16.

Jacobian Elliptic Functions and Theta

Functions
L. M . M~LNE-TEOMSON
Contents
Pege
Mathematical Properties . . . . . . . . . . . . . . . . . . . . 569
16.1. Introduction . . . . . . . . . . . . . . . . . . . . . 569
16.2. Classification of the Twelve Jacobian Elliptic Functions . . 570
16.3. Relation of the Jacobian Functions to the Copolar Trio . . 570
16.4. Calculation of the Jacobian Functions by Use of the Arith-
metic-Geometric Mean (A.G.M.) . . . . . . . . . . . 571
16.5. Special Arguments . . . . . . . . . . . . . . . . . . 571
16.6. Jacobian Functions when m=O or 1 . . . . . . . . . . . 571
16.7. Principal Terms . . . . . . . . . . . . . . . . . . . 572
16.8. Change of Argument . . . . . . . . . . . . . . . . . 572
16.9. Relations Between the Squares of the Functions . . . . . 573
16.10. Change of Parameter.. . . . . . . . . . . . . . . . . 573
16.11. Reciprocal Parameter (JacobisReal Transformation) . . . 573
16.12. Descending Landen Transformation (GaussTransforms-
tion) . . . . . . . . . . . . . . . . . . . . . . . 573
16.13. Approximation in Terms of Circular Functions . . . . . . 573
16.14. Ascending Landen Transformation . . . . . . . . . . . 573
16.15. Approximation in Terms of Hyperbolic Functions . . . . 574
16.16. Derivatives . . . . . . . . . . . . . . . . . . . . . 574
16.17. Addition Theorems . . . . . . . . . . . . . . . . . . 574
16.18. Double Arguments . . . . . . . . . . . . . . . . . . 574
16.19. Half Arguments . . . . . . . . . . . . . . . . . . . 574
16.20. Jacobis1.maginary Transformation. . . . . . . . . . . . 574
16.21. Complex Arguments . . . . . . . . . . . . . . . . . 575
16.22. Leading Terms of the Series in Ascending Powers of u. . . 575
16.23. Series Expansion in Terms of the Nome p . . . . . . . . 575
16.24. Integrals of the Twelve Jacobian Elliptic Functions . . . . 575
16.25. Notation for the Integrals of the Squares of the Twelve
Jacobian Elliptic Functions . . . . . . . . . . . . . 576
16.26. Integrals in Terms of the Elliptic Integral of the Second
Kind . . . . . . . . . . . . . . . . . . . . . . . 576
16.27. Theta Functions; Expansions in Terms of the Nome p . . 576
16.28. Relations Between the Squares of the Theta Functions . . 576
16.29. Logarithmic Derivatives of the Theta Functions . . . . . 576
16.30. Logarithms of Theta Functions of Sum and D8erence . . 577
16.31. JacobisNotation for Theta Functions . . . . . . . . . 577
16.32. Calculation of JacobisTheta Function @(ulm) by Use of
the Arithmetic-Geometric Mean . . . . . . . . . . . 577

1 University of Arizona. (Prepared under contract with the National Bureau of

Standards.)
587
568 JACOBIAN ELLIPTIC FUNCTIONB AND THETA FUNcllONS

16.33. Addition of Quarter-Periods to Jacobi's Eta and Theta

Functions . . . . . . . . . . . . . . . . . . . . . 57 7
16.34. Relation of iTacobi's Zeta Function to the Theta Functions. 578
16.35. Calculation of Jacobi's Zeta Function Z(+) by Use of the
Arithmetic-Geometric Mean . . . . . . . . . . . . . 578
16.36. Neville's Notation for Theta Functions . . . . . . . . . 578
16.37. Expression as Infinite Products. . . . . . . . . . . . . . 579
16.38. Expression as Infinite Seriea . . . . . . . . . . . . . . 579
Numerical Methods . . . . . . . . . . . . . . . . . . . . . . 579
16.39. Use and Extension of the Tables . . . . . . . . . . . . 579
References . . . . . . . . . . . . . . . . . . . . . . . . . . 581
Table 16.1. Theta Functions . . . . . . . . . . . . . . . . . . . 582
s9,(~O\ao)J s a se(~'I\a0)
sm(E"\a"), Jea tpXEya0)
a=0(50)850,E, ~,=00(5~)900,9-10D
Table 16.2. Logarithmic Derivatives of Tbeta Functions . . . . . . 584
d
-In 9,(u) =j(co\ao)
du
d SC(u)=-j(~aao)
-In
du
d
-In 8,(u)=g(co\ao)
du
d
- In sd(u)= -g(c1ao)
du
a=0"(5")85", c, ~~=00(5~)900,
5-6D

The author wishes to acknowledge hie great indebtedneas to hie friend, the late P r p f m r
E. H. Neville, for invaluable Bssietance in readiig and criticizing the manuscript. Pro-
fessor Neville generously supplied material from hie own work and was responsible for many
improvements in matter and arrangement.
The author'a beat thanb are also due to David S. Liepman and Ruth Zucker for the
preparation and checking of the tablea and graphs.
 16. Jacobian Elliptic Functions and Theta Functions
Mathematical Properties

Jacobian Elliptic Functions Let p, q be any two of the letters s, c, d, n.

