a

r
X
i
v
:
0
7
1
2
.
1
3
3
2
v
2


[
m
a
t
h
.
N
T
]


1
9

M
a
y

2
0
0
8
Ramanujan-type formulae for 1/:
A second wind?

Wadim Zudilin
14 May 2008
Abstract
In 1914 S. Ramanujan recorded a list of 17 series for 1/. We
survey the methods of proofs of Ramanujans formulae and indicate
recently discovered generalizations, some of which are not yet proven.
The twentieth century was full of mathematical discoveries. Here we ex-
pose two signicant contributions from that time, in reverse chronological
order. At rst glance, the stories might be thought of a dierent nature. But
we will try to convince the reader that they have much in common.
1 Ramanujan and Apery: 1/ and (3)
In 1978 R. Apery showed the irrationality of (3) (see [5] and [21]). His ratio-
nal approximations to the number in question (known nowadays as Aperys
constant) have the form v
n
/u
n
Q for n = 0, 1, 2, . . . , where the denomina-
tors {u
n
} = {u
n
}
n=0,1,...
and numerators {v
n
} = {v
n
}
n=0,1,...
satisfy the same
polynomial recurrence
(n + 1)
3
u
n+1
(2n + 1)(17n
2
+ 17n + 5)u
n
+ n
3
u
n1
= 0 (1.1)
with the initial data
u
0
= 1, u
1
= 5, v
0
= 0, v
1
= 6.

The work was supported by a fellowship of the Max Planck Institute for Mathematics
(Bonn) and supported in part by the INTAS foundation, grant no. 03-51-5070.
2000 AMS subject classication: 11F11, 11Y60, 33C20 (primary), 05A19, 11B65, 11J82,
11M06, 14H52, 14J32, 33C75, 33F10, 34M50, 40G99, 65B10, 65Q05 (secondary).
1
Ramanujan-type formulae for 1/: A second wind? 2
Then
lim
n
v
n
u
n
= (3)
and, surprisingly, the denominators {u
n
} are integers:
u
n
=
n

k=0
_
n
k
_
2
_
n + k
k
_
2
Z, n = 0, 1, 2, . . . , (1.2)
while the numerators {v
n
} are close to being integers.
In 1914 S. Ramanujan [22], [7] recorded a list of 17 series for 1/, from
which we indicate the simplest one

n=0
(
1
2
)
3
n
n!
3
(4n + 1) (1)
n
=
2

(1.3)
and also two quite impressive examples

n=0
(
1
4
)
n
(
1
2
)
n
(
3
4
)
n
n!
3
(21460n + 1123)
(1)
n
882
2n+1
=
4

, (1.4)

n=0
(
1
4
)
n
(
1
2
)
n
(
3
4
)
n
n!
3
(26390n + 1103)
1
99
4n+2
=
1
2

2
(1.5)
which produce rapidly converging (rational) approximations to . Here
(a)
n
=
(a + n)
(a)
=
_
a(a + 1) (a + n 1) for n 1,
1 for n = 0,
denotes the Pochhammer symbol (the rising factorial). The Pochhammer
products occurring in all formulae of this type may be written in terms of
binomial coecients:
(
1
2
)
3
n
n!
3
= 2
6n
_
2n
n
_
3
,
(
1
3
)
n
(
1
2
)
n
(
2
3
)
n
n!
3
= 2
2n
3
3n
_
2n
n
_
(3n)!
n!
3
,
(
1
4
)
n
(
1
2
)
n
(
3
4
)
n
n!
3
= 2
8n
(4n)!
n!
4
,
(
1
6
)
n
(
1
2
)
n
(
5
6
)
n
n!
3
= 12
3n
(6n)!
n!
3
(3n)!
.
Ramanujans original list was subsequently extended to several other series
which we plan to touch on later in the paper. For the moment we give two
more celebrated examples:

n=0
(
1
3
)
n
(
1
2
)
n
(
2
3
)
n
n!
3
(14151n + 827)
(1)
n
500
2n+1
=
3

, (1.6)

n=0
(
1
6
)
n
(
1
2
)
n
(
5
6
)
n
n!
3
(545140134n + 13591409)
(1)
n
53360
3n+2
=
3
2

