Escolar Documentos
Profissional Documentos
Cultura Documentos
RECI'ANGLE FIGURE
18.1 RHOMBUS
y=w
w=o+w' +w'- 0
*=w'
w REAL Q REAL
w' PURE MAG. 0; PURE IMAG.
10'1 2w, since g, 20 1w;I Lw, since g a L O
Fundamental Rectan6lea
Study of all four functions (@,@', f, u) can be reduced to consideration of their values in a Funda-
mental Rectangle including the origii (see 18.2 on homogeneity relations, reduction formulas and
p-1.
A>O OA
<
FIGURE18.2
There is a point on the right boundary of Fundamental Rectangle where @ =O. Denote it by 20.
WEIERSTRASS ELLIPTIC AND RELATED FUNCIIONS 631
18.2. Homogeneity Relations, Reduction The Case ga<O
Formulas and Processes Put t=i and obtain, e.g.,
{
Homogeneity Relations (Suppose t #O)
Note that Period Ratio is preserved. 18.2.16 @(2; 92, g3) = - @(iz; 92, -Sa)
18.2.1 @(tzltw, tw)=t-3@(z10,0) Thus the case g3<0 can be reduced to one where
18.2.2 @(tzlkd~fw)=t-@ (ZlaJ 93>0.
18.2.3 {(tzltw, tu)=t--lf(z(w, 0) Period Relations and Reduction to the FPP (M,N
integers)
18.2.4 u(tzltw, tu)=tu(zIw, w)
18.2.17 @(z+2Mw+2Nw)= @(z)
18.2.5 g2(ta, tw)=t--g2(w, w)
18.2.6 g3(tw, tw) =t-g3(0J @I 18.2.18 @(z+2Mw+2Nw) = @(z)
18.2.7 et&, to)=t-e(w,a),i=1, 2, 3 18.2.19
18.2.8 A(tw, tw)=t-12A(a, 0) (2+2 M w +2Nw)={(2) + 2 M 7 + 2N7
18.2.9 H,(tw, to)=t-2Ht(w, w), i=1, 2, 3 18.2.20
(See 18.3)
u(z+ 2 M w +2Nw)
18.2.10 q(tw, tw)=q(w, a) (See 18.10)
18.2.11 m(tw, tw)=m(w,a) (See 18.9)
= (- 1)+N+MN 4 4 exp [(z+Mw+Nw) ( 2 M v
+2N7)l
18.2.12 @(tz; t - l g ~ ,t-6g3)=t-3@(Z; 92, g3)
18.2.21 where q={(w), q={(w)
18.2.13 @(tz; t-4g2, t-g3)=t-@ (z; 92, g3)
ConjugateValueei
18.2.14 {(tz; t-lg2, t-g3)=t-{(z;gz, g3)
f(Z)=f(z), where f is any one of the functions
18.2.15 u(tz; T4g2,t-g3)=tu(z; 92, g3)
@, @,c, 0.
632 WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS
FIGURE18.3
18.2.38
@(zlw, w)=w-3@
(zw-(l, -3 @(zlw, w)=y-*@
(z& : I -t
3
-
18.2.39
@(zIo, w)=w-2@ (zw-11, g) @(zIw,o)=Wi2@
(ZWi-t
1::) -
18.2.4
r(Zlw,@)=w-{ (ZU-/~,
9
; r(ZIw,w)=Wir
)(: : : 1
ZW
; -j -
18.2.46
(
A(w, w)=w-A 1,
3
- A(@, W ) = ~ - A
(: 9-
3
-
050 (approx.)
18.3.11 ~w'~/~1.91014 si/$O if (w;~/w2~3.81915
447 (appmx.)
18.3.15 u(w')"e''u'll/Ha'l
18.3.16 Z(os)=ea "J(-Ha)
q'w n-
18.3.17 arg [u(w2)]=-
i +z
634 WEIERSTRASS ELLIPTIC AND RELATED FUN%TIONS
18.3.32 C(2~/3)=--
3
I(2 4 3 ) =-
-exp [2vw/9]
18.3.35 u(2w/3)=
[ @ ( 2 ~ / 3 ) ] / ~ e ~ ~ ~ / ~
-exp [21w/9]
u(2w/3)=
[@ (2~/3)]~/~e2*~/~
LegendresRelation
18.3.38 E+z+w=3g2/4
18.3.39 @lZ+@ZG+ZWG=O
WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS 635
18.3.M -AI16
18.4.1
18.4.3
18.4.5
18.4.9
18.4.10
18.5.1
18.5.2 where
and
3 k-2
18.5.3 C - crnck-mn, k l 4
'-(2k+l)(k-3) m-2
18.5.4
18.5.5
18.5.6
18.5.10 c,=3cac3/11
18.5.11 ce= [2c:+34]/39
18.5.12 C, =2&/33
18.5.13 CB =5 ~ (1
2 1t$+ 36~:)/7293
18.5.14 cg=c3(29c:+ llc:)/27 17
18.5.15 c~O= (242Ci+1455~g)/240669
18.5.16 ~11= 14~83
(3894+ 3694)/3187041
18.5.17 cl2=(1 14950c:+ 1 0 8 0 0 0 0 ~ ~166617~:)/891678645
~+
18.5.18 10~~3(297~:+530~:)/11685817
2C2(528770~:+7164675~$:+29896024)
18.5.19 c14= (306735) (215441)
4 ~ (6292181
3 5 4 +179865450~33+140513674)
18.5.20 CIS=
(179685)(38920531)
ci(589578554 +108651 1320c$:+ 875341836~2)
18.5.21
c1e= (5909761)(5132565)
~ # 3 (3017 1955~:+126138075~$:4-281 517394)
18.5.22 c17= (920205)(667867 1 )
1155&C:+ 122378650673*378~$g+2348703*887777~:
1541470*949003~~+30458088737
18-5-23 CIS=
(1342211013)(4695105713)
2&3( 3365544215 4 4-429852433 4 5 4 ~+8527743477&)
: *
18.5.24 c19= (91100295)(113537407)
a NOTES: 3. cI0 is given incorrectly in [18.12] (factor 13 is
1. c4-16 were computed and checked independently missing in denominator of third term of bracket);
by D. H. Lehmer; these were double-checked by this value was computed independently.
substituting g,=20 el, g ~ = 2 8ca in values given in 4. No factors of any of the above integers with more
[18.10]. than ten digits are known to the author. This is not
2. cITcI8 were derived from values in (18.101 by necessarily true of smaller integers, which have, in
the same substitution. These were checked (numer- many instances, been arranged for convenient use
ically) for particular values of g:, ga. with a desk calculator.
WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS
637
m
638 WEIERSTRASS ELLIPTIC AND RELATED FIJNCTIONS
18.5.29 z=A,u+A~u~+A~u'+A~u'+
...
18.5.33 A7=---4aaA1
7
18.5.34 Ag=O
18.5.35 All=%a3A:/11
10a1
18.5.36 Ala=- 39 (aS"+U)
+-723x7 (llcY;+loa:) u97
12 18.5.37 -96~.$@75
18.5.40
18.5.41
18.5.42
11 18.5.43
+= (1l054+ 16380&+ 1092OaXa:
18.5.44
+168~4)u a 7 + F(854+280&+56&) ua9 18.5.45
18.5.46
18.5.58
(@ '- @ A)=[ -1Qc2u-56c3u3+30~us+264cae~u'
(840Cl,-1004) U L -545gG ull
+ 3 11
+ @ E-
15~~u~-28c~u~+30c&~+
114~~3u'~
24404%
+7(12+54)u'"---d4]+. ..
11
18.5.60
3@#
(2 -zo)=A- bA3- 22 A4+3(q+bz)A6
+1004-279@ i2b+132b4p+. . .
(z
+21c@r+5&~~+ (18~&+3o&t
+3 3 c a e 3 ) u 4 + ~ ~ ~ + 9 ~ ~ l + 1 0 5 c :
+424e,+F 46) d1+70 Qe:
+..e,
18.5.66 where u= ( 2 - a ) , [
=5! @3-3c2@ -5
18.5.67 C0=C(Zg)
18.6.7
@(SI (2) = 12(@@ "'+2 @'@"+ @"@')
Series near uf
=3.5!@'[@2-c2]
18.5.68 18.6.8
-(7c2e:+21c8ef+5c:)a7
7
11 18.6.12
18.5.69
where a= (z-of) + (774@ +57~8)/331
18.6.15
Reversed Series for Small lul
18.5.70
z,c,+2 d+E lJ'I+3ra
--' 4
5 7 14
@ @ '-&a
(2)= 151 [@ '--8~2 @'+524 @ '/3
18.5.71 where y2=g2/48 3 a/l1+ (4444-3284) iP '139
+3 2 8 ~ @
-4884~8 @ /33+~,(55&-23164/429]
18.5.72 ~.~=ga/120 18.6.18
For reversion of Maclaurin series, see 3.6.25 and @ ('6)(~)=8.15!@'[@7-6~2@6-5C3@4+26d@ '/3
[18.181. +1 2 3 @ ~ + (11 14-824) @ /39-61&/33]
~ '/ll
WEIERSTRASS ELLfPTIC AND RELATED FUNCIXONE) 641
Partial Derivatives with Reepect to Invariants y"=[a@ (z)+b]y (LarntYs equation)--see [18.8],
2.26
18.6.19 For other (more specialized) equations (of
wders 1-3) involving @(z), see [18.8], nos.
1.49, 2.28,2.72-3, 2.439440, 3.9-12.
For the use of @ (2) in solving differential equa-
18.6.20 tions of the form y'"+A(z,y)=O, where A(z,y) is
a polynomial in y of degree 2m, with coefficienta
which are analytic functions of 2, see [18.7],
p. 312ff.
18.6.21 18.7. Integrals
Ind+$nite
A %= -33-
ai73
(g2@ +: g3) 18.7.1
18.6.22
If @ '(a) #O
18.6.24
18.7.3
4-3a3 18.7.5
Sa =- 27
18.6.27
FIGURE
18.4
In region A In region A
~ ( @ ) r O i f y 2 . 4 a n d s 1 . 5 9(@)20elsewhere
; (1) If a21.05, use criterion for region A for
A>O.
(2) If l<a<1.05: B ( g ) Z O if y 1 . 4 and
sI.4, -&<a173 (@)<34 if .4<y5.5 and
.4<25.5. Y(@)2 0 elsewhere
WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS 643
In region B In region B
The sign (indeed, perhaps one or more significant Use the criterion for region B for A>O.
h i t s ) of @' is obtainable from the first term;
-212, of the Laurent series for @'.
A>O w=l
Map: @(z)=u+iv
1
Near zero: @ (2)=2+t,
1
@ (2) = ~ + c z Z 2 + c 2
w'=i
w' = 1.4i
0' = 2.0i
644 WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS
OA
< m=l
Map: @(Z)=U+~D
1
Near zero: @ ( Z ) = ~ + C ,
1
p (2)=2+cta'+rr
FIQUBE18.6
WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS 645
A>O o=l
Map: c(z)=u+iw
1
Near zero: r(z)=-+tl
2
c(2)=---12 c22
3 +E2
FIGURE
18.7
646 WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS
A
O< ~z=l
Map: {(z)=u+iv
1
Near zero: {(z)=-+rl
2
1 c2.23
{(2)=---
2 3 +c2
FIGURE
18.8
WEIERSTRASS ELLIPTIC AND RELATED F"CI?IONS 647
A>O w=l
FIGURE
18.9
648 WEIERSTRABS ELLlPTIC AND RELATED FUNCTIONS
OA
< w=l
Map: u(z)=u+&
..-
FIGURE
18.10
WEIERSTRASS ELLJPTIC AND RELATED FUNCTIONB 649
18.9. Relations with Complete Elliptic Integrals K and K' and Their Parameter rn
and with Jacobi's Elliptic Functions (see chapter 16)
(Here K(m) and K'(m)=K(l-m) are complete elliptic integrals of the 1st kind; see chapter 17.)
