Evaluations of a continued fraction of Ramanujan
N.D. BAGIS(*) - M.L. GLASSER(**)
ABSTRACT- We study the properties of a general continued fraction of Ramanujan.
In certain cases we evaluate it completely.
MATHEMATICSSUBJECTCLASSIFICATION(2010). 11A55; 33E05.
KEYWORDS. Continued Fractions; Ramanujan; Evaluations
1. Introduction Let
a;q
k:Yk 1
n0
(1 aqn) 1
Then the Ramanujan eta function is
f( q):(q;q)1 2
and the Weber function is
F( q):( q;q)1: 3
Also we denote by
K(x) Zp=2
0
1 1 x2sin2(t)
p dt
4
the elliptic integral of the first kind.
(*) Indirizzo dell'A.: Department of Informatics, Aristotele University, Pane- pistimioupolis, Thessaloniki, Greece.
E-mail: [email protected]
(**) Indirizzo dell'A.: Department of Physics, Clarkson University Potsdam, NY 13699-5820, New York, United States.
E-mail: [email protected]
The functionkris defined as the solution to the equation (see [2], [7]) K(k0r)
K(kr)
pr 5
whereris positive,qe ppr
andk0
1 k2
p . It is known that wheneverr is positive rational,kris an algebraic number.
In Berndt's book [5] pg. 21 and in [6], one can find the following ex- pansion
THEOREM 1.1. Suppose that q;a and b are complex numbers with q
j j<1, or that q;a, and b are complex numbers with abqm for some integer m. Then
UU(a;b;q)( a;q)1(b;q)1 (a;q)1( b;q)1 ( a;q)1(b;q)1(a;q)1( b;q)1 6
a b 1 q
(a bq)(aq b) 1 q3
q(a bq2)(aq2 b) 1 q5
q2(a bq3)(aq3 b)
1 q7 . . .
Our main concern in the present article is to simplify and evaluateUfor some specific cases ofa,bandq. For this purpose we define
X( a;q)1(b;q)1 (a;q)1( b;q)1; 7
then
X 1 X1U 8
2. Propositions
PROPOSITION2.1. Set
f(q) X1
n 1
qn2 9
then
f(q) 1 f(q)1 q
1q
q3 1q3
q5 1q5
q7 1q7. . . 10
PROOF. Take q!q2 in (6) and then seta!q andb!q2. Using [5]
chapter 16 Entry 22, we get the result (10).
The expansion (10) can also be found independently and directly by using Euler's general continued fraction expansion [8]. p
PROPOSITION2.2.
F( q) f( q) F( q)f( q) q
1 q
q3 1 q3
q5 1 q5
q7 1 q7. . . 11
PROOF. Set b0 in (6) and thenaq. The result follows imediately from the definitions off andFby relations (2) and (3). p
PROPOSITION2.3.
f( q) F( q)
f( q)F( q)f( q) 1 f( q)1 12
PROOF. This follows from Propositions 2.1 and 2.2 p We continue by setting
u0(a;q): 2a 1 q
a2(1q)2 1 q3
a2q(1q2)2 1 q5
a2q2(1q3)2
1 q7 . . .
13
and
P ( a;q)1 (a;q)1
2
: 14
Then
P 1
P1u0(a;q);
15
On solving (15) with respect toP, we get COROLLARY2.4. Ifjqj<1, then
16 ( a;q)1 (a;q)1
2
1 2 1
2a 1 q
a2(1q)2 1 q3
a2q(1q2)2 1 q5
a2q2(1q3)2
1 q7 . . .
COROLLARY2.5. Ifjqj<1andja=qj<1then 4X1
n0
a2n1
(2n1)(1 q2n1)log 1 2 1 u0(a;q)
17
PROOF. Take the logarithm on both sides of (16) and expand the two products in Taylor series. Then the double sums are easily rearranged to
get the desired result. p
From Corollaries 2.4 and 2.5, we proceed to the more general formula:
PROPOSITION2.6.
2X1
n0
a2n1 b2n1
(2n1)(1 q2n1)log 1 2 1 U(a;b;q)
18
and hence
1 2
1 U(a;b;q)
2
1 2
1 u0(a;q)
1 2
1 u0(b;q)
19
From relation (19) it is clear that to studyU we have only to examine u0(a;q).