10005
. (1.7)
Ramanujan-type formulae for 1/: A second wind? 3
Formula (1.6) is proven by H. H. Chan, W.-C. Liaw and V. Tan [13] and
(1.7) is the Chudnovskys famous formula [15] which enabled them to hold
the record for the calculation of in 198994. On the left-hand side of each
formula (1.3)(1.7) we have linear combinations of a (generalized) hypergeo-
metric series
m
F
m1
_
a
1
, a
2
, . . . , a
m
b
2
, . . . , b
m

z
_
=

n=0
(a
1
)
n
(a
2
)
n
(a
m
)
n
(b
2
)
n
(b
m
)
n
z
n
n!
(1.8)
and its derivative at a point close to the origin. The rapid convergence of the
series in (1.4)(1.7) may be used for proving the quantitative irrationality of
the numbers

d with d N (see [26] for details).

In both Ramanujans and Aperys cases, there were just hints on how
the things might be proven. Rigorous proofs appeared somewhat later. We
will not discuss proofs of Aperys theorem and its further generalizations (see
[17] for a review of the subject), just concentrating on the things around the
remarkable Ramanujan-type series. But we will see that both Ramanujans
and Aperys discoveries have several common grounds.
2 Elliptic proof of Ramanujans formulae
Although Ramanujan did not indicate how he arrived at his series, he hinted
that these series belong to what is now known as the theories of elliptic
functions to alternative bases. The rst rigorous mathematical proofs of
Ramanujans series and their generalizations were given by the Borweins [11]
and Chudnovskys [15]. Let us sketch, following [15], the basic ideas of those
very rst proofs.
One starts with an elliptic curve y
2
= 4x
3
g
2
x g
3
over Q with funda-
mental periods
1
,
2
(where Im(
2
/
1
) > 0) and corresponding quasi-periods

1
,
2
. Besides the Legendre relation

2

2

1
= 2i,
the following linear relations between
1
,
2
,
1
,
2
over Q are available in the
complex multiplication case, i.e., when =
2
/
1
Q[

d] for some d N:

2

1
= 0, A
2
C
1
+ (2A + B)
1
= 0, (2.1)
where the integers A, B and C come from the equation A
2
+ B + C = 0
dening the quadratic number and Q(, g
2
, g
3
) Q. Equations (2.1)
allow one to express
2
,
2
by means of
1
,
1
only. Substituting these ex-
pressions into (2.1), and using the hypergeometric formulae for
1
,
1
and
also for
2
1
,
1

1
(which follow from Clausens identity) one nally arrives at
Ramanujan-type formulae for 1/: A second wind? 4
a formula of Ramanujan type. An important (and complicated) problem in
the proof is computing the algebraic number
=
8
2
81
2
1
_
E
2
()
3
Im
_
, where E
2
() = 1 24

n=1
e
2in

d|n
d.
Note that viewed as a function of is a non-holomorphic modular form of
weight 2. Although the Chudnovskys attribute the knowledge of the fact that
() takes values in the Hilbert class eld Q(, j()) of Q() to Kronecker
(in Weils presentation [24]), we would refer the reader to the work [8] by
B. C. Berndt and H. H. Chan.
3 Modular proof of Ramanujans formulae
An understanding of the complication of the above proof came in 2002 with
T. Satos discovery of the formula

n=0
u
n
(20n + 10 3

5)
_

5 1
2
_
12n
=
20

3 + 9

15
6
(3.1)
of Ramanujan type, involving Aperys numbers (1.2). The modular argument
was essentially simplied by H. H. Chan with his collaborators and later by
Y. Yang to produce a lot of new identities like (3.1) based on a not necessarily
hypergeometric series F(z) =

n=0
u
n
z
n
. Examples are

n=0
n

k=0
_
n
k
_
2
_
2k
k
__
2n 2k
n k
_
(5n + 1)
1
64
n
=
8

3
(3.2)
due to H. H. Chan, S. H. Chan and Z.-G. Liu [12];