Then p, q determine in the lattice a minimum
16.1. Introduction rectangle whose sides are of length K and Kand,
A doubly periodic meromorphic function is whose vertices s, c, d, n are in counterclockwise
called an elliptic function. order.
Definition
Let m, ml be numbers such that
The Jacobian elliptic function pq u is defined by
m+ml=l. the following three properties.
(i) pq u has a simple zero at p and a simple
We call m the parameter, ml the complemkntary pole at q.
parameter. (ii) The step from p to q is a half-period of pq u.
In what follows we shall assume that the param- Those of the numbers K, iK,K+iKwhich differ
eter m is a real number. Without loss of gen- from this step are only quarter-periods.
erality we can then suppose that Om < l< (see (iii) The coefficient of the leading term in the
16.10, 16.11). expansion of pq u in ascending powers of u about
We define qwcrter-periods K and iKby u=O is unity. With regard to (iii) the leading
term is u, l/u, 1 according as u=O is a zero, a
16.1.1
pole, or an ordinary point.
Thus the functions with a pole or zero at the
origin (Le., the functions in which one letter is s)
are odd, and the others are even.
Should we wish to call explicit attention to the
value of the parameter, we write pq (ulm) instead
so that K and Kare real numbers. K is called of pq u.
the real, iKthe imaginary quarter-period. The Jacobian elliptic functions can also be
We note that defined with respect to certain integrals. Thus if
16.1.2 K(m)=K(m,)=
K(1-m). de
16.1.3
(1 - m sin2e) 1/2
We also note that if any one of the numbers m,
ml, K(m), K(m),K(m)/K(m) is given, all the the angle cp is called the amplitude
rest are determined. Thus K and Kcan not
both be chosen arbitrarily. 16.1.4 cp=am u
In the Argand diagram denote the points 0, K, and we define
K+iK,iKby s, c, d, n respectively. These
points are at the vertices of a rectangle. The 16.1.5
translations of this rectangle by AK, piK,where sn u=sin Q, cn u=cos cp,
A, p are given all integral values positive or nega-
tive, will lead to the lattice dn u=(l-m sin2 cp)1/2=A(cp).

.S .C .S .c Similarly all the functions pq u can be expressed

.n .d .n .d in terms of cp. This second set of definitions,
.S .C .S .c although seemingly different, is mathematically
.n .d .n .d equivalent to the definition previously given in
terms of a lattice. For further explanation of
the pattern being repeated indefinitely on all notations, including the interpretation, of such
sides. expressions aa sn (q\a),cn (ulm), dn (u, k), see 17.2.
569
 .
Pole Pole Pole Pole
iK' K+iK' K 0

Half period iK' an u cdu dc u ns u Periods ZiK', 4K+4iKt, 4K

-
Halfperiod K+iK' cnu sdu nc u da u Periods 4K', 2K+2iK', 4K
- -
Half period K dn u ndu SCU ca u Periods 4iK', 4K+4iKr, 2K

16.2. Jacobian diptic junctions

FIQURE v u
ns u, nc u, nd u provided that when two lettars are the same, e.&,
1 pp u, the corresponding function ki put equd to
 JACOBIAN ELLIPTIC FUNCTIONS AND THETA FUNCTIONS 571
16.4. Calculation of the Jacobian Functions by and then compute successively pN+ *-a, . . .,
Use of the Arithmetic-Geometric Mean p,, +e,. from
the recurrence relation
(A.G.M.) Cn
16.4.3 sin (2p,,-1-p,,)=- sin pa.
For the A.G.M. scale see 17.6. a,
To calculate sn (ulm), cn (ulm), and dn (ulm) Then
form th0A.G.M. scale starting with 16.4.4
16.4.1 @)=l, bo=&, co=J;sI, si1 (ulm)=sin cti (ulm)=cos cpo
terminating at the step N when cN is negligible to
the accuracy required. Find pN in degrees where
16.4.2 m=2NaNu-
1SO0
T
I h m Wese all the other functions can be deter-
mined.

