You are on page 1of 7

SOME INFINITE SERIES

H. f. SANDHAM

1. We sum eight

general

infinite

series

[l],

particular

which are
Is

33

cosh (x/2)

cosh (3x/2)

cosh (5x/2)

73

. . . = 0,

-+

53

cosh (7x/2)

l6cosh(x/2)

36cosh(3x/2)

56 cosh (5x/2)

x6

1.21

"

l6

35

56

er + 1

e3* + 1

eiT + 1

76

31

+-H-=

e7- + 1

1 22

~ 768 '

tanh (x/2)

tanh (3x/2)

l3

33

504

tanh (5x/2)
53

tanh (7x/2)

x3

+--1-=

>

73

1.31

32

sinh x

sinh 2x

sinh 3x

-+

...=

sinh 4x

1.32

l6 sinh x

25 sinh 2x

35 sinh 3x

x3

46sinh4x

1.41

4x

Z-=

w13

i e2 - 1

" 360'
1

'
24

430

cases of

some infinite series

1.42

431

"

coth xw

19tt7

56700

E-

Noting that

e + 1

=-tanh

>

-=

llxl

coth-,

e* - 1

it will readily be seen that each of the above four pairs of series
contains one of the hyperbolic functions: sech, tanh, cosech, coth,
multiplied by a positive and by a negative power. If we include the
remaining
two functions:
sinh and cosh, we have of course two
pairs of series diverging to infinity.

The general series are (3.22), (3.32); (4.22), (4.32); (5.22), (5.32);
(6.222), (6.32).
2. The bases of the proofs are the partial

e2TX 1

2xx

2.11

-=-+

2.12

-=-
e2rx _

e2rax

1
i

1 -

"

x2 + n2

(* cos 2nira

+ El-\
x i ( x2 +

2tx

expansions

E -'
x

fraction

n sin 2nira)

n2

x2 +

n2 )

(0 < a < 1),

from which we deduce

2.21

-=

certain

12*1

e*x + 1

+ E'-'
2

sinh irax

2.22

;-

sinh 7rx

erax + eT(l-a)x

2.23 -

other partial

=-2-i
4

"

fraction

*
(2n + l)2 -

x2

( \)nn

-sm

lran

x2 + n2

(2w +

ir

1)

= E-sin

e" + 1

expansions

ira(2n + 1)

x2 + (2 + l)2

cosh (itax/2)
4 (-1)(2+1)
2.24-
= E-cosh (xx/2)

x2 + (2w + l)2

(-

cosh (irax/2)

--

cos (wax/2)

"

4n

(2 + l)4 - x4

= E -'

432

H. F. SANDHAM

Multiplying

[June

An and find
cosh (rax/2)

cos (rax/2)

cosh (xx/2)

cos (xx/2)

8 (-1)(2
x

(2 + l)4 - x4

(-1

< a < 1).

Putting x = 0 yields
8

x
therefore,

(-l)n

o 2w + 1

equating

coefficients

3.22

of aim,

i-l)n(2n+

l)4m+3

22-

=0

(w h 0).

The case m = 0 is series (1.11).

Putting a = 0 yields

1
cosh (xx/2)

3 31

cos (xx/2)

_8_ -

(-1)(2+

'no

(2n+

l)3

1_

'

Writing
1

cosh x

1!

then, with the understanding

En+(n\

En-lEia

we have, on equating

3 32

x2

x3

= 1 + Ei + E2 + Ez + ,

2!

3!

that

+ ( " ) En-2E2CC2 +=(

coefficients

(-1)"

+ ),

of x4m,
1

V_._=_(_)
o (2m + 1)4OT+1cosh (x(2 + l)/2)

x/x\4m

iE + Ei)im

8 V2 /

(m > 0).
The case m = \ is series (1.12).

i954]

SOME INFINITE SERIES

433

4. Similarly, we have
gTax

I gir(la)x

giraxi

e" + 1

4.1

(e*a(2n+l)

_|_ g*(la)xi

e*xi +1

_|_ gT(l-o)(2n+l)\(-gTo(2n+l)i

--

gT(2n+l)

00

(2n

l)3

(2w + l)4 -

_J_ gx(l-<i)

x4

(2n+l)i\

(0 < a < 1).

_L. J

Putting x = 0 yields
4

2+l

"iri

/gira(2n+l)

_|_ gir(l-a)

(2n+l))

(gira(2n+l)

i _|_ gir(l-a)

(2n+l)i)

e2x(+i) + !

(0 < a < 1).

Differentiating

with respect

"

sinh xa(2w+l)

"

to a, this becomes

cos xa(2w+l)+cosh
g-(2n+l)

xa(2+l)

sin xa(2+l)

+ 1

4.21

1/
= (-)

1-i

1+t

Ai \gir(1*)_gira(li)

^xafl+i)_g-to

(0<a<l).

