EXERCISE 5.2.

(a) Similar to the harmonic series

n=1∞1n=1+12+13+14+⋯+1n+⋯

of Example 5.1.2, the series n=2∞1n also diverges. Since

1lnn>1n

then by Comparison Test, the series

n=1∞1lnn

is divergent.

(b) Using the Ratio Test, we find that

limn→∞n+1!10n+1n!10n=n+110=∞ .

Hence, the series

n=1∞n!10n

is divergent.

(c) For large n the dominant term in the denominator of the given series
n=1∞12n2n+1=n=1∞14n2+2n

is 4n2 so we compare it with the series

n=1∞14n2 .

Observe that
14n2+2n<14n2 for all n≥1

because the left side has bigger denominator. We know that

n=1∞14n2=14 n=1∞1n2

is convergent because it is a constant times a p-series with p=2>1 .

Therefore, by Comparison Test,

n=1∞12n2n+1

is also convergent.
2
EXERCISE 5.6.1

3
(a) If sinx can be represented as a power series about 0, then sinx is
equal to the sum of its Maclaurin series

fx=n=0∞fn(0)n!xn=f0+f'(0)1!x+f''02!x2+⋯

Taking the derivatives of sinx and their corresponding values at 0, we

obtain

fx=sinx f0=0
f'(x)=cosx f'(0)=1
f''x=-sinx f''0=0
f'''x=-cosx f'''0=-1
f4x=sinx f40=0

Since the derivatives repeat in a cycle of four, we can write the Maclaurin
series as follows:

f0+f'(0)1!x+f''02!x2+f'''03!x3+…=x-x33!+x55!-x77!+⋯

=n=0∞(-1)nx2n+12n+1!

To check whether this series converges for all x or not, we can make use
of Taylor’s inequality which states that if fn+1(x)≤M for x-a≤d, then the
remainder Rn(x) of the Taylor series satisfies the inequality

Rnx≤Mn+1!x-an+1 for x-a≤d.

Now since fn+1(x) is ±sinx or ±cosx, we know that fn+1(x)≤1 for all x. So we
can take M=1, a=0 in Taylor’s Inequality:

Rnx≤xn+1n+1!
However,
limn→∞xnn!=0 for every real number x

so the right side of this inequality approaches 0 as n→∞, and by the

Squeeze Theorem, Rnx→0. It follows that Rnx→0 as n→∞, so sinx is equal to
the sum of its Maclaurin series. Indeed,

sinx=x-x33!+x55!-x77!+⋯

=n=0∞(-1)nx2n+12n+1! for all x

(b) To find the Maclaurin series for cosx, we could proceed directly as in
(a) but it’s easier to differentiate the Maclaurin series for sinx given above.
That is,

cosx=ddxsinx=ddxx-x33!+x55!-x77!+⋯

=1-3x23!+5x45!-7x67!+⋯=1-x22!+x44!-x66!+⋯

4
Since the Maclaurin series for sinx converges for all x, the differentiated
series for cosx also converges for all x. Thus,

cosx=1-x22!+x44!-x66!+⋯

=n=0∞(-1)nx2n2n! for all x

EXERCISE 5.7.1

There are two equivalent ways to find the power series expansion of
f(x)g(x)=coshxsinhx .

METHOD 1
Assuming that fx=coshx is given by its Maclaurin series, we expand coshx
into such a series. In taking the derivatives of coshx and their
corresponding values at 0, we rely on the fact that

coshx=12ex-e-x and sinhx=12ex+e-x.

But e0=e-0=1, so the values of the hyperbolic sine and cosine functions at 0
reduce to
cosh0=121-1=0 and sinh0=121+1=1.

Hence, we have
fx=coshx f0=1
f'(x)=sinhx f'(0)=0
f''x=coshx f''0=1
f'''x=sinhx f'''0=0
f4x=coshx f40=1

f5x=sinhx f50=0

f6x=coshx f60=1

from which we obtain the Maclaurin series expansion

f0+f'01!x+f''02!x2+f'''03!x3+f404!x4+f505!x5+f606!x6+…=1+01!x+12!x2+03!x3+14!x4+05!

x5+16!x6+…

fx=coshx=1+12!x2+14!x4+16!x6+…=n=0∞x2n2n!