One can find many useful results on internet sites, such as http://pi.physik.uni-bonn.de/ dieckman/InfProd/InfProd.html In some cases the fractionu0(a;q) can calculated in terms of elliptic func- tions. For example
1 2
1 u0(q;q) p 2k0rK(kr): 20
More generally, 4X1
n0
qn(2n1)
(2n1)(1 q2n1) 4Xn 1
j1
arctanh(qj) log 2k0rK(kr) p
21
from which we obtain:
PROPOSITION2.7. Letnbe positive integer, then 22 1 2
1 u0(qn;q) 1 2
1 2qn 1 q
q2n(1q)2 1 q3
q2n1(1q2)2 1 q5
q2n2(1q3)2
1 q7 . . .
p
2k0rK(kr)exp 4Xn 1
j1
arctanh(qj)
" #
LEMMA2.8. Letn1,n2be positive integers, then 23 1 2
1 U(qn1;qn2;q)exp 4 nX1 1
j11
arctanh(qj1) Xn2 1
j21
arctanh(qj2)
!
" #
PROOF. The proof follows easily from Propositions 2.6 and 2.7. p PROPOSITION2.9.
X1
n0
qn
1 a2q2n 1 1 q
a2(1 q)2 1 q3
qa2(1 q2)2 1 q5
q2a2(1 q3)2
1 q7 . . .
24
PROOF. Solve relation (18) forU(a;b;q), divide bya band then take
the limitb!a. p
PROPOSITION2.10. If qe ppr , then K(kr)
2p 1
4 1
1 q
(1 q)2 1 q3
q(1 q2)2 1 q5
q2(1 q3)2
1 q7 . . .
25
PROOF. Set in (24)ai,qe p
pr
and use the identity X1
n0
qn
1q2n K(kr) 2p 1
4 p
It is also known that whennis an integer,
1 2
1 u0(qn1=2;q)exp 4X1
n0
q(2n1)(n1=2)
(2n1)(1 q2n1)
" #
26
exp 4Xn 1
j0
arctanh(qj1=2)arctanh(kr)
" #
Hence
krtanh 4Xn 1
j0
arctanh(qj1=2)log 1 2 1 u0(qn1=2;q)
" #
For every positive integer n. By this means we get a continued fraction expansion for the singular moduluskr
k0r
1 kr 1 2
1 u0(q1=2;q) 27
From Propositions 2.6 and 2.7 and Lemma 2.8, we have
PROPOSITION2.11. If c is positive real andn1,n2are positive integers then:
1 2
1 U(qn1c;qn2c;q) 28
exp 2 Xn1 1
j11
arctanh(qj1c) Xn2 1
j21
arctanh(qj2c)
!
" #
: Moreover:
If UU(a;b;q), where qe ppr and c log(a)
p
pr
log(b) p
pr
29
then
1 2
1 U(a;b;q)
exp 2 X log(a)
ppr
1 j11
arctanh(qj1c) X log(b)
ppr
1 j21
arctanh(qj2c) 0
B@
1 CA 2
64
3 75
where xf g is the fractional part of x and x is the largest integer not exceeding x.
REMARK2.12. Observe that forn1n2n
1 2
1 U(qnc;qnc;q)1
Also, we observe that
1 2
1 U(a; b;q)
1 2 1 u0(a;q)
1 2
1 u0(b;q)
30
This relation is similar to (19).
PROPOSITION2.13.
31log 1 2 1 U(aq;b;q)
log 1 2
1 U(a;b;q)
2arctanh(a)
PROOF. From (20)
log 1 2
1 U(aq;b;q)
2X1
n0
b2n1
(2n1)(1 q2n1)2X1
n0
q2n1a2n1 (2n1)(1 q2n1) or
log 1 2
1 U(aq;b;q)
2X1
n0
b2n1
(2n1)(1 q2n1) 2X1
n0
a2n1
(2n1)(1 q2n1)
2X1
n0
q2n1a2n1
(2n1)(1 q2n1) 2X1
n0
a2n1
(2n1)(1 q2n1) 2X1
n0
a2n1 2n1
p PROPOSITION2.14.