n=0
[n/3]

k=0
(1)
nk
3
n3k
(3k)!
k!
3
_
n
3k
__
n + k
k
_
(4n + 1)
1
81
n
=
3

3
2
(3.3)
due to H. H. Chan and H. Verrill (2005);

n=0
n

k=0
_
n
k
_
4
(4n + 1)
1
36
n
=
18

15
(3.4)
due to Y. Yang (2005).
It should be mentioned that PicardFuchs dierential equations (of or-
der 3) satised by the series F(z) always have very nice arithmetic proper-
ties [25]. Therefore, it is not surprising that F(z) admits a modular parametri-
zation: f() = F(z()) is a modular form of weight 2 for a modular (uni-
formizing) substitution z = z().
Ramanujan-type formulae for 1/: A second wind? 5
Let us follow Yangs argument to show the basic ideas of the new proof
in the example of (3.1). Our choice is
z() =
_
()(6)
(2)(3)
_
12
, f() =
(2)
7
(3)
7
()
5
(6)
5
,
which are modular forms of level 6; the expressions were obtained by F. Beuk-
ers in his proof of Aperys theorem using modular forms [9]. Here
() = e
i/12

n=1
(1 e
2in
)
is the Dedekind eta-function. The function g() = (2i)
1
f

()/f() satises
the functional equation
g() =
c(c + d)
i
+ (c + d)
2
g() for =
_
a b
c d
_

0
(6) + w
6
,
where w
6
denotes the AtkinLehner involution. Taking
=
1

6
_
0 1
6 0
_
, =
0
=
i

30
we obtain
g(
0
) + 5g(5
0
) =

30

. (3.5)
On the other hand, h() = g() 5g(5) is a modular form of weight 2 and
level 30 (on
0
(30) + w
5
, w
6
). This implies that h()/f() is an algebraic
function of z() and after explicit evaluations at =
0
we arrive at
g(
0
) 5g(5
0
) = h(
0
) =
900

2 402

10
5
f(
0
) = (900

2402

10)f(5
0
).
(3.6)
Combining (3.5) and (3.6) we deduce that

30

= (900

2 402

10)f(5
0
) + 10g(5
0
),
and it only remains to use the expansion
g(5
0
) = z
df/dz
f

1
2i
z

()
z()

=5
0
= (108

2 48

10)

n=0
nu
n
z(5
0
)
n
(since z

()/(2i) and f() are modular forms of weight 2 on

0
(6) +w
6
, the
function
1
f

1
2i
z

z
Ramanujan-type formulae for 1/: A second wind? 6
is an algebraic function of z) and the evaluation
z(5
0
) = z(
0
) = 161 72

5 =
_

5 1
2
_
12
.
As pointed out to us by H. H. Chan, the main diculty one meets in the
above proof is to prove the algebraicity evaluations rigorously (cf. [8]).
4 Creative telescoping
There is yet another method of proof, but applicable only to a small number
of Ramanujan-type series. It is based on the algorithm of creative telescoping,
due to GosperZeilberger. Note that an essential part of the rst proof of
Aperys theorem [21], namely, the proof of the recurrence (1.1), was given by
D. Zagier also using a telescoping argument. D. Zeilberger (and his automatic
collaborator S. B. Ekhad) could prove the simplest Ramanujans identity (1.3)
in the following way [16]. One veries the (terminating) identity

n=0
(1/2)
2
n
(k)
n
n!
2
(3/2 + k)
n
(4n + 1)(1)
n
=
(3/2 + k)
(3/2)(1 + k)
(4.1)
for all non-negative integers k. To do this, divide both sides of (4.1) by the
right-hand side and denote the summand on the left by F(n, k):
F(n, k) = (4n + 1)(1)
n
(1/2)
2
n
(k)
n
n!
2
(3/2 + k)
n
(3/2)(1 + k)
(3/2 + k)
;
then take
G(n, k) =
(2n + 1)
2
(2n + 2k + 3)(4n + 1)
F(n, k)
with the motive that F(n, k + 1) F(n, k) = G(n, k) G(n 1, k), hence