- U
16.5. Special Arguments

sn u cn u dn u

16.5.1 0 0 1

1
16.5.2
aK
16.5.3 K

16.5.4 l(iK')

16.5.5 31 (K+ iK')

1
16.5.6 K+ Z(iK')

16.5.7 iK' m

16.5.8 aK+iK'
1

16.5.9 K+iK'

16.6. Jacobian Functions when m=O or 1

- m=l
m=O

16.6.1 sin u tanh u

16.6.2 cos u sech u
16.6.3 1 uech u
16.6.4 cps U 1
16.6.5 Bin u sinh u
16.6.6 1 cosh u
16.6.7 Bec u 1
16.6.8 Bec u cosh u
16.6.9 tan u sinh u
16.6.10 csc u coth u
16.6.11 cac u csch u
16.6.12 oot u csch u
16.6.13 U gd u
572 JACOBIAN ELLIPTIC FUNCTION6 AND THETA FUNCTION8
$
+
*
+
J
4
;N"
+
3
3
I
5
I i .I.
 JACOBIAN ELLIPTIC FUNCMONS ANI) THETA FUNCMONB 573
16.9. Relations Between the Squaree of the
Functions

16.9.1 -dn%+rnl= -m cn2u=m sn2u-m

16.9.2 -mlndu+ml= -mmlsd2u=m cdu-m
16.9.3 mlsc2u+ml=mlnc2u=dc2u-m
16.9.4 cs2u+m1 =dszu=ns2u-m
In using the above results remember that
m+ml=l.
If pq u, rt u are any two of the twelve functions, ie
one entry expresses tq2u in terms of pqu and conveniently to find an (ulm) in terms of an (ulp)
another expresses qt2u in terms of rtk. Since and dn (ulm) in terms of dn (VIP), but that the
tq%. qt2u=1, we can obtain from the table the calculation of cn (ulm) requires all three functions.
bilinear relation between pq2u and rtk. Thus
for the functions cd u, an u we have 16.13. Approximation in Terms of Circular
Functions
16.9.5 ndZu=- d2u, dnzu=l -m snu When the parameter m is so small that we may
m1
neglect mz and higher powers, we have the
and therefore approximations
16.9.6 (1-m cd2u)(1-m sn2u)=ml. 16.13.1
1
16.10. Change of Parameter sn (ulm) =sinu-- m(u-sin u COB u) cos u
4
Nqative Parameter
If m is a positive number, let 16.13.2
1
cn (ulm) =COB u+- m(u-sin u COB u) sin u
m 1 U 4
16.10.1 p=- 1+ mPI=- 1+ m
1
16.13.3 dn (ulm) = 1-- m sin2u
16.10.2 811 (ul-m)=&d (VIP) 2

16.10.3 cn (uJ-m)=cd (VIP) 1

16.13.4 am (ulm) =u-- m(u-sin u cos u).
4
16.10.4 dn (ul-m)=nd (VIP).
One way of calculating the Jacobian functions
16.11. Reciprocal Parameter (JacobisReal is to use Landensdescending transformation to
Transformation) reduce the parameter sufEciently for the above
16.11.1 m>O, p=rn-, v=um112 formulae to become applicable. See also 16.14.

16.11.2 sn (ulm)=p%n (VIP) 16.14. Ascending Landen Transformation

16.11.3 cn (ulm)=dn (4P)

To increase the parameter, let

16.11.4 dn (ulm)=cn (VIP)

Here if m>l then m-=p<l.
Thus elliptic functions whose parameter is real
can be made to depend on elliptic functions whose
parameter lies between 0 and 1.
16.12. Descending Landen Transformation
Transformation)
(Gauss
To decrease the parameter, let
574 JACOBIAN ELLIPTIC FUNCMONS AND THETA FUNCTIONS