A X K O
(2m-1) +6iJm -ma
18.9.1 el= (2--m)Ffm) el= m m )
3wa 34
(2m-l)E( m) 2(1-2m)F(m)
18.9.2 ea= ea=
3wa 34
16me(m-1)'P2( m) -256(m-ma)Fa(m)
18.9.6 A= A=
W12 a;,
18.9.7
18.9.13 q= l(w)==
3w
[3E( m) +(m -2)K( m)] qz=S(o2) =KO
3w2
[6E(m) +(4m--5) K(m)]
18.9.14
18.10.3
18.10.4 (V'lrZ/2@)
18.10.5
18.10.6
18.10.7
18.10.8
18.10.9
18.10.10
18.10.11
18.10.12
18.10.13
18.10.14
18.10.15
18.10.16
18.10.17
18.10.18
18.10.19
WEIERSTRASS ELLIPTIC AND RELATED FTJNCTIONSr 651
Series
18.10.20 &(O)=O
18.10.21 . . . +pncn+l)+
92(0)=2q~[i+q1az+f~3+pa~4+ . . .]
18.10.22 &(0)=1+2[q+q'+$+ fp1+. .I
{
18.10.23 84(0>=1+2[-q+q'-@+ . . . +(-1>"qn2+ .I
Attainable Accuracy
A>O K O
Note: S,(O)>O, j=2,3,4 Note: a2(0)=AefrJ8,A>b;
-
9&(0>>0; 84(0)=~3(0)
MO): 2 terms give at least 5s 2 terms give at least 3s
18.11.2
a constant. If any af or bi is congruent to the c gin, the corresponding factor is omitted from the
product. Factors corresponding to multiple pol@ zeros) are repeated according to the multiplicity.
18.12. Caee A=O(c>O) 18.12.9 p=l, m=l
Subcase I 18.12.10 rJ(o)=O
18.12.1 g2>0, g3<0: (el=e2=c, ea= -2c)
18.12.11
18.12.2 Hi =H2=0, H3=3~
18.12.3 18.12.12 .(w*)=O
@(z; 128,--88)=~+3c{sinh [(~C)*Z]}+ 18.12.13 9 (w/2)=c
18.12.4 @'(w/2)=0
-
18.12.14
T(z; 128,-88)=-~~+(3~)4 coth [(~C),Z] 18.12.15 f(w/2)= -
18.12.5 18.12.16 u(w/2)=0
"(2; 128,--88) = (3c)-i sinh [(3c)*z]e-"'/*
18.12.17 @ (w'/2) = -5c
18.12.6 w=m, w'=(12c)-*ri
18.12.7 q=f(o) = -
18.12.18
18.12.21
u(w/2)=
P(02/2)=c
n-
18.12.47
cw
r(w,/2)=-+i
2 -
18.12.22 @ (wzP>=O 18.12.48 u(w2/2)=o
cw Subcase I11
{
18.12.23 {(*/2)=- Q) --2
18.12.49 g2=0, g3=0 (el =e2 =e3=0)
18.12.24 u(w2/2)=o 18.12.50 P(2;0) O)=z-*
Subcaae I1 18.12.51 {(z; 0, 0) =z-1
18.12.25 18.12.52 u(z; 0, O)=z
g2>0, g3>O: (el=2c, e2=e3=-c)
18.12.53 (,$j= -iw=.n
18.12.26 Hi=%,
Ha=Ha=O
18.13. Eguianharmonic Case (g,=O, &=l)
18.12.27 @(z; la2,
88)=-~+3c{sk[(~c)+z]}
-2
If g2=O and g3>0, homogeneity relations allow
18.12.28 us to reduce our considerations of @ to@ (z; 0, 1)
{(z; 128, 88)=cZ+(3c)* cot [(~c)*z] (9) and u are handled similarly). Thus
@(z;G,g3)=g~/5~(zg~~o;0, 1). The case g2=0,
18.12.29 g3=1 is called the EQUIANHARMONIC case.
(2; 128, 88) = (3c)-+sin [(3c)+z]ecEa*
i FPP; Reductionto Fundamental Triangle
18.12.30 w=(12c)-+*, w=i
-
AI = AOez, is the Fundamental Triangle
18.12.31 v={(w) =cw
Let Q denote Cffl throughout 18.13.
18.12.32 =io,
q={(w)
~ = 1 . 5 2 9 954037 05719 28749 13194 17231
18.12.33 P=o, m=O
2wed124
18.12.34 u( w)=-
lr
18.12.35 U(W)=o
18.12.36 44 =o
18.12.37 g (w/2)=5c
18.12.38 9( w/2)=--9
2w3
18.12.39 T(w/2)=New+ n - b )
er21mwJZ
18.12.40 u( w/2)=
lr
18.12.41 9(w/2)= -c
18.12.42 gD ;w/t>
-3
FIGURE
18.11
18.12.43 f(w/2) =.if
18.12.44 =o
u(w/2) 6 This value was computed and checked by multiple
precision on a desk calculator and is believed correct to
18.12.45 gD (02/2)= -c 309.
Reductwn.for zz in 4: zl=e% iS in AI. Special Values and Formulas
18.13.9
18.13.1 @ (2%) =cap
(4 A= -27, &(4-'l3);,
H2=&(4-'13), Ha= &(4-'la)e
Values at Half-periods
JYJ I *@I I f
18.13.16
W E O l
18.13.17
w2
18.13.18
a' = wa
18.13.19
wal
18.13.20
202/9
18.13.21
w2/3
18.13.22
4~/9
18.13.23
0212 ea+ HZ
18.13.24
2~/3
18.13.25
8wd9
l1 - 43
18.13.26 e . / "fidr/6
42 -21l13 3i [5+21/8] Cir/e
3114
1
3~014 Z(a--Hz) i(38/4)4ai [;+31/4$7] 61.16
3"4Va3
18.13.28
20 i % e-'r/6 e-/sJi.dr/6
-
4* =
e)=i qT+
18.13.37 {(ez) =e-'t(z)
18.13.33 @
:--q ~
cos cos -
3
++=
18.13.38 u(ez)=eu(z)
@ '(2)
-2
Near zero: @ '(z)=-+Q
23
-2 23
@ '(z)=,+,+s,
T(z)
1
Near zero: j-(z)=-+e1
2
1 26
I(2)=;- KO2+
4 2)
656 WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS
EXACT car
I APPROXIMATE car
First 5 approximate values determined from exact values of c 3 k ; subsequent values determined by using exact ratios
using at least double precision arithmetic with a desk calculator. All approximate c's were checked with the
cak/csk-a,
use of the recursion relation; c 3 - c Z 7 are believed correct to at least 21s; c 3 0 - c 3 9 are believed correct to 205.