n
F(n, k) is a constant, which is seen to be 1 by plugging in k = 0. Finally,
to deduce (1.3) one takes k = 1/2, which is legitimate in view of Carlsons
theorem [6, Section 5.3].
5 Guilleras series for 1/
2
If one wishes to use the latter method of proof for other Ramanujan-type
formulae, ingenuity is required in order to put the new parameter k in the
right place. This was done only recently by J. Guillera [18, 20], who used the
method to prove some other identities of Ramanujan (in those cases when z
Ramanujan-type formulae for 1/: A second wind? 7
has only 2 and 3 in its prime decomposition). If the reader doubts the appli-
cability of the method, then take into account that the purely hypergeometric
origin of the method and its independence from the elliptic and modular stu
allowed Guillera [18, 19, 20] to prove new generalizations of Ramanujan-type
series, namely,

n=0
(
1
2
)
5
n
n!
5
(20n
2
+ 8n + 1)
(1)
n
2
2n
=
8

2
, (5.1)

n=0
(
1
2
)
5
n
n!
5
(820n
2
+ 180n + 13)
(1)
n
2
10n
=
128

2
, (5.2)

n=0
(
1
2
)
3
n
(
1
4
)
n
(
3
4
)
n
n!
5
(120n
2
+ 34n + 3)
1
2
4n
=
32

2
, (5.3)
and also to nd experimentally [19] four additional formulae

n=0
(
1
2
)
n
(
1
4
)
n
(
3
4
)
n
(
1
6
)
n
(
5
6
)
n
n!
5
(1640n
2
+ 278n + 15)
(1)
n
2
10n
=
256

3
3
2
, (5.4)

n=0
(
1
2
)
n
(
1
4
)
n
(
3
4
)
n
(
1
3
)
n
(
2
3
)
n
n!
5
(252n
2
+ 63n + 5)
(1)
n
48
n
=
48

2
, (5.5)

n=0
(
1
2
)
n
(
1
3
)
n
(
2
3
)
n
(
1
6
)
n
(
5
6
)
n
n!
5
(5418n
2
+ 693n + 29)
(1)
n
80
3n
=
128

2
, (5.6)

n=0
(
1
2
)
n
(
1
8
)
n
(
3
8
)
n
(
5
8
)
n
(
7
8
)
n
n!
5
(1920n
2
+ 304n + 15)
1
7
4n
=
56

2
. (5.7)
As Guillera notices, the series in (5.5)(5.7) are closely related to the series

n=0
(
1
2
)
n
(
1
4
)
n
(
3
4
)
n
n!
3
(28n + 3)
(1)
n
48
n
=
16

3
,

n=0
(
1
2
)
n
(
1
6
)
n
(
5
6
)
n
n!
3
(5418n + 263)
(1)
n
80
3n
=
640

15
3
,

n=0
(
1
2
)
n
(
1
4
)
n
(
3
4
)
n
n!
3
(40n + 3)
1
7
4n
=
49
3

3
,
respectively, proven by the methods in Sections 2 and 3. However, there is
no obvious way to deduce any of formulae (5.1)(5.7) by modular means; the
problem lies in the fact that the (Zariski closure of the) projective monodromy
group for the corresponding series F(z) =

n=0
u
n
z
n
is always O
5
(R) (this
Ramanujan-type formulae for 1/: A second wind? 8
is an immediate consequence of a general result of F. Beukers and G. Heck-
man [10]), which is essentially richer than O
3
(R) for classical Ramanujans
series.
There exists also the higher-dimensional identity

n=0
(
1
2
)
7
n
n!
7
(168n
3
+ 76n
2
+ 14n + 1)
1
2
6n
=
32

3
,
discovered by B. Gourevich in 2002 (using an integer relations algorithm).
Guillera also found experimentally an analogue of Satos series:

n=0
v
n
(36n
2
+ 12n + 1)
1
2
10n
=
32

2
,
where v
n
=
_
2n
n
_
2 n

k=0
_
2k
k
_
2
_
2n 2k
n k
_
2
.
(5.8)
6 Transformations of hypergeometric series
As we have seen, Ramanujans original formulae as well as Guilleras formulae
(5.1)(5.7) involve classical hypergeometric series (1.8), while series like (3.1)
(3.4) and (5.8) are based on double hypergeometric series. A natural way
to pass from one formula to another is by algebraic transformations of the
hypergeometric series involved. For instance, formula (3.1) may be deduced
from the transformation

n=0
u
n
z
n
=
1
2 + 2z

1 34z + z
2

3
F
2
_
1
4
,
1
2
,
3
4
1, 1

256t(z)
_
where
t(z) =
z
2(1 + 14z + z
2
)
4
_
1 36z + 199z
2
+ 184z
3
+ 199z
4
36z
5
+ z
6
+ (1 + z)(1 z)
2
(1 18z + z
2
)

1 34z + z
2
_
,
given by Y. Yang (2005), together with the Ramanujan-type formula for the
3
F
2
-series on the right-hand side specialized at the point
_
5 + 4

2
7

3
_
4
= 256t
__

5 1
2
_
12
_
.
A similar argument is used by M. D. Rogers in [23] to deduce some further
identities for 1/ of RamanujanSato type.
Ramanujan-type formulae for 1/: A second wind? 9
Using the quadratic transformation z 4z/(1 z)
2
of the hypergeo-
metric series, we were able to produce from (5.1), (5.2) two more series of the
latter type [27]:

n=0
w
n
(4n)!
n!
2
(2n)!
(18n
2
10n 3)
1
(2
8
5
2
)
n
=
10

2
,

n=0
w
n
(4n)!
n!
2
(2n)!
(1046529n
2
+ 227104n + 16032)
1
(5
4
41
2
)
n
=
5
4
41

41

2
,
where the sequence of integers
w
n
=
n

k=0
_
2k
k
_
3
_
2n 2k
n k
_
2
4(nk)
, n = 0, 1, 2, . . . ,
satises the recurrence relation
(n+1)
3
w
n+1
8(2n+1)(8n
2
+8n+5)w
n
+4096n
3
w
n1
= 0, n = 1, 2, . . . .
In [28] we show that a huge family of formulae for 1/
2
(as well as for 1/
3
,
1/
4
, etc) can be derived by taking powers of Ramanujan-type formulae for
1/. For instance, the square of the Chudnovskys formula (1.7) takes the
monstrous form

n=0
w
n
(3n)!
n!
3
(222883324273153467n
2
+ 16670750677895547n
+ 415634396862086)
(1)
n
640320
3n+3
=
1
64
2
.
7 Further observations and open problems
It is worth mentioning that identities like (4.1) are valid for all non-negative
real values of k. This fact has several other curious implications; for instance,
the series
G(k) =

n=0
(1/2 + k)
5
n
(1 + k)
5
n
_
820(n + k)
2
+ 180(n + k) + 13
_
(1)
n
2
10n
(7.1)
has a closed-form evaluation at k = 0 and k = 1/2:
G(0) =
128

2
and G(
1
2
) = 256(3),
where the rst formula follows from (5.2) while the second one was given by
T. Amdeberhan and D. Zeilberger [4]. Guillera has conjectured (and proven)
Ramanujan-type formulae for 1/: A second wind? 10
evaluations for series like (7.1) viewed as functions of the continuous (complex
or real) parameter k.
It seems to be a challenge to develop a modular-like theory for proving
Guilleras identities and nding a (more or less) general pattern of them. For
the moment, we have only speculations in this respect on a relationship to
mirror symmetry, namely, to the linear dierential equations for the periods
of certain CalabiYau threefolds. A standard example here is the hypergeo-
metric series (cf. (5.1) and (5.2))
F(z) =
5
F
4
_
1
2
,
1
2
,
1
2
,
1
2
,
1
2
1, 1, 1, 1