32 log 1 2 1 U(a;b;q)
2X1
j11
arctanh(aqj1) 2X1
j21
arctanh(bqj2)
PROOF. In (31) replaceawithaqand add the result to (31) to arrive at
log 1 2
1 U(0;b;q)
log 1 2
1 U(a;b;q)
2X1
n0
arctanh(aqn)
By doing the same with
log 1 2
1 U(0;b;q)
2X1
n0
b2n1
(2n1)(1 q2n1); using the fact that
log 1 2
1 U(0;0;q)
0
we obtain (32). p
REMARK2.15. From (32) one can show that whena,b,sare positive reals and
b a
s nn1 n22N 33
then for
xa andya(bn2) bn1 a2 a b 34
andn1[y],n2[x], we can evaluate
log 1 2
1 U(qa;qb;qs)
log 1 2
1 U(qs(xy [y]);qs(xy [x]);qs)
in the closed form
2[x] 1X
j1
arctanhqs(jf gx f g)y
2[y] 1X
j1
arctanhqs(jf gx f g)y 35
From the relations (see [2])
sn(u)sn(q;u) 2p Kk
X1
n0
qn1=2sin 2K
p (2n1)u
1 q2n1 36
cn(u)cn(q;u) 2p Kk
X1
n0
qn1=2cos 2K
p (2n1)u
1q2n 1 37
whereu2Kz=pand Z
sn(u)dukr1
2 log 1 krcd(u) 1krcd(u)
1
2klog w3 w2
pk w3w2
pk
! 38
and from X1
n0
qn1=2
(2n1)(1 q2n1) i 4 arctan
m( q)
1 m( q)
s !
39
withm(q)k2rand (17) we get next
PROPOSITION2.16. If qe ppr
, r>0andu2R, then
Re log 1 2
1 u0(q1=2eiu;q)
40
1
2 log w3 w2
kr
p w3w2
kr p
u
0 iarctan
m( q)
1 m( q)
s !
wherew2;3are the Jacobi theta functions:
w2(q;z)2X1
n0
q(n1=2)2cos((2n1)z) 41
w3(q;z)12X1
n1
qn2cos(2nz) 42
PROPOSITION2.17. Ifu1;u2are real numbers andjqj<1, then 43 Re log 1 2
1 U(q1=2eiu1;q1=2eiu2;q)
1
4 log w3 w2
kr
p w3w2
kr
p
u2
u1
More generally,
cc(u)cc(q;u): 2p Kk
X1
n0
qn1=2cos (2n1)2Ku p
1q2n1
44
tan 4Ku p
ss(u)q 1sec 4Ku p
cn(u)
2ppqcos 2Ku p
sec 4Ku p
Kk(1q) where
ss(u)ss(q;u): 2p Kk
X1
n0
qn1=2sin (2n1)2Ku p
1q2n1 45
But cc(q;u) isign(q)cd( q;u), so by solving (44) for the function ss we get
46ss(q;u)
cn(q;u)csc 4Ku p
q
ppqcsc 2Ku p
Kkr(1q) icd( q;u) cot 4Ku p
which leads to
PROPOSITION2.18. Ifuis inCsuch thatjq1=2e2iuK=pj<1and0<q<1, then
p2 4K2k
d
dt log 1 2
1 u0(iq1=2e2itK=p;q)
tu 47
ippqcsc 2Ku p
kK(1 q) 1icot 4Ku p
cd(q;u)
csc 4Ku p
cn( q;u) q
PROOF. From (17), the definitions of cc, ss andeix cos(x)i sin(x), we have
p2 4K2k
d
du log 1 2
1 u0(iq1=2e2iuK=p;q)
ss( q;u) icc( q;u)
ss( q;u)cd(q;u)
Using (46), we get (47). p
REMARK2.19. The function ss(q;u) plays very important role here. If one finds its generalized integral, then the problem of the evaluation ofU will be solved.
REFERENCES
[1] M. ABRAMOWITZ and I.A. STEGUN, Handbook of Mathematical Functions, Dover Publications, 1972.
[2] J.V. ARMITAGE and W. F. EBERLEIN, Elliptic Functions, Cambridge Uni- versity Press, 2006.
[3] B.C. BERNDT,Ramanujan`s Notebooks Part I, Springer Verlag, New York, 1985.
[4] B.C. BERNDT,Ramanujan`s Notebooks Part II, Springer Verlag, New York, 1989.
[5] B.C. BERNDT,Ramanujan`s Notebooks Part III, Springer Verlag, New York, 1991.
[6] L. LORENTZENand H. WAADELAND,Continued Fractions with Applications, Elsevier Science Publishers B. V., North Holland, 1992.
[7] E.T. WHITTAKER and G. N. WATSON,A course on Modern Analysis, Cam- bridge U. P., 1927.
[8] H.S. WALL, Analytic Theory of Continued Fractions, Chelsea Publishers, New York, 1948.
Manoscritto pervenuto in redazione il 15 gennaio 2013.