5
Similarly, we assume that gx=sinhx is also given by its Maclaurin series.
Thus, we find
gx=sinhx g0=0
g'(x)=coshx g'(0)=1
g''x=sinhx g''0=0
g'''x=coshx g'''0=1
g4x=sinhx g40=0

g5x=coshx g50=1
g6x=sinhx
g60=0
g7x=coshx
g70=1

which corresponds to the Maclaurin series expansion

g0+g'01!x+g''02!x2+g'''03!x3+g404!x4+g505!x5+g606!x6+g707!x7+…=0+11!x+02!x2+13!

x3+04!x4+15!x5+06!x6+17!x7+…

gx=sinhx=x+13!x3+15!x5+17!x7+…=n=0∞x2n+12n+1!

From these Maclaurin series expansions of the hyperbolic sine and cosine
functions, we can therefore write the ratio of coshx to sinhx as

fxgx=coshxsinhx=1+12!x2+14!x4+16!x6+…x+13!x3+15!x5+17!x7+… .

Now in principle, power series behave like polynomials when added or

subtracted, even when multiplied or divided. Thus, we may perform long
division to evaluate the above expression. In our case however, we find
only the first few terms because the calculations for the later terms
become tedious and the initial terms are the most important ones.

1x + x3 -x345+2x5945+⋯

x+x33!+x55!+x77!+… 1+x22!+x44!+x66!+…

1+x23!+x45!+x67!+⋯

x23+x430+x6840+⋯

x23+x418+x6360+⋯

-x445-x6630+⋯

-x445-x6270+⋯

2x6945+⋯

fxgx=coshxsinhx=x-1+13x-145x3+2945x5+⋯
6
METHOD 1
Since by definition,

coshx=12ex-e-x and sinhx=12ex+e-x,

it follows that
coshxsinhx=12ex+e-x12ex-e-x=ex+e-xex-e-x

=ex+1exex-1ex=e2x+1exe2x-1ex

=e2x+1e2x-1 .

Hence, we can obtain the power series expansion of the ratio

f(x)g(x)=coshxsinhx using the Maclaurin series expansion of the exponential
function given by

ex=n=0∞xnn!=1+x+x22!+x33!+x44!+x55!+⋯
That is, we can write

e2x+1e2x-1=1+21!x+222!x2+233!x3+244!x4+255!x5+266!x6+277!x7+⋯+11+21!x+222!x2+233!

x3+244!x4+255!x5+266!x6+277!x7+⋯-1

=+1+1+2x+2x2+86x3+1624x4+32120x5+64720x6+1285040x7+⋯-

1+1+2x+2x2+86x3+1624x4+32120x5+64720x6+1285040x7+⋯

=2+2x+2x2+43x3+23x4+415x5+445x6+8315x7+⋯2x+2x2+43x3+23x4+415x5+445x6+8315x7+

Again, by long division

1x+13x -145x3 +2945x5

2x+2x2+43x3+23x4+415x5+445x6+8315x7+⋯ 2+2x+2x2+43x3+23x4+415x5+445x6+⋯

2+2x+43x2+23x3+415x4+445x5+8315x6+⋯

23x2+23x3+25x4+845x5+463x6+⋯

23x2+23x3+49x4+29x5+445x6+⋯

-245x4-245x5-8315x6+⋯

-245x4-245x5-4135x6+⋯

4945x6+⋯

we find that indeed

fxgx=coshxsinhx=e2x+1e2x-1=x-1+13x-145x3+2945x5+⋯

7
Finally, we expand the function hx=1x into a Taylor series about x=1 ,
assuming that such an expansion is valid. That is, we have

hx=x-1 h1=1
h'x=(-1)x-2 h'1=-1
h''x=(-1)(-2)x-3 h''1=-12∙2!
h'''x=(-1)(-2)(-3)x-4 h'''1=-13∙3!
h4x=(-1)(-2)(-3)(-4)x-5 hn1=-14∙4!