Note that, when successive applications are to 16.17. Addition Theorems

{
be made, it is simplest to calculate d n (ulm) since
16.17.1 sn(u+v)
this is expressed always in terms of the same func-
tion. The calculation of cn (ulm) leads to that of -
-
sn uecn v-dn v+sn vmcn u.dn u
dn (VIP). 1 -m snZu.sn2v
The calculation of sn (ulm) necessitates the 16.17.2 cn(u+v)
evduation of all three functions. cn u . cn v-sn u . dn u . sn v -dn v
1-m sn2u'snb
16.15. Approximation in Terms of Hyperbolic
Functions dnu. dnv- msnu. cnu.snv . cnv
16.17.3 dn(u+v)=
1 - msn2u. sn2v
When the parameter m is so close to.unity that
m,2 and higher powers of ml can be neglected we Addition theorems are derivable one from
have the approximations another and are expressible in a great variety of
forms. Thusns(u+v) comes from l/sn(u+v) in the
16.15.1 form (1 -msn2usnzv)/ (sn u cn v dn v+ sn v cn u dn u)
1 from 16.17.1.
sn (ulm) = tanh u+- ml (sinh u cosh u-u) sech2 u
4 Alternatively ns(u + v)=ml%n (W-u)-v}
16.15.2 which againfrom16.17.1 yields the form (ns u cs v dsu
cn (ulm) =sech u -nsvcsudsu)/(ns2u-ns2v).
The function pq(u+v) is a rational function of
-a1
m, (sinh u cosh u-u) tanh u sechu the four functions pq u, pq v, pq'u, pq'v.
16.18. Double Arguments
16.15.3
dn (ulm) =sech u 16.18.1 sn 2u
1 -2 s n u . c n u . d n u --
29nu.cnu.dnu
+& m, (sinh u cosh u+u) tanh u sech u 1 - 773811% cnzu+sn2u.dn2u
16.15.4 16.18.2 cn2u
1
~ (sinh u cosh u-u) sech u.
am (ulm) =gd u + ml -cnzu-sn2u. dn2u-cn2u-sn2u.
- dn2u
1 -msn4u cn2u+sn2u- dn2u
Another way of calculating the Jacobian func- 16.18.3 dn 2u
tions is to use Landen's ascending transformation -dn~u-msnZu.cnzu~dn~u+cnzu(dn2u-l)
- -
to increase the parameter suficiently for the above 1 -msn'u dn2u-cn2u(dn2u- 1)
formulae to become applicable. See also 16.13. 1 -cn 2u-sn2u dnZu
16.18.4
l+cn 2u- cn2u
16.16. Derivatives

1
I 2;- Derivative 16.18.5
1 -dn 211- mn2u cn2u
l+dn 211- dn2u
16.19. Half Arguments
16.16.1 sn u en u d n u 1-cn u
16.16.2 cn u -sn u dn u Pole n 16.19.1 sn2&u=- l+dn u
16.16.3 dn u -m sn u cn u
16.16.4 cd u -ml sd u nd u dn u+cn u
16.16.5 sd u cd u nd u Pole d 16.19.2 cn2&u= l+dn u
16.16.6 nd u m sd u cd u
16.16.7
16.16.8
16.16.9 I 1
dc u
,",","
ml sc u nc u
scudcu
dcuncu
Pole c 16.19.3 dn2&U=
ml+dn u+mcn u
l+dn u
16.16.10
16.16.11
16.16.12 1 2; I
ns u -& u cs u
Pole B

Note that the derivative is proportional to the

16.20. Jacobi's Imaginary Transformation
16.20.1
16.20.2
sn(iu1m) =isc(ulml)
cn(iulm)=nc(u(ml)
product of the two mpolar functions. 16.2Q.3 dn(iu1m)=dc(u( m,)
 JACOBIAN ELLIPTIC FUNCTIONS AND "EXETA FWNCmONS 575
16.21. Complex Arguments 16.23.6
With the abbreviations
16.21.1 16.23.7
s=sn(zl m), c=cn(z\m), d=dn(z1 m), sl=sn(YJml), 7r
dc (uIm)-- S~CV
cl=cnblml), dl=dn(ylml) -2K
2A OD
cos (2n+l)o
n-0

16.23.9
16.22. Leading Terms of the Series in Aecending
Powers of u
16.22.1
u3 Ub +m:'2K
2r
(-1)" n"" sin 2nu
l+d"
sn(u1 m) =u- (1 +m) -+ (1 +14m+mz)3 16.23.10
3!
-(1+135m+135m2+ma)
u7
...
7r
ns (u.lm)=- csc v-- K n-0 I-@"+'
2K
2A
-sin (2n+l)v5 p+1

16.22.2
U2 U'
16.23.11
A
ds (uIm)=- csc v--
2U n""+1
K n-0 1+qz"+'
-
sin (2n+l)v
cn (uI m) = 1 --+ (1 +4m) - 2K
~

2! 4! 16.23.12
-(1+44m+16mz) $+ ...