2
10
z
_
=

n=0
_
2n
n
_
5
z
n
,
which satises the 5th-order linear dierential equation
_

5
32z(2 + 1)
5
_
Y = 0, where = z
d
dz
.
If G(z) is another solution of the latter equation of the form F(z) log z+F
1
(z)
with F
1
(z) zQ[[z]], then

F(z) = (1 2
10
z)
1/2
det
_
F G
F G
_
1/2
(the sharp normalization factor (1 2
10
z)
1/2
is due to Y. Yang) satises the
4th-order equation
_

4
16z(128
4
+ 256
3
+ 304
2
+ 176 + 39) + 2
20
z
2
( + 1)
4
_
Y = 0 (7.2)
(entry #204 in [3, Table A]). For a quadratic transformation of the new
function

F(z) we have the following explicit formula [2]:
1 + z
(1 z)
2

F
_
z
(1 z)
2
_
=

n=0
_
n

k=0
4
nk
_
2k
k
_
2
_
2n 2k
n k
__
2
z
n
, (7.3)
where the right-hand side is the Hadamard square of the series
1
1 16z
2
F
1
_
1
2
,
1
2
1

16z
1 16z
_
=

n=0
_
2n
n
_
2
(1)
n
z
n
(1 16z)
n+1
=

n=0
_
n

k=0
4
nk
_
2k
k
_
2
_
2n 2k
n k
__
z
n
which admits a modular uniformization (cf. Section 3). It is worth mention-
ing that (7.2) and the dierential equation of order 4 for the right-hand side
Ramanujan-type formulae for 1/: A second wind? 11
of (7.3) are of CalabiYau type [1], [14], i.e., they imitate all properties of
a dierential equation for the periods of a CalabiYau threefold. Are there
analogues of Hilbert class elds for this and similar situations? Can formulae
(5.1)(5.8) be deduced from formulae for 1/ by means of algebraic trans-
formations of hypergeometric series? There is still some work to do in the
subject originated by Ramanujans note [22] almost 100 years ago.
Acknowledgements. It is a pleasure for me to thank Gert Almkvist,
Heng Huat Chan, Jes us Guillera, Jonathan Sondow, Yifan Yang, Don Zagier,
and the anonymous referee for helpful comments.
References
[1] G. Almkvist and W. Zudilin, Dierential equations, mirror maps
and zeta values, in: Mirror Symmetry V, N. Yui, S.-T. Yau, and
J. D. Lewis (eds.), AMS/IP Stud. Adv. Math. 38 (International Press
& Amer. Math. Soc., Providence, RI, 2007) 481515.
[2] G. Almkvist, D. van Straten, and W. Zudilin, Generalizations of
Clausens formula and algebraic transformations of CalabiYau dieren-
tial equations, in preparation (2007).
[3] G. Almkvist, C. van Enckevort, D. van Straten, and
W. Zudilin, Tables of CalabiYau equations, arXiv: math/0507430
[math.AG] (2005).
[4] T. Amdeberhan and D. Zeilberger, Hypergeometric series accel-
eration via the WZ method, The Wilf Festschrift (Philadelphia, PA,
1996), Electron. J. Combin. 4:2 (1997) Research Paper 3, 4 pp.
[5] R. Apery, Irrationalite de (2) et (3), Asterisque 61 (1979) 1113.
[6] W. N. Bailey, Generalized hypergeometric series, Cambridge Math.
Tracts 32 (Cambridge Univ. Press, Cambridge, 1935); 2nd reprinted
edition (Stechert-Hafner, New YorkLondon, 1964).
[7] B. C. Berndt, Ramanujans notebooks. Part IV (Springer-Verlag, New
York, 1994).
[8] B. C. Berndt and H. H. Chan, Eisenstein series and approximations
to , Illinois J. Math. 45:1 (2001) 7590.
[9] F. Beukers, Irrationality proofs using modular forms, Journees
arithmetiques (Besan con, 1985), Asterisque 147-148 (1987) 271283.
Ramanujan-type formulae for 1/: A second wind? 12
[10] F. Beukers and G. Heckman, Monodromy for the hypergeometric
function
n
F
n1
, Invent. Math. 95:2 (1989) 325354.
[11] J. M. Borwein and P. B. Borwein, Pi and the AGM (Wiley, New
York, 1987).
[12] H. H. Chan, S. H. Chan, and Z. Liu, Dombs numbers and
RamanujanSato type series for 1/, Adv. Math. 186:2 (2004) 396410.
[13] H. H. Chan, W.-C. Liaw, and V. Tan, Ramanujans class invariant