C3f 5 13t-I ". 28-1' k=2, 3, 4, . .
18.13.39 3 1143
u 3u2 5u3 7u4 63d +-;i we--
[
z=(@-1)1/2 I+-+-+-+-+-
7 26 38 40 248
+231u' + ~ + 0 ( " q '18.13.47 where w =( @-a)/3e2
3 g 542%'
Other Series Involving pr
18.13.M where u= @-a/8 and z is in the Funda- Revemed Series for Large I @ ' 1
mental Triangle (Figure 18.11) if @has an appro- 18.13.4
priate value.
Series near so
18.13.41
18.13.44
Series near wz [- - 2 - i ~ +5 z2+%
(Pr-i)=z ~
3i
$+o(z*)
1
6i
(@-e2)=3eiu
r
1+z+z2+;i 18.13.50 where z= (Z-Z~)~
132
4
+?5 z4+? z6+=285 z8+O(z7)
1 18.13.51 z=& l-k-; d+7+0(a4)]i
fl
-214*310*5*23*257-18049
43!
18.13.M where v = ~ ( z - w ) ~
18.13.55
(z-02)=( @'/sei) 1 - 2 w + W - 7360 18.13.66
[
+330w"-2268d+ 2FO(w7)], we+
18.13.56 where w= @ '2/9
Other Series Involving t Economized Polynomiab (0<~51.53)
Reversed Series for Large It1
18.13.57
~0=(-1)9.99999 96 ~4=-(-9)2.20892 47
18.13.58 y= p / 2 0
~1=(-2)3.57143 20 0,=(-10)1.74915 35
Series near so
18.13.59 a2=(-5)9.80689 93 o+,= -(-12)4.46863 93
~3=(-7)2.00835 02
18.13.69 ~(z)=T 0
~,Z""+e(z)
30192d
-- 1634208 P1P+0(w7)]r
275 +3575 1(~)1<3 x10-8
@ @I f U
18.14.8
W=Y e l 4 0 F*Ih eq2q
18.14.9
W"4 6-0 0 o+d gr/4(&)ei*/4
18.14.10
Wt.=cl* ea= -4 0 $= -* i / 4 ~ ie4q2u4)
@ @' f U
18.14.11
4 $(&+ 2114)(1 + 2 q
18.14.12 er/0(21/16)
sw+z
T P
012 a12 -a
af
18.14.13
18.14.15
er/M(21/83)
%I4 ++JG a (fi+ 42)e's14
&4(*+ 43114
e1r/4
18.14.16
%I2 -i/2 eir/4 [hU'2 2
~+!]e-,r/4 e*/16(21/8)eir/4
18.14.17
2413
-i
T dsec 30'-1
eir/4 4 263- 3 2+
-;[ @ (24/3)]fi er/oeir/4(39
P 3 1;1-
18.14.18
-42 (a-m
eOr/M(21/P)
a (fi- 42)ei.14 &4(&- .,/2)1/4 eir/4
32014
2
Near zero: @'(z)=-+i
-2
Near zo:@'(z)= -(2- 20)
ca
2 +
-2
@'(z)=,+s+*
z @'M=-(z-2 zo)+3( z40
-zo)*+ 4
h(z)
1
Near zero: f(z)=-+tl
Z
1 2-d
1(2)=;-@$-1' 121<
4 9
WEIERSTRASS ELLIPTIC AND RELATED FWNCl'IONS 661
Coefficients for Laurent Series for @, @', and 5
k EXACT @k APPROXIMATE Qr
-
1 1/20 .os
2 1/(3*20')= 1/1200 .8333 . . x10-'
3 2/(3-13*2@)= 1/156000 .641026 641025 . . . X10-6
4 5)(3.13.17.204)
5/(3*13-17.204) = 1/21216000 .47134 23831 07088 98944x10-7
5 2/(3'*13*17*2W) = 1/(31824X1W)
6 10/(3'.13'*17*20')
1-_,\_ I (4964544x 11W)
n I (28. I-_32.17.206) = 11/(4964544X W)
7 4/(3.13*.17.29.207) = ij(7998432
1/(7998432Xx 107) .12502 45048 02941 37661X10-1*
.75927 19109 76468 59917X10-16
8 2463/(3'*11.13*.17'*29.208) =968203125/(1262002599X 10")
9 138. 17t-29.37.200)= 833984376/(18394643943X 1017)
2.5.7.61/(38.13a.17t-29.37.200)
2.5.7.61/(3a. .46338 43633 93461 06092X 10-l8
r 6Zp 8$ 1729
v=y 1-y+---+-
5 5 7 5
Other Series Involving
Reversed Seriem for Large
18.14.47 z=T-'
18.14.37 18.14.48 v=p112
662 WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS
(p-) x x3
=------ x5 x7 59 %= (-4p.33352 77 aa= (-12)4.94511 45
2 6 20'70-240 %=(-6)6.49412 86
-__----
xll
o(x17),
11~13 216
825 31200 97503-
6
18.14.52 x= (2--0)
18.14.58 ."@'(x) =~~,x*"+E(x)
0
18.14.53
x.w--+---+---
23 7w6 13w' 929w9 1 9 4 ~ " 942883d3 lc(x)1<4X10-7
3 30 63 4536 891 + 3891888 ao= -2.00000 00 a,= (-7)6.58947 52
+0(w'5) a.l=~-~)l.ooooo02 a5=(-9)5.59262 49
18.14.54 w=-~({-v) %=(-3)4.99995 38 a6=(-11)5.54177 69
Series Involving u a,=(-5)6.41145 59
18.14.55
a=Z----
26 3'2' 3 . 2 3 ~ ' ~3 . 1 0 7 ~ ' ~33-7*23-37~21
6
2.5! 22-9!+2a.13!+2'.17! + 25.21!