16.22.3
16.24. Integrals of the Twelve Jacobian Elliptic
Functions

-m(16+44m+mz) &-. . .
U6 16.24.1 fan u du=m-I*l In (dn u-mlJ2cn u)
16.24.2 fch u du=m-lD arcc08 (dn u)
No formulae are known foq the general coeffi-
cienta in these.seria. 16.24.3 f dn u du=arcsin (sn u)

16.23. Series Expansions in Terms of the Nome 16.24.4 f cd u du=rn-lJ2 I n (nd u+m%d u)
p=e-rK'/K and the Argument v=.ru/(2K) 16.24.5 fad u du= (mrnl)-lJ2 amin (-ml%d u)

16.23.1
2r
sn (ulm)=-ml/ZK
m
3 pn+1/2
sin (2n+1)v 16.24.6 Jnd u du=ml-l/zarcms (cd u)

16.23.2
2r
cn (ul m) = -
m1j2K
- _pn+lR cos (2n+l)v
__

1+p2"+'
16.24.7
16.24.8
fdc u du=h (nc u+sc u)
f nc u du=m;1/2 (dc u+m:/2sCu)

-
n-0 I n

16.23.3
t
dn (ul m) =-+-
2t -
-cos2nv
2K K,-ll+q'"
'"
16.24.9
16.!24.10
f sc u dU=mi1l2 (dc u+m:/%c U)
f u du=h (ds U-cs
I n

U)
16.23.4 16.24.11 f ds u du=ln (ns u-cs u)
16.24.12 u dU=h
JCS (m U--dS U)

16.23.5 I nnumerical us3 of the above table certain re-

stiictions must be put on u in order to keep the
arguments of the logarithms paeitive and to avoid
576 JACOBIAN ELLIPTIC FUNCTIONB AND THETA F"CTI0NS

trouble with many-valued inverse circular func-

16.27.3 4(z, q)=8,(2)=1+2
-
C pR2cos 2nz
tions. 7I-1

16.27.4
16.25. Notation for the Integrals of the Squares m
of the Twelve Jacobian Elliptic Functions &(z, q)=8,(2)=1+2C (-l)"qn2 cos 2nz
nil

Theta functions are important because every

16.25.1 Pq u=
s,'pq2t dt when q #s one of the Jacobian elliptic functions can be ex-
pressed as the ratio of two theta functions.
16.25.2 Ps u=s,' (pq2t-h) dt-&1 See 16.36.
The notation shows these functions as depend-
Examples ing on the variable z and the nome p, lql<l.
In this case, here and elsewhere, the convergence
cd2t dt, Ns u=l(ns2t-$)dt-i 1 is not dependent on the trigonometrical terms.
In their relation to the Jacobian elliptic functions,
16.26. Integrals in Terms of the Elliptic Inte- we note that the nome p is given by
gral of the Second Kind (see 17.4)
P =e--rK 'IK,
16.26.1 mSn u= -E(u) +u where K and iK' are the quarter periods. Since
16.26.2 mCn u=E(u) -mlu Pole n q=q(m) is determined when the parameter m is
given, we can also regard the theta functions as
16.26.3 Dn u=E(u) dependent upon m and then we write
16.26.4 mCd u= -E(u) +u+mn u cd u ~ A z q>=8.(zlm),
, a=1, 2, 3J4
16.26.5 but when no ambiguity is to be feared, we write
mmlSd u=E(u)-mlu-mn u cd u Pole d 8Jz) simply.
The above notations are those given in Modern
16.26.6 mlNd u=E(u)-man u cd u
Analysis [16.6].
16.26.7 Dc u= -E(u) +u+sn u dc u There is a bewildering variety of notations, for
example the function 9,(z) above is sometimes
16.26.8 denoted by $,(z) or 8(z); see the table given in
mlNc u=-E(uL)+mlu+sn u dc u Pole c Modern Analysis [16.6]. Further the argument
16.26.9 mlSc u= -E(u) +sn u dc u u=2Kz/r is frequently used so that in consulting
books caution should be exercised.
16.26.10 Ns u= -E(u) +u-cn u ds u
16.28. Relations Between the Sguares of the
16.26.11 Theta Functions
Ds u=-E(u)+mlu-cn u ds u Pole s
16.28.1 8:(~)8:(0) =8:(~)9:(0) -ai( ~)8:(0)
16.26.12 Cs u= -E(u) -cn u ds u
16.28.2 8:( 2)81(0)= 8:( 2)8:(0) -8:(~)8:(0)
All the above may be expressed in terms of
Jacobi's zeta function (see 17.4.27). 16.28.3 8: (Z) 8: (0)=8: (Z) 8; (0)-8: (Z) 8;(0)
E 16.28.4 8i(~)
8:(0) =8:( 2)&(O) -8; (z)@(O)
Z(u)=E(u)--u,
K where E=E(K)
16.28.5 +W)
=8:(0)
16.27. Theta Functions; Expansions in Terms
of the Nome q Note also the important relation
16.27.1 16.28.6 8:(0)=8,(0)8,(0)8,(0) or 8i=828394
OD