and so the Taylor series expansion of 1x with a=1 is

n=0∞hnan!x-an=1-11!x-1+2!2!x-12-3!3!x-13+4!4!x-14+⋯

x-1=1-x-1+x-12-x-13+x-14-x-15+⋯

hx=x-1=n=0∞-1nx-1n

Upon substitution of our derived series expansions into the given

expression for magnetic polarization, we obtain

Px=ccoshxsinhx-1x

=c1x+13x-145x3+2945x5+⋯-c1-x-1+x-12-x-13+x-14-x-15+⋯

=c1x+13x-145x3+2945x5+⋯-c1-x-1+x2-2x+1-x3-3x2+3x-1+x4-4x3+6x2-4x+1-x5-5x4+10x3-

10x2+5x-1+⋯

=c1x-1+1+1+1+1+1+x13+1+2+3+4+5-x21+3+6+10-x3145-1-4-10-x41+5+x52945+1

=c1x-6+463 x-20 x2+67445 x3-6x4+947945 x5+⋯

Therefore, the magnetic polarization

Px=c coshxsinhx-1x

can be written in power series expansion as

Px=c 1x-6+463 x-20 x2+67445 x3-6x4+947945 x5+⋯ .

8
EXERCISE 5.7.11

I NVERSION OF P OWER S ERIES

In Exercise 5.7.1, we found the Maclaurin series expansion of the
hyperbolic sine to be

x=sinhy=n=0∞y2n+12n+1!=y+13!y3+15!y5+17!y7+… .

To develop a series expansion of y=sinh-1x by series inversion, we assume

that

y=sinh-1x=n=0∞bnxn=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6+b7x7+⋯

where
x=y+13!y3+15!y5+17!y7+… .

It follows then that xn have the following values for n=2, 3, 4, 5, 6, 7 :

y +y33!+ y55! + y77! + y99! + y1111! + y1313! +y1515! +y1717! +…

y +y33!+ y55! + y77! + y99! + y1111! + y1313! +y1515! +y1313! +…

y2+y43!+ y65! + y87! + y109! + y1211! + y1413! +y1615! +y1817! +…

y43!+y63!3!+y83!5!+y103!7!+y123!9!+y143!11!+y163!13!+y183!15!+…

y65! +y85!3!+y105!5!+y125!7!+y145!9! +y165!11!+y185!13!+…

y87! +y107!3!+y127!5!+y147!7! +y167!9! +y187!11!+…

y109! +y129!3!+y149!5! +y169!7! +y189!9! +…

y1211! +y1411!3!+y1611!5!+y1811!7!+…

y1413! +y1613!3!+y1813!5!+…

y1615! +y1815!3!+…

y2+y43+2y645+
+y8315+2y1014175+2y12467775+4y1442567525+y16638512875+2y1897692469875+…

9
y34126513546505547170185216000000+y32232561666370491121664000000+173y30116280833
185245560832000000+107y28276859126631537049600000+3673y2646143187771922841600000
+1187y2488736899561390080000+27553y2214789483260231680000+3449y20160059342643200
00+2y1897692469875+y16638512875+4y1442567525+2y12467775+2y1014175+y8315+2y645+y
43+y2

y=b0+b1y+y36+y5120+y75040+…+b2y2+y43+2y645+⋯+b3y4+2y63+⋯

Collecting like terms just as for polynomials, we get

y=b0+b1y+b2y2+b16y3+b23+b3y4+b1120y5+2b245+2b33y6+b15040y7+⋯

D IRECT M ACLAURIN E XPANSION

Since the hyperbolic functions are defined in terms of exponential
functions, it’s not surprising to learn that the inverse hyperbolic functions
can be expressed in terms of logarithms. In particular, if we let y=sinh-1x ,
then

x=sinhy=ey-e-y2
so
ey-2x-e-y=0
or, multiplying by ey,
e2y-2xey-1=0 .

This is really a quadratic equation in ey :

ey2-2xey-1=0 .

Solving by the quadratic formula, we get

ey=2x±4x2+42=x±x2+1 .

Note that ey>0, but x-x2+1 <0 (because x<x2+1 ). Thus, the minus sign is
inadmissible and we have
ey=x+x2+1 .

Taking the logarithm of both sides, we obtain

lney=lnx+x2+1
or
y=sinh-1x=lnx+x2+1.

