 1, 1, . dx2y = 2 — 18x + 36x2 2ox3i

MISCELLANEOUS APPLICATIONS    199
•72, giving a maximum value.


d3y
= — 18 + 72X — 6ox2;
dx3
d3y    6,
dx3
and since this is not zero x = 1 gives a point of inflexion.
200    DIFFERENTIAL CALCULUS
d 'y
=3x—2b—a+3(x—a)=6x—2b—4a.
If x=a,    d
d2y
x =2a—2b, andifx=a 32b,    d2=2b— 2a.
We cannot therefore determine the sign of the second differential coefficient unless we know the relative magnitudes of a and b.
If a > b,    x = a + 2b gives y a maximum value,
3 a<b,    x=a
a = b,    there is a point of inflexion for the value x = a.
The required values of (x — a)2 (x — b) will be found by substitution of the values of x in the original expression.
Example S.
Divide the number 21 into three parts a, b, c in continued proportion such that 3a + 6b + 4C may be a maximum.
Since a, b, c are in continued proportion, a : b :: b : c, so that we may write a = bk and b = ck.
.'. a = ck2. But    a+b+c=zl.
:. (I + k + k2) c = 21     (i).
We have to find the maximum value of 3a + 6b + 4C or, in terms of c and k, of (4 + 6k + 3k2) c.
Let    (4 + 6k + 3k2) c = z
    (ii).
Then if there were one variable k, the necessary procedure would be to find the values of k which give dk a zero value. But c may vary as well as k, and if we differentiate z with respect to k a further differential coefficient, namely dk , is involved. We can, however, make use of equation (i) and thus eliminate dk from our differential equations.
Differentiating equations (i) and (ii) respectively with respect to k,
dimc
(6k+6) c + (4 + 6k +    dc_dz_0
3k,) dk dk
for a critical value.

ILLUSTRATIVE EXAMPLES    201
Eliminating dk we obtain easily that

3k2+2k—2=o,

so that k = —        the positive root giving a maximum value. 3
Substituting in (i) and simplifying,

c=14-V7,
whence    a = 14 — 4 'V7,

b=5'V77,
and the required value of 3a + 6b + 4C becomes 56 + 14 V7.
(The minimum value, found from the value k = —    'V5, is
3
56 — 14 V/7). Example 9.
Find the maximum value of 2 (a — x) (x + Vx2 + b2) where x is real.
In certain circumstances it may happen that a simple and straight-forward method of obtaining maximum or minimum values can be evolved by reference to algebra or geometry.
For example, although by differentiating the above expression and equating the result to zero the required value can be obtained, a neater proof results from the use of a well-known algebraic property.
(a — x)2 and x2 + b2 are positive since x is real.
 
(a — x)2 + (x2 + b2)
2 '(a — x)2 (x2 + b2),
i.e.   2 (a — x) 1'x2 + b2,
i.e.
a2 — tax + x2 + x2 + b2
2 (a — x) 1'x2 + b2,
or
2 (a — x) x + a2 + b2
2 (a — x) 1/'x2+b'2,
i.e.
a2 + b2
2 (a — x) (x +V x2 + b2).

In other words, 2 (a — x) (x + 1'x2 + b2) is not greater than a2 + b2, i.e. the maximum value of the expression is a2 + b2.

202    DIFFERENTIAL CALCULUS
EXAMPLES 12
Find the maximum and minimum values, where such exist, of
7. 12x5 45x4 + 40x3 + 6. 9. x3—5x2+8x—4.
x3
2. + ax2 3a2x.
3
4. x2 (I + x2)-3.
x2 I
6. (3 + x2)3.
8 (x + a) (x + b) (x—a)(x—b)'


10. ax + by, where xy = c2.
EXAMPLES    203
Illustrate your results by drawing a graph of the function.
36. If x be the independent variable find the maximum and minimum values of y given
y — 12 — x4 (5x2 + 6x — 15) = 0.
37. Explain how to discriminate between the maxima and minima values of f (x), if
df
and a, b, c be in ascending order of magnitude.
x2    2
EXAMPLES    205