&(z, p)=&(z) =2q1'4C(- l)"qn("+') sin (2n+l)z 16.29. Logarithmic Derivatives of ' the Theta
n-0
Functions
1637.2
(D

~Y,(z,p ) = s , ( ~ ) = 2 p l /pn(i+l)
~~ COS (2n+i)z
n-0
 JACOBIAN ELLIPTIC FUNCTIONS AND THETA FUNCTIONS 577
16.29.2 16.31. Jacobi's Notation for Theta Functions
16.31.1 @ ( ~ l m ) = @ ( ~ ) = 8 , ( ~ ) , v=- m
2K
16.31.2 @ , ( u ~ ~ ) = @ , ( u ) = ~ , ( v ) = O ( U + K )
16.31.3 H(U~m) = H(U) = 81(v)
) 16.31.4 uH,(ul m) =HI ( =&(v) = H(u+K)
16.32. Calculation of Jacobi's Theta Function
16.30. Logarithms of Theta Functions of Sum @(ut m) by Use of the Arithmetic-Geometric
and Difference
Mean
16.30.1
Form the A.G.M. scale starting with
16.32.1 bo=&,
aQ=l, c,= 6
terminating with the Nth step when cNis negligible
to the accuracy required. Find pN in degrees,
16.30.2 where
1800
16.32.2 (ON=yaNu -
77

and then compute successively (ON+ (ON-2, . . .,

(4, 1p0 from the recurrence relation

16.30.3 .
16.32.3 sin (2++,-(0,,)=- C" sin pn.
an
Then
16.30.4 16.32.4

The corresponding expressions when /3=irare

easily deduced by use of the formulae 4.3.55 and
4.3.56.
16.33. Addition of Quarter-Periods to Jacobi's Eta and Theta Functions

U u+2K L+ 2K + 2iK'

where H(u) and H,(u) have the period 4K. Q(u) and
&(u) have the period 2K.
2iK' is a quasi-period for all four functions,
that is to say, increase of the argument by 2iK'
multiplies the function by a factor.
578 JACOBIAN ELLIPTIC FUNCTIONS AND THETA FUNCTIONS

16.34. Relation of Jacobi's Zeta Function to the 16.36.2 8d(u)= e@+q 8,(u)=*).
e(K)
Theta Functions 0 (0)
b If A, p are any integers positive, negative, or
Z(u)=& In Q(u) zero the points u,,+2Mi+2piKt are said to be
congrzcent to 210.
a 8: (E)
2K c n u d n u 8,(u) has zeros at the points congruent to 0
16.34.1 Z(W)=-
2K91 (g)- sn u 8,(u) has zeros at the points congruent to K
8,(u) has zeros at the points congruent to iK'
tYd(u) has z e m at the points congruent to
K+iK'
2K d n u s n u
-_ a 8: (E),
16.34.2 -
cn u Thus the s u f f ~secures
~ that the function a,(@)
2KQ2 (g) has zeros at the points marked p in the intro-
d u c h y diagram in 16.1.2, and the constant by
which Jacobi's function is divided secures that the
16.34.3
-2K 8a (z)
_ ?r * ( S ) - m s n u c n u
-
dn u
leading coefficient of 8,(u) at the origin is unity.
Therefore the functions have the fundamentally
important property that ifp, q are any two of the
letters E, c, n, d, the Jacobian elliptic function

16.34.4
_ r 8:
- ($2 pq u is given by
Q,(4
-2K 84 (g) 16.36.3 PP u=8,(u>'

These functions also have the property

16.35. Calculation of Jacobi's Zeta Function
16.36.4 m;"'S,(K-u) =8,(~)
Z(ulm) by Use of the Arithmetic-Geometric
Mean 16.36.5 m;"'Sd(K-u) =8,(~),
Form the A.G.M. scale 17.6 starting with
for complementary arguments u and K-u.
16.35.1 %=I, bo= 6,
%= 6 In terms of the theta functions defined in 16.27,
let v=a-t~/(2K),then
terminating at the Nth step when CN is negligible
to the accuracy required. Find VN in degrees 16.36.6 ~Y,(u)=-J 2KfJl(d O,(U)=- 82 (4
where $1 (0) 8 2 (0)