18.14.59 xt(x)=Can2++ t(x)
0
32*313.503~" - 3'.7-685973~~
+ 28-25! 2'.29! +O(P)
(t(X)l<3Xlo-s
18.14.56 ao=(-1)9.99999 99 a.4=-(-9)2.57492 62
d 4 17*113d3
z=u+-+-+
24*34j 2'.3-7 213-34-7.11.13 al= -(-2)l.66666 74 ai= -(-11)5.67OO8 00
Hl=-i/a, H2=+,
H3=i/4, m=3, q = k r f 2 18.15.5 where R= Jw ~21,
@ @I I U
Numerical Methods
18.16. Use and Extension of the Tables
Example 1. Lemniscatic Case
(a) Given z=z+iy in the Fundamental Triangle, find @(@',{,u) more accurately than can be
done with the maps.
a-Use Maclaurin series throughout the Fundamental Triangle. Five terms give at least six
significant figures, six terms at least ten. @, {-Use Laurent's series directly "near" 0, (if (z(<l, four
terms give at least eight significant figures for @, nine for {; five terms at least ten significant figures for
@, eleven for {). Use Taylor's series directly "near" q,. Elsewhere (unless approximately seven or
eight significant figures are insufficient) use economized polynomials to obtain @ (z),@'(z) and/or {(z)
as appropriate. To get @(iy), @'(iy) and/or {(iy), use Laurent's series for "small" y, otherwise use
economized polynomials to compute @(y), @'(y) and/or {(y), then use complex multiplication to
obtain @ (iy), @'(iy) and/or {(iy). Finally, use appropriate addition formula to get @ (z) and/or l(z).
@'-Use Laurent's series directly "near" 0 (if z<l 1, four terms give at least $ix significant figures,
five terms at least eight significant figures). Elsewhere, either use economized polynomials and addition
formula as for @ and {, or get @ '2=4@3- @ and extract appropriate square root (g@'20).
(b) Given @(@', {, a) corresponding to a point in the Fundamental Triangle, compute z more
accurately than can be done with the maps. Only a few significant fiEsures are obtainable from the use
of any of the given (truncated) reversed series, except in a small neighborhood of the center of the series-
For greater accuracy, use inverse interpolation procedures.
Example 2. Equianharmonic Case
(a) Given z=z+iy in the Fundamental Triangle, find @ (@', I, a) more accurately than can be
done with the maps.
a-Use Maclaurin series throughout the Fundamental Triangle. Four terms give at least eleven
significant figures, five terms at least twenty one.
@,{-Use Laurent's series directly "near" 0 (if 1, four terms give at least 10s for @, 11s for {;
five terms at least 13s for @, 14s for {). Elsewhere (unless approximately seven or eight significant
figures are insuffcient) use economized polynomials to obtain @ (z), @'(z) and/or S(z), as appropriate.
To get @ (iy), @'(iy) and/or {(iy), use Laurent's series. Then us3 appropriate addition formula to get
@(z) and/or Hz).
664 WEIERSTRASS ELLIPTIC AND RELATED FUNCIIONS
@'-Use Laurent's series directly "near" 0 (if Izi< 1, four terms give at least 8S, five terms at least
11s). Elsewhere, either proceed aa for @ and 6, or get @'2=4@a-1 and extract appropriate square
root (JJ@ ' 20).
(b) Given @ (@',{,a) corresponding to a point in the Fundamental Triangle, compute z more
accurately than can be done with the maps. Only a few significant figures are obtainable from the use
of any of the given (truncated) reversed series, except in a small neighborhood of the center of the series.
For greater accuracy, use inverse interpolation procedures.
Example 3. Given period ratio a, find parameters m (of elliptic integrals and Jacobi's functions
of chapter 16) and p (of 9 functions).
m-In both the cases A>O and A<O, the period ratio is equal to K'(m)/K(m)(see 18.9). Knowing
K'IK, if 1< K'/K$3, use Table 17.3 to find m; if K'/K>3, use the method of Example 6 in chapter 17.
An alternative method is to use Table 18.3 to obtain the neceasary entries, thence use
m= (&-ea)/(el-ea) in w e A>O,
m=3--3e&H2 in case A<O.
q-In both the cases A>O and AO <, the period ratio determines the exponent for p[p=e-*" if A>O,
p = i ~ f l /if~ AO
< .] Hence enter Table 4.16 [e-, 2=0(.01)1] and multiply the results as appropriate
[e.g, , e-' .nr= (e-*) '(e- .'*)I
Determination of Valuee at Half-Periods, Invariante and Related Quantities from Given Periods (Table 18.3)
A>O OA
<
Given w and w', form d / i w and enter Table 18.3. Given wl and w', form w'liw, and enter Table
Multiply the results obtained by the appropriate 18.3. Multiply the results obtained by the appro-
power of u (see footnotes of Table 18.3) to obtain priate power of wl (see footnotes of Table 18.3)
value desired. to obtain value desired.
Example 4. Faample 4.
Given 0=10, w ' = l l i , find et, gt, and A. Given *=lo, ~ ' = 1 1 3 , find et, gt, and A.
Here w'/iw=l.l, so that direct reading of Table so that direct reading of Table
Here d/&=l.l,
18.3 gives 18.3 gives
el(l)=1.6843 041 el(l) = -.2l66 25764-3.0842 589;
e2(l)=-.2166 258 (=-el-es) &(l)=.4332 5152= -29?(el)
ea(l)=-l.4676 783 e&) =&(U
g2(1)=10.0757 7364 g2(l) = -37.4874 912
ga(1) =2.1420 1000. ga(1) = 16.5668 099.