(1+0 /

Writing
/y-

<V*

V"

WV

_-rT=1

-V"

*v

VV

+ BlTi + A_ + B.-+...,

then

1-i
gB(l-i)

1+ i
g-(l-)

2i/

gK(l+i)

y2

= (2732(2-

1)-

y \

g-y(.l+i)

237J6(25-

y6

y10

6!

10!

1) + 267310(29- 1)-

2!

),

and since
y

y5

1!

5!

y9

25-

91

therefore
-

4.22

(2n +

E -^-r-

l)4m+1

*(**+-
+ 1

7i4OT+2

=-t

2 (4w + 2)!

The case w = l is formula (1.21).

We also have

(24M+1- 1)

(w ^ 0)-

v -

434

H. F. SANDHAM

(_J__iV_L__l')
\e*x +1

4.31

2 / \6'

[June

+1

2/

2i

x2(2m + 1)

= - Z TT-V^:tanh (<2
+ lWx

(2w + l)4 -

x4

Writing
-

x2

x3

1!

2!

3!

= 1 + Di + D2 + D3 + ,

ex+l

so that
Pn+l

Z)n = (2 -

2+2) _,
M+ 1

then
"

4.32

tanh (x(2 + l)/2)

V-=-

x4'"+3 (D + Di)im+2

(2m + 1)4M+3

8i

(4m+

2)!

(m 5: 0).

5. Again, from
sinh rax sin xax

5.1--:-=

sinh xx sin xx

we have,

on letting

5.21

- Z V^

--

x4 w4

sinh xw

(-Ka<

1),

x+0,
4

"

m sinh xm

a2 =-Z(-l)n-

(-l<a<l).

Since
<y2

<y6

-y10

2!

6!

10!

therefore
M

(_

l)n/J4m+1

Z r:-=
i

Equating
Dividing

sinh xm

o,

(m > 0).

coefficients of a2, we have the case m = 0, i.e. series (1.31).

across by a and letting a>0, we have
X2

"

(-1)"M6

sinh xx sin xx

x4 m4 sinh xm

1954]

435

Writing
X

sinh x

1!

=i+c1-

X2

+ c,-

2!

X3

+ c,-+...

3!

5.32

(-1)

n4"1-1 sinh x

E -"-"-

x4'"-1

(C + Ci)im

(4w)!

=-"-

( > 0).

The case w = 1 is series (1.32).

6. Lastly, we have
o2irai

2x(la)x

g2raii

e2xx 1

g2x(la)xi

e2Txi 1

xi

4*3

x4 -

6.1
fg2ran

a2r{\a)

n\(p2v

a ni

n(e2 -

g2ir(la)

7ii\

1)

9

oo

(pi-van

g2x(l

(2a - l)2 = - E-,

x

Differentiating

a) n\ fnl-sani

n(e2rn

with respect
"

4(2a - 1) = - 16E-

g2x(la)ni\

(0<a<

1)

1).

to a, this becomes

6.21

,,-L

e2 -

/
1+ i
+ 4 --I-)
\g2xa(l-i)

1- i
I

g2xo(l+>)

y5

\
\J

(0 < a < 1).

Since
sin y cosh y + cos y sinh y = 2-23-1-25-

1!

51

v9

9!

and

1+ i
g(l-i)

1- i
1

g(l+i) _

- -

therefore

1+

2 /

y2

y \

2!

[27i8 -

y6

y10

6!

10!

23756 + 267310--

),

436

h. f. sandham

6.221

Am
-=-,

11

6.222

23 -=-1

i e2 - 1

"

24

8x

Bim+2

nim+1

i e2 - 1

(m > 0).

2 4w + 2

The case m = 3 is series (1.41).

We also have

Ve2 -

2 / \e2rxi -

2/

6.31

= -
4x2x2i

H-;

1 "

xi

2^ i

cth XM,

x4 m4

therefore

6.32

"

coth xm

m4"1

Z -

(2X)4"1 (B + Bi)im

=-A-^-^-
2

(4w)!

(>0).

The case m = 2 is series (1.42).

References
1. S. Ramanujan,
series (1.41), (1.42),

Collected papers of Srinivasa Ramanujan, Cambridge, 1927. The

(1.12) are theorems (4), (5), (6), page xxxv, and the series

(3.22), (6.221) are questions 358, 387, page 326. Although he only stated some particular cases there is no doubt that all the general formulae of this paper were known
to Ramanujan.
Those familiar with his work will recall his preference for stating
most interesting particular cases rather than writing out general formulae.
Other proofs of some of these series have been given:

2. G. N. Watson, J. London Math. Soc. vol. 3 (1928) pp. 216-225.

3. G. H. Hardy, J. London Math. Soc. vol. 3 (1928) pp. 238-240.
4. H. F. Sandham, Quart. J. Math. (2) vol. 1 (1950) pp. 238-240.
Dublin Institute