16.35.2 w
CpN=PaNU -
180'
16.36.7 &(4
9 " ( U ) = 8 ~ ~an(U)=-.
Q4(4
94(0)

and then compute successively w - 1 , VN--~, . . .,

e,(PO from the recurrence relation
Cn
16.35.3 sin (!&--1-Vn)=- sin Vn.
a,
Then
16.35.4
Z(ulm)=cl sin R+G sin e+ . . . +cN sin C~N.
16.36. Neville's Notation for Theta Functions
These functions are aeiinea in wrms of ciacubi'a 16.4. NezriUe'a ihtafunctionS
FIQURE
theta functions of 16.31 by
aa(u)t 8c(u), ~ # ( U ) JQn(u)
16.36.1 H(u) 8,(~)=
&(u)=H'(o)' H(u+K) m=-1
H(K) 2
 {
16.38. Expression as Infinite Series
Let a=m/(2K)
16.38.1
2**1/2
8*(u)=[m1,2m;/2K]
1/2
3 (-1)
m
I P-("+I) sin (2n+ l)a

16.38.2 S,(U)=
1/2],,,[ - 5 q"(-+l) cos (27b+1)0
n=O

FIGURE16.5. hgarithm@ Oh+%t'ivf%

functions
O f fhta16.38.3 8a(u)=[&] 112 (1+2 5
R-1
*n2 co8 2nv)

d In 8,(u),
- -d In 8,(u) 16.38.4
au du

m =5
1 %(u)==+
[ K] ll2 1 +2 2 (-- 1)"Pa
cos %a)
/
16.37. Expression as Infinite Products . . . = ~ Y ~ q)
16.38.5 (2~/~)1/2=1+2~+2p~+2n9+ (o,
q= n(m)9 0 = 4 2 K )
16.37.1 16.38.6
8 , ( u ) =m(m1 6
a y 6 sin 2, n-1 (1-26" COB 2v+fp) (2K'/~)'lZ=1+2q1+2~+2~+.=83(0, all * *

16.37.2
16.38.7
8,(,)=(=y6Co8
m
a ii
a-1
(1+2qs"cos 2u+fp)
(2m1"~/~)1n=2a1/4(1+~+$+p1*+q~0+ . . .)
16.37.3
mm, l/lP m
8a(u)=(-) II
n-1 (1 +2n"-' co8 2u+p-') = 82(0, P)
16.37.4 16.38.8
1/12 m
&(u)=(&) II
n-1 (1-2p-1 COB 2a+fp-3
(2m:/2~/~)1"=1-2n+2n4-2qg+ . . . =8,(0, n).

16.39. Use and Extension of the Tables

Example 1. Calculate nc (1.996501.64) to 4s.
p2=( &y =10-"X 7.67

From Table 17.1,1.9965O=K+.OOl. From the which is negligible.

table of principal terms From 16.12.4
ncu=-m;ln/(u-K)+ . . . dn2 [*1g+(i31-(1-i3.
dn(.20(.19)=
nc (K+.0011.64)=-(.36)-'/'
.ool +...
- 10000 Now from 16.13.3
(1+h)-dn2[ .I9 1(6)l
---6
=-1667+. . .
+...
I
dn [.19 ($)]=.999951
whence dn (.201.19)=.996253.
and since the next term is of order .001 this value Example 3. Use the ascending Landen trans-
-1667 is correct to at least 4s. formation to calculate dn (.201.81) to 5D.
Example 2. Use the descending Landen trans- Fkom 16*14-1
formation to calculate dn (.201.19) to 6D. 4(.9) 360
p=--
Here m=.19, m:/*=.9and so from 16.12.1 (1.9)2-J 361 Cl=(AY

Ale0
-(3 20
M- - , 1 +p'/2=lS' 0=.19.
p
:
20
1 +p;/*=--
19'
.-gX .20 = .19,

ki negligible to 4D. Thus

580 JACOBIAN ELLIPTIC FUNCTIONS AND THETA FUNCTIONS

19
dn2( .19 I E)+% 1 Example 4. Use the ascending Landen trans-
formation to calculate cn (.20(.81)to 6D.
dn (.20/.81)=5X
dn (.19 1 g) Using 16.14.4, we calculate dn (.201.81) and
deduce cn (.20).81)from 16.14.3 settling the sign
from Figure 16.1.
From 16.15.3
As in the preceding example, we reduce the
1 1
=sech (.19)+-X-
4 361 t,anh.19 sech .19 calculation of dn (.201.81) to that of dn

1
3
[sinh .19 cosh.19+.19] when
1 1
= .982218+-X - (.187746)(.982218)
4 361
(
dn .19 - =.982267
[(.191145) (1.01810) + .191 dn (.201.81)=.984056
1 1
= .982218+-X- (.184408)[.384605] cn (.201.81)=.980278.
4 361
= .982218+.000049=.982267.
Thus dn (.201.8l)= .98406.