Multiplying by appropriate powers of @=lo we Multiplying by appropriate powers of %=lo
obtain w0 obtain
el=.01684 3041 el= -..00216 62576+.03084 25893
e2= -.00216 6258 e2=.00433 25152
ea= -.01467 6783
-
es=el
g2=1.O075 77364 x p=-3.7487 4912 x lo-'
ga=2.1420 1000 x gae1.6566 8099 x lo-'
whence whence
A=8.9902 3191 x A=-6.0092 019 x lo-'
Example 5. (A>O) Example 5. O
)A
<(
Given w=lO, w'=553, find q, q', a(w), u(w') Given w=lOOO, w,'=1004i, find ql, qz', u(w),
and u ( w ) . u(w') and u(w').
Forming w'/iw=5.5 and entering Table 18.3 we
obtain 1= .82246704, u(w) = .96045 40. using With 4 i w = 1.004, four point interpolation in
Legendre's relation we find q'=qw'-&/2=2.9527 Table 18-3gives 72-1.5626 756,&= -1.5726 6643,
723i. Since interpolation for u(w') and u(w+w') u(w)=l.l805 028, u(w;)=1.1901 52i and u(w')
is diflicult,, use is made of 18.3.15-18.3.17 together =~ 7 5 0 8 +
4 ,4767173.
with 18.3.4 and 18.3.6. VdUB of gi, g3 and el Can MdtiplSing the results obtabed by the appm-
be read directly to eight significant figures and eS
to about five significant figures giving gz=8.1174 Piate Powem Of Y gives ~ ~ = * 0 0 1 526756,
6 7;
243, ga=4.4508 759, e1=1.6449 341, and ea =--.00157 266% d0a)=1180-5028, (.w);
= -32247. Use of 18.3.6 'elds H8=.00174 69 =1190.1523 and u(w')=475.084+476.717d
3:
and H1=.00174 693. App 'cation of 18.3.15-
18.3.17 yields u(w')/3=.0071177 and u(m)
= -.002016--01055i. Multiplying the results
obtained by the appropriate powers of w we ob-
tain q=.08224 6704,7'=.29527 7233, u(w) ~ 9 . 6 0 4 5
40, u(w')=.O711773 and u(w)=-.02016-.1055i.
A>O O< A
Given g>lO and g3>0 such that A=$-27&>0 Given g1 and ga>O such that A=$-27&<0 (if
-(ifga=O, [w'l=w; see lemniscatic case), compute ga=O,Jw'l=w,; see seudo-lemniscatic case), com-
gl=gg;2/a. From Table 18.1, determine wg:/' pute gl=ggrZD. &om Table 18.1, determine
and wg'",' thence w and w'. Uzgall' and w,'galtd, thence w, and w,'.
Example 6. Example 6.
Given g2=10, g3=2, find w and w.' With g2=
gg3-2m=6.2996 05249, from Table 18.1 wg:/'=
1.1267 806 and w'g3"'=1.2324 2953 whence w=
'-
Given 2=-10, g3=2, find w, and 4. With
-gZ=g293-2 --1011.5874 0105= -6.2996 053, from
TabIe 18.1 0lga~~'=1.5741 349 and ~.&~lJ8=1.7124
1.003847 and w'= 1.0979703. 3963 whence w,=1.40239 48 and W;=1.52561 O2i.
Example 7. Example 7.
Given g2=8, gs=4, find w and w.' With Given g2=7, a=6, find w and W;.
Pi =g2gs-2m=3.1748 02 104, from Table 18.1 WJ/~= With 5i=gig3-"=7/3.3019 2725=2.119974, from
1.2718 310 and ~'ga'/"=1.8702 4253 T d l e 18.1 w~g2/'=1.3423 442 and w~'ga'"=3.1441
whence W= 1.009453 and w'= 1.4844133. 141i whencew=.99579 976 and 4=2.33241 83i.
666 WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS
{
Using the truncated approximations 18.10.21- The theta functions are computed for argument
18.10.23 we compute the theta functions for zero using 18.10.21-18.10.23 a n d the theta
argument zero. Using 16.27.1-16.27.4 we com- functions for arguments ul and v2 corresponding
pute the theta functions for arguments 8 where to z=zl+z2using 16.27.1-16.27.4. Using 18.10.5-
z=z and z=iy. Then, with 18.10.kl8.10.7 18.10.6 together with 1810.10, we find @ (3)=
together with 18.10.9 and 18.10.18 we obtain .lo576 946, @(2i)=-.24497 773, @(3),=
{(9)=.09889 5484, {(19i)=-.00120 0155i, -.07474140, @(2i)=-.25576 007;. The addi-
@(9)=.01706 9647, @(9)=-.00125 8460, tion formula 18.4.1 yields @(3+2i)=.01763 210
@ (19i)=-.00861 2615, @(19i)=-.00003 757i. -.07769 187i, and 18.4.2 yields @(3+2i)=
Using the addition formula 18.4.3, we obtain -.00069 182 t.04771 305;.
{(9+ 19i)= .07439 49- .00046 88i.
Use of Table 18.2 in Computing @, @,for Special Period Ratios
If the problem is reduced to computing @, @ , in the Fundamental Rectangle for the case when the
real half-period is unity and pure imaginary half-period is ia, for certain values of a Table 18.2 may be
used. Consider @ as an example. If lzl is small,then use Laurentsseries directly for @ (2) [invar-
iants for use in the series are given in Table 18.31.
If z is largeand y smalluse Table 18.2 to obtain 2@ (z) and 2@(z), thence @ (2) and @(z) ;
use Laurentsseries to obtain @ (iy) and @(iy); finally, use addition formula 18.4.1.
For z smalland y large,reverse the procedure. For both 2 and y large,use Table 18.2 to
obtain @(z), @(x), @ (iy) and @(iy), thence use addition formula 18.4.1.
Similar procedures apply to @or {. For @, one can also first obtain @, then compute @
=4@3-g2@-g3 and extract the appropriate square root (see 18.8 re choice of sign for @).
A>O A< 0
Example 11. Compute @(.8+i) when a=1.2. Example 11. Compute @(.9+.li) for a=1.05.