n
and a new class of series for 1/, J. London Math. Soc. (2) 64:1
(2001) 93106.
[14] Y.-H. Chen, Y. Yang, and N. Yui, Monodromy of PicardFuchs
dierential equations for CalabiYau threefolds, J. Reine Angew. Math.
(to appear); arXiv: math/0605675 [math.AG] (2006).
[15] D. V. Chudnovsky and G. V. Chudnovsky, Approximations and
complex multiplication according to Ramanujan, in: Ramanujan re-
visited (Urbana-Champaign, Ill., 1987) (Academic Press, Boston, MA,
1988) 375472.
[16] S. B. Ekhad and D. Zeilberger, A WZ proof of Ramanujans for-
mula for , in: Geometry, Analysis, and Mechanics, J. M. Rassias (ed.)
(World Scientic, Singapore, 1994) 107108.
[17] S. Fischler, Irrationalite de valeurs de zeta (dapr`es Apery, Rivoal,
. . . ), Asterisque 294 (2004) 2762.
[18] J. Guillera, Some binomial series obtained by the WZ-method, Adv.
in Appl. Math. 29:4 (2002) 599603.
[19] J. Guillera, About a new kind of Ramanujan-type series, Experiment.
Math. 12:4 (2003) 507510.
[20] J. Guillera, Generators of some Ramanujan formulas, Ramanujan J.
11:1 (2006) 4148.
[21] A. van der Poorten, A proof that Euler missed... Aperys proof of
the irrationality of (3), Math. Intelligencer 1:4 (1978/79) 195203.
[22] S. Ramanujan, Modular equations and approximations to , Quart. J.
Math. Oxford Ser. (2) 45 (1914) 350372; reprinted in: Collected papers
of Srinivasa Ramanujan, G. H. Hardy, P. V. Sechu Aiyar, and B. M. Wil-
son (eds.) (Cambridge University Press, Cambridge, 1927; Chelsea Publ.,
New York, 1962) 2339.
Ramanujan-type formulae for 1/: A second wind? 13
[23] M. D. Rogers, New
5
F
4
hypergeometric transformations, three-
variable Mahler measures, and formulas for 1/, Ramanujan J. (to ap-
pear); arXiv: 0704.2438 [math.NT] (2007).
[24] A. Weil, Elliptic functions according to Eisenstein and Kronecker,
Ergebnisse der Mathematik und ihrer Grenzgebiete 88 (Springer-Verlag,
Berlin, 1976).
[25] Y. Yang, On dierential equations satised by modular forms, Math.
Z. 246:1-2 (2004) 119.
[26] W. Zudilin, Ramanujan-type formulae and irrationality measures of
certain multiples of , Russian Acad. Sci. Sb. Math. 196:7 (2005) 983
998.
[27] W. Zudilin, Quadratic transformations and Guilleras formulas for
1/
2
, Math. Notes 81:3 (2007) 297301.
[28] W. Zudilin, More Ramanujan-type formulae for 1/
2
, Russian Math.
Surveys 62:3 (2007) 634636.
Aliation: Steklov Mathematical Institute, Russian Academy of Sciences,
Gubkina str. 8, 119991 Moscow, RUSSIA;
Department of Mechanics and Mathematics, Moscow Lomonosov State Uni-
versity, Vorobiovy Gory, GSP-1, 119991 Moscow, RUSSIA
E-mail : [email protected]