n an bn Cn -
Cn
an
9n sin qn sin ( 2 ~ n - 1 - 4 2~n-1-9n
-----
0 1 .8 .6 .6 . 65546 . 60952
1' . 9 . 89443 . 1 . 11111 1.2069 .93452 . 10383 . 10402
2 .89721 . 89721 .00279 .00311 2.4117 . 66679 00207 00207
3 .89721 . 89721 0 0 4. 8234 -. 99384 0' 0

n 9r sin q,, sin (~v.-I- 4 2*-1-* - (2*-1-*) 1

=lnsec (2+-1-*)

0 . 58803 . 55472
1 1.0780 . 88101 . 09789 .09805 1.0048 .00120
2 2. 1533 .83509 00260 .00260 1. 0
3 4.3066 -.91879 0' 0 1. 0
 JACOBIAN EUIPTlC FUNCTIONS AND THETA FUNCTION6 581
Example 7. Use the q-seriea to compute 8,(300\45O)= 39128
ca (33601 621.09). (sec 45)'8,(300\450)= 1.02796.
Here we use the series 16.23.12, K=1.60804 862,
r U r
Therefore
q=.00589 414, v=z=pdians or 30". sc (.61802I .5) =m
.59128 (am 450)t
Since g' is negligible to 8D, we have to 7D
ca (33601 621.09)
. . = .68402.
r
-- Example 10. Fmd sn (.753421.7) by inverse
-2K intarpolation in Table 17.5.
=(.97683 3852)(1.73205 081) This method is explained in chapter 17, Example
-3.90733 541[(.00003 4740)(.86602 5404)) 7.
Example 11. Find u, given that ca (uI.5)=.75.
=1.69180 83. From 16.9.4 we have
Example 8. Use theta functions to compute
sn (.618021.5) to 5D.
Here K(3) =1.85407 Thus
sna (ul.5)
= .64
and
sn (ul.5)=.8.
Thus We have therefore replaced the problem by
8, (3Oo\45O) that of finding u given sn (ulm), where m is known.
=
~.618021.5) 6,0 If p=am u
=- *59128 .56458 sin (p=sn u and so
1.04729-
.~

from Table 16.1. (p=.9272952radians or 53.13010.

Example 9. Use theta functions to compute From Table 17.5,
sc (.618021.5) to 5D.
As in the preceding example ~ = ~ ( 5 3 . 1 3 0 1 0 ~ \ 4 5.99391.
~)=
t0=300, a0=45' Alternatively, starting with the above value of
80 that (pwe can use the A.G.M. scale to calculate F(p\a)
as explained in 17.6. This method is to be pre-
ferred if more figurea are required, or if a diff'ars
We use Table 16.1 to give from a tabular value in Table 17.5.

Tat8 Tpbk
[lS.ll A. Erdblyi et al., Higher transcendental functions, E. p. Adam and R. HiPPbley, smithsonisn
vol. 2 (McGraw-Hill Book Co., Tnc., New York, mathematical formulae and tablea of elliptic
N.Y., 1953). functions, 3d reprint (The Smitheonian Institu-
t16.21 L. V. King, On the direct numerical calculation of tion, Washington, D.C., 1957).
elliptic functions and i n k p b (Cambridge [16.8]J. Hotid, h u e i l de formulea et de tablea ~ ~ m 6 -
Univ. Pr-, Cambridge, England, 1924). riquea (Gauthier-Vdlars, Paris, France, 1901).
[16.31 w. M~~ and F. 0brhettinger, ~~~~b and [16.91 E. Jahnke and F. Emde, Tabla of functio-, 4th
theorems for the special functions of mathe- ed. (Dover Publications, Inc., New York, N.Y.,
matical physics (Chelsea Publishing Co., New 1945).
York, N.Y., 1949). (16.101 L.M. Milne-Thomson, Die elliptischen Funktionen
(16.41 E. H. Neville, Jacobian elliptic functions, 2d ed. von Jacobi (Julius Springer, Berlin, Germany,
1931).
(Oxford Preae, En@and, lQ5l). (16.111 L. M. Milne-Thomeon, Jacobian elliptic function
[16.5] F. Tricomi, Elliptische Funktionen (Akademiache tables (Dover Publications, Inc., New York,
Verlagsgeaelhchaft, Leipdg, Germany, 1948). N.Y., 1956).
[16.61 E. T. Whittaker and G. N. Weteon, A e o m of [16.12] G. W.and R. M.Spenceley, Smitheonian elliptic
modern analysis, chs. 20, 21, 22, 4th ed. (Cam- function tsbl-, Smitheonian M i s ~ e l l a n e( 3~1~- ~
bridge Univ. Preee, Cambridge, England, 1952). laction, vol. 109 (Washington, D.C.,1947).