Using Table 18.2 or Laurentsseries 18.5.1-4 with Using Table 18.2 or Laurentsseries 18.5.1-4 with
g2=9.15782 851 and g2=-42.41653 54 and
g3=3.23761 717 from Table 18.3, g3=9.92766 62 from Table 18.3,
@ (.8) = 1.92442 11, @ (.9)= .34080 33,
@(A) = -2.76522 05, @ (.9)= -2.164801,
@(i)= -1.40258 06 and @ (.li)= -99.97876,
@(i)=-l.19575 58i. Using the addition for- @ (.li)= -2000.4255i. With the addition for-
mula 18.4.1 mula 18.4.1
@ (.8 +i)= -.381433- .149361i. @ (.9f.li)= .231859-.215149i.
Example 12. Compute {(.02+3i) for a=4. Example 12. Compute @(.4+.9i) for a=2.
Using Table 18.2 or Laurentsseries 18.5.1-5 with Using Table 18.2 or Laurents series 18.5.1-4,
g2=8.11742 426 with
g3~4.45087587 g2=4.54009 85,
from Table 18.3, g3=8.38537 94
{(.02)=49.99999 89, from Table 18.3,
@ (.02)=2500.00016, @ (.4)=6.29407 07,
@(.02)= -249999.98376, @(.4)=--30.99041,
((34 = .89635 173i, @(.9i)=-l.225548,
@ (3i)= -.82326 51 1, @,(.9i)= -3.19127 03;.
@ (3;) = -.00249 829i. Using the addition formulas 18.4.1-2,
Applying the addition formula 18.4.3, @(.4+ .9i)= 1 .lo519 76 -.56489 OOi.
{(.02+3i) = .016465+.89635i.
668 WEIERSTRASS ELLIPTIC AND RELATED FUNCi?IONS
(or periods from which g2 and ga can be computed-in any case, periods must be known, at least
approximately)
F h t reduce the problem (if necessary) to computation for a point z in the Fundamental Rectangle
(see 18.2). After final reduction let z denote the point obtained.
A>O < 0
A
If Wz>u/2 orl If 9%2>~/2 or
Iz>d/4, use duplication formula as in case
Iz>u/2, use duplication formula A>O. Otherwise, use Maclaurin series for u
directly.
U(2) = - @ (42)u(zp) ,
e=-
3
.lo115 7160-.04190 06673i
WEIERSTRASS ELLIPTIC AND RELATED FUNCTIONS 669
A< 0
Stopping with the term in 21) zl=..81+.233.
Assuming Az=-.03-.013, using 18.5.1, @(.81+
.233)=.91410 95-.86824 37i, @ (.78+.22i)=
1.03191 60- .91795 22i; with inverse interpolation
z:*)=. 77254- .2404i. Repeated applications of
inverse interpolation yield z=. 772247-.239258{.
Example 15. Given c=10-153, gz=8, g3=4, Example 15. Given u=.4+.13, g2=7, g3=6,
find z. Using the reversed series 18.5.90 with find z. Using the reversed series 18.5.70 with
y2=.14583, y3=.05
&=-.13333 333,
u=+.40000 OOO+.lOOOO 0003
A,= -.02857 14286,
u= .03076 923076+.04615 3846153
?d=
5
+.00011 783+.00032 6963
192y3u11=-.00000013+.00000 006i
55
z=.40011 469+.10034 2603
Methods of Computation of @ (@, or u) for Given ratio a is not too large. However, if aZ&(A>O)
5 and Given 82, g3 (or the equivalent), with the Uee
or ah245(A<O), the series will diverge at the
of Automatic Digital Computing Machinery
far corner of the Fundamental Rectangle, so that
(a) Integration of Differential Equation use may be made of an appropriate duplication
@ and @may be generated for any z close formula. Alternatively, one may compute the
enough to a known point z*(@ (z*) and@(z*) functions on Ox and Oy, then use an addition
being given) by integrating @=6 g2-gZ/2. formula. Even so, the series will diverge at
A program to do this on SWAC, via a modifica- z=ia if a22(A>O) and at 2=ia/2 if ah4(A<O).
tion of the Hammer-Hollingsworth method For great accuracy, multiple precision operations
(MTAC, July 1955, pp. 92-96) due to Dr. P. might be necessary. Double precision floating
Henrici, exists at Numerical Analysis Research, point mode has been used in a program, written
UCLA (code number 00600, written by W. L. for SWAC, to compute @ , @ andc.
Wilson, Jr.). The program has been tested For computation of u, use of the Maclaurin
numerically in the equianharmonic case, using series throughout the Fundamental Rectangle is
integration steps of various sizes. For example, probably simplest (series converges for all z).
if one starts with z*=oa, using an integration Mention should be made of the possible use of
step(h,k), where h and k are respectively the the series defining the 9 functions. These series
horizontal and vertical components of a step, converge for all complex v, and the computation
with (h,k) having one of the six values (k2h0,0), of @, @, and u by 18.10.5-18.10.8 could easily
(fho,fko), b=oa/2OOO, k0=)4/2000, one can be mechanized. The series involved have the
expect almost 8s in @ and 7 s in @ after 1000 advantage of converging very fast, even in case
steps, unless z is too near a pole. AO
<, where IpIS e-r/2(qj e- if A>O).
(b) Use of Series
The process of reducing the computation prob- Use of Maps
lem to one in which z is in the Fundamental If the problem (of computing @, @, 5 or CT for
Rectangle can obviously be mechanized. Inside given z) is reduced to the case where the real half-
the Fundamental Rectangle the direct use of period is unity and imaginary half-period is one
Laurentsseries is appropriate when the period of those used in the maps in 18.8 inspection of the
670 WEIERSTRASS ELLIPTIC AND RELATED FUNCTION8
appropriate figure will give the value of @(z) approximation to lzol by Graeffe's process, we
[l(z) or a(z)] to 2-3s. If @' is wanted instead, may use the fact that zo=w+iyo(A>O), %=y
get @,use 18.6.3 to obtain @ and selectsign (8) +iyo(A<O) to obtain an approximation to 4.
of @' appropriately. (See Conformal Mapping It is noted that yo/w is a monotonic decreasing
(18.8) for choice of sign of square root of @"I. function of (period ratio) a2 1 for A>O and
Computation of ~0 [I ZY~/W>;
2 tlrcmsh @(%.7297)).