Because the hyperbolic functions are differentiable, the inverse hyperbolic

functions are all differentiable. This implies that if we let y=sinh-1x , it would
be legitimate to find y', y'', y''', etc.. Doing so gives us the following
derivatives and their corresponding values at x=0 :

y=lnx+x2+1 y(0)=0

y'=1x2+1 y'(0)=1

y''=-xx2+132
10
y'''=3x2x2+152-1x2+132 y''(0)=0

y(4)=-15x3x2+172+9xx2+152
y'''0=-1
y(5)=105x4x2+192-90x2x2+172+9x2+152

y(6)=1050x3x2+192-945x5x2+1112-225xx2+172 y(4)(0)=0

y(7)=4725x2x2+192+10395x6x2+1132-14175x4x2+1112-225x2+172
y(5)(0)=9

y(6)(0)=0

y70=-225

From the first few yn(0) values calculated above, we can infer that only
odd derivatives of the inverse hyperbolic sine function survive, the rest
vanish. Substituting these values to the general form of the Maclaurin
series expansion, we obtain

y=x-13!x3+95!x5-2257!x7+⋯

=x-12x33+3∙32∙3∙4x55-3∙3∙5∙52∙3∙4∙5∙6x77+⋯

=x-12x33+1∙32∙4x55-1∙3∙52∙4∙6x77+⋯

=n=0∞-1n2n!22nn!2x2n+12n+1 , x<1

Therefore, the inverse hyperbolic sine function can be expressed as

sinh-1x=x-13!x3+95!x5-2257!x7+⋯=n=0∞-1n2n!22nn!2x2n+12n+1 , x<1

11
PROBLEM 5
Solve the equation y''+y=0 .

The given differential equation can be solved using power series. That is, we
assume there is a solution of the form
(1-
y=c0+c1x+c2x2+c3x3+⋯=n=0∞cnxn
a)

We can differentiate power series term by term, so

y'=c1+2c2x+3c3x2+⋯=n=1∞ncnxn-1

y''=2c2+2∙3c3x+⋯=n=2∞nn-1cnxn-2

(1-
b)
In order to compare the expressions for y and y'' more easily, we rewrite y''
as follows:

(1-
y''=n=0∞n+2n+1cn+2xn
c)

Substituting the expressions in Equations (1-a) and (1-c) into the differential
equation, we obtain

12
n=0∞n+2n+1cn+2xn+n=0∞cnxn=0

n=0∞n+2n+1cn+2+cnxn=0

(1-
d)
If two power series are equal, then the corresponding coefficients must be
equal. Therefore, the coefficients of xn in Equation (1-d) must be 0:

n+2n+1cn+2+cn=0 (1-
cn+2=-cnn+1n+2 n=0, 1, 2, 3, … e)

If c0 and c1 are known, the recursion relation (1-e) allows us to determine the
remaining coefficients recursively by putting n=0, 1, 2, 3, … in succession.

Put n=0 : c2=-c01∙2

Put n=1 : c3=-c12∙3

Put n=2 : c4=-c23∙4=c01∙2∙3∙4=c04!

Put n=3 : c5=-c34∙5=c12∙3∙4∙5=c15!

Put n=4 : c6=-c45∙6=-c04!5∙6=-c06!

Put n=5 : c7=-c56∙7=-c15!6∙7=-c17!

By now we see the pattern:

For the even coefficients, c2n=-1nc02n!

For the odd coefficients, c2n+1=-1nc12n+1!

Putting these values back into Equation (1-a), we write the solution as

y=c0+c1x+c2x2+c3x3+c4x4+c5x5+⋯

=c01-x22!+x44!-x66!+⋯+-1nx2n2n!+⋯+c1x-x33!+x55!-x77!+⋯+(-1)nx2n+12n+1!+⋯

=c0n=0∞(-1)nx2n2n!+c1n=0∞(-1)nx2n+12n+1!

where c0 and c1 arbitrary constants.

Remark: We recognize the series above as being the Maclaurin series for cosx
and sinx which we have derived in Exercise 5.6.1. Therefore, we could write the
solution as

y=x=c0cosx+c1sinx

13