Given 92, (or equivalent)
Since 8@ (h)=o, the Laurent's series gives yo/@ is a monotonic increasing function of a for
A
O< and
o=l+~u~+c,ua+c,u'+ . , . 2
[OSYo/%;< ~ C C o S h431
where u=%. We may solve this equation [by
Graeffe's (root-squaring) process or otherwise] Further data is available from Table 18.2 or from
for ita absolutely smallest root [having found an Conformal Maps defined by ,p(2).
References
Texts and Articles (18.141 E. 5. Selmer, A simple trisection formula for the
elliptic function of Weierstrass in the equian-
(18.11 P. Appell and E. Lacour, Principes de la thQrie des harmonic case, Norske Vid. Selsk. Forh. Trond-
fonctions elliptiques et applications (Gauthier- heim 19,29, 116-119 (1947).
Villars, Paris, France, 1897).
(18.21 A. Erdhlyi et al., Higher transcendental functions, [18.15] J. Tannery and J. Molk, ElCments de la thhrie des
vol. 2, ch. 13 (McGraw-Hill Book Co., Inc., fonctions elliptiques, 4 vols. (Gauthier-Villam,
New York, N.Y., 1953). Paris, France, 1893-1902).
[18.3] E. Grrreser, Einfiihrung in die Theorie der ellip- [18.16] F. Tricomi, Elliptische Funktionen (Akademkche
tischen Funktionen und deren Anwendungen Verlagsgesellschaft, Leipeig, Germany, 1948).
(R. Oldenbourg, Munich, Germany, 1950). (18.17 F. Tricomi, Fundoni ellittiche, 2d ed. (Bologna,
(18.41 G. H. Halphen, Trait6 des fonctions elliptiques et Italy, 1951).
de leurs applications, 1 (Gauthier-Villars, Paris, [18.18] C. E. Van Orstrand, Reversion of power series,
France, 1886). Phil. Mag. (6) 19, 366-376 (Jan.-June 1910).
(18.51 H. Hancock, Lectures on the theory of elliptic
functions, vol. 1 (John Wiley & Qons, Inc., New (18.191 E. T. Wbittaker and G. N. Watson, A course of
York, N.Y., 1910, qeprinted, Dover Publica- modern analysis, ch. 20, 4th ed. (Cambridge
tions, Inc., New York, N.Y., 1968). Univ. Press, Cambridge, England, 1952).
(18.61 A. Hurwitz and R. Courant, Vorlesungen flber
allgemeine Funktionentheorie und elliptische
Funktionen, 3d ed. (Springer, Berlin, Germsny, Guidcs, C o l l e c t i ~of~Formulae, ctc.
1929).
[18.7] E. L. Ince, Ordinary differential equations (Dover 118.201 P. F. Byrd and M. D. Friedman, Handbook of
Publications, Inc., New York, N.Y., 1944). elliptic integrals for engineers and physicists,
(18.81 E. Kamke, Differentialgleiehungen, Lijsunp Appendix, sec. 1030 (Springer-Verlag, Berlin,
methoden und Lijsungen, vol. 1, 2d ed. (Akade Germany, 1954).
mische Verlagsgesellschaft, Leipdg, Germany, 118.211 A. Fletcher, Guide to tables of elliptic functions,
1943). Math. Tables Aids a m p . 3, 247-249 (1948-49).
(18.91 D. H. Lehmer, The lemniscate constant, Math. [18.22] S. Flolgge, Handbuch der Phyaik, vol. 1, pp. 120-
Tables Aids Comp. 3, 5-551 (1948-49). 146 (Springer-Verhg, Berlin, Germany, 1966).
(18.101 8. C. Mitra, On the expansion of the Weierstrassian
and Jacobian clliptic funations in powers of the [18.23] H. Kober, Dictionary of conformal representations
(Dover Publicatiom, Inc., New York, N.Y.,
argument, Bull. Calcutta Math. Soc. 17, 159-172
1952).
(1926).
(18.111 F. Oberhettinger and W. Magnus, Anwendung der [18.24] L. M. Milne-Thomson, Jacobian elliptic function
elliptischen Funktionen in Physik und Technik tables (Dover Publications, Inc., New York,
(Springer, Berlin, Germany, 1949). N.Y., 1960).
118.121 Q.Prasad, An introduction to the theory of elliptic [18.26] K. Weierstrsss and H. A. Schware, Formeln und
functions and higher transcendentals (Univ. of Lehrsiitse sum Gebrauche der elliptischen Funs
Calcutta, India, 1928). tionen. Nach Vorleaungen und Aufseichnungen
(18.131 U. Richard, Osservaeioui sulla bisesione delle dea Herm K. Weierstrass bearbeltet und
fundoni ellittiche di Weierstrass, Boll. Un. Mat. herauegegeben von H. A. Schwarr, 2d ed.
It.3. 3, 4, 395-397 (1949). (Springer, Berlin, Germany, 1893).
WEIERSTRMS ELLIPTIC AND RELATED P"CI'ION6 671
Tables [18.28] T. I. Southard, Approximation and table of the
[18.26] Chih-Bing Ling, Evaluation at half-periods of Weierstress @ function in the equianharmonic
Weierstrass' elliptic function with rectangular case for real argument, Math. Tables Aids Comp.
primitive period-parallelogram, Math. a m p . 149 11, 58, 99-100 (Apr. 1957). ! ( u ) = @ ( u ) - ~ 1
69, 67-70 (1960). Values of e i (i=l, 2, 3) to 15D
for various period ratios in case A>O. to 7D . with modified central differences,
118.271 E. Jahnke and F. Emde, Tables of functions, 4th ~c=0(.1).8(.05)1.55.
ed., pp. 1W106 (Dover Publications, Inc., New
York, N.Y., 1945). Equianharmonic case, real (18.291 D. A. Strayhorne, A study of an elliptic function
(Thesis, Chicago, Ill., 1946). Air Documents
agument,@(u),P(u), t(u),u(u),u=~(%) Division T-2, AMC, Wri ht Field, Micro-
film No. b734F15000. &(e; 37, -42), 4D,
h 1 mostly 4D. t=.04i(.04~31.3%
3