57 views

Uploaded by Renee Bravo

- General Math DLL for SHS - (more dll at depedtambayanph.blogspot.com) Q1, Week 10.xlsx
- Arithmetic Geometry.pdf
- Ecrin 4.10 Update Notes
- Introduction to Functional Programming Using Haskell - Richard Bird
- Hash Function
- Cfx9970 Chapter16 En
- Hack Reactor MakerSquare Telegraph Academy Questions and Solutions
- Introduction to the Theory of Groups - G. Polites
- Business15 Exponential Functions
- Real Functions - Prelim 16362871361872
- ConMcohomology1.pdf
- precalculus practice 2
- Tangential Algebras and Tropical Potential Theory.pdf
- Arrival of the Fittest
- 1.3 Properties of Parent Functions 1.4 Domain Range
- Kilpatrick Algebra Game
- Representation Benefits of Deep Feedforward Networks
- Review Double Integration
- 20378737766
- individualized instruction project

You are on page 1of 26

http://www.elsevier.com/locate/jnt

Natália Archinard

McGill University, 805 Sherbrooke St. W., Montreal, QC, Canada H3A-2K6

Received 14 February 2002; revised 9 October 2002

Communicated by H. Darmon

Abstract

algebraic points at which the series takes algebraic values (the so-called exceptional set). This

answers a question of Siegel in special cases. For this, we ﬁrst prove identities, each one

relating locally one hypergeometric series to modular functions. In some cases, the identity

and the theory of complex multiplication allow the determination of an inﬁnite subset of the

exceptional set. These subsets are shown to be the whole sets in using a consequence of

Wüstholz’s Analytic Subgroup Theorem together with mapping properties of Schwarz triangle

functions. Further consequences of the identities are explicit evaluations of hypergeometric

series at algebraic points. Some of them provide examples for Kroneckers Jugendtraum.

r 2003 Elsevier Science (USA). All rights reserved.

MSC: 11

Introduction

XN

ða; nÞðb; nÞ n

F ða; b; c; zÞ ¼ z;

n¼0

ðc; nÞð1; nÞ

Q

where ðx; nÞ :¼ nj¼1 ðx þ n jÞ and a; b; cAQ; ceN; deﬁnes an analytic function

of z on the unit disc and satisﬁes a linear differential equation with monodromy

0022-314X/03/$ - see front matter r 2003 Elsevier Science (USA). All rights reserved.

doi:10.1016/S0022-314X(03)00042-8

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 245

group Dða; b; cÞ: Schwarz [20] showed that this function is algebraic exactly when its

monodromy group is ﬁnite. Hence, for most values of the parameters a; b; c; this

series deﬁnes a transcendental function. As for other G-functions, it has been

believed that transcendental hypergeometric series would take transcendental values

at algebraic points. Siegel [22] asked the question of the cardinality and of the nature

of the so-called exceptional set:

% F ða; b; c; zÞAQg:

Eða; b; cÞ :¼ fxAQ; %

Using a consequence ([29, Satz 2]) of Wüstholz’s Analytic Subgroup Theorem [30],

Wolfart [27,28] showed, with some conditions on a; b; c; that the points z in Eða; b; cÞ

correspond to isogenous abelian varieties TðzÞ of the same dimension, which are

deﬁned over Q % and have complex multiplication (CM) in the sense of Shimura–

Taniyama (see further details in [10]). Let Zða; b; cÞ be the closure of the image under

z/TðzÞ of P1C f0; 1; Ng on the Siegel moduli variety of such abelian varieties, as

studied for example by Cohen and Wolfart [8]. The study of Eða; b; cÞ is then reduced

to that of the CM points on Zða; b; cÞ; that is, of the points corresponding to abelian

varieties with CM. Along these lines, Wolfart proved the inﬁniteness of the

exceptional set Eða; b; cÞ in the cases where the monodromy group Dða; b; cÞ is

arithmetic and the conditions co1; c4a40; c4b40; j1 cj; jc a bj; ja bj

AZ1 ,f0g and j1 cj þ jc a bj þ ja bjo1 hold. Under the latter conditions, the

monodromy group is triangular. The 85 conjugacy classes of arithmetic triangle

groups were determined by Takeuchi [24]. Wolfart also asserted that the exceptional

set can only be inﬁnite when the monodromy group is arithmetic, but a gap in his

proof was found by Gubler. Cohen and Wüstholz [10] proposed a weak version of

André–Oort conjecture to show that a nonarithmetic group implies a ﬁnite

exceptional set. This weak conjecture for an irreducible variety Z of dimension 1

states roughly that if Z contains a Zariski dense subset of CM points corresponding

to isogenous abelian varieties, then Z is of Hodge type (for a precise statement, see

[10]). It was proved recently by Edixhoven and Yafaev [12]. An exceptional set for

Appell–Lauricella hypergeometric series in two variables is deﬁned by Desrousseaux

[11] in an analogous way, so that the points of the exceptional set correspond to

abelian varieties with CM. Part of Desrousseaux’s work relies on a weak André–

Oort conjecture for CM points on surfaces in Shimura varieties as given in Cohen [7].

After the result of Wolfart et al. [6], Flach [13], Joyce and Zucker [16] published on

special algebraic values of hypergeometric series with monodromy group isomorphic

to SL2 ðZÞ: Properties of n-adic evaluations of hypergeometric series were also

studied by Beukers [5] and André [2].

The purpose of the present article is the explicit determination of the exceptional

sets (Theorem 5.4) for a family of hypergeometric series whose monodromy group is

isomorphic to SL2 ðZÞ; namely the family of hypergeometric series which appear in

Kummer’s solutions to the Picard–Fuchs differential equation of the universal family

of elliptic curves. These series are transcendental functions and there are 16 of them.

The exceptional sets will be shown to be trivial in the cases where c41 (Section 5). In

the cases with co1; the exceptional set is inﬁnite by Wolfart’s result. In these cases, a

ARTICLE IN PRESS

246 N. Archinard / Journal of Number Theory 101 (2003) 244–269

a result of Wüstholz on linear independence of periods to Euler’s integral

representation of the hypergeometric series viewed as a quotient of periods on

abelian varieties deﬁned over Q% (this idea is due to Wolfart [28]). The construction of

the abelian varieties will be made precise in Section 2. The exact knowledge of the

lattice of these complex tori up to isogeny (Section 2) together with that of the image

of the Schwarz maps (Section 1) are essential for the translation of the necessary

condition in usable terms (Section 5). On the other hand, the sufﬁcience of these

necessary conditions on z will be shown in Section 4.1 via identities. These identities

(Section 3) relate locally one hypergeometric series to the modular functions J and

Z2 : They contain information on the arithmetic nature of the hypergeometric series.

For instance, they allow us to calculate explicit evaluations of hypergeometric series

at algebraic points, some of them providing examples for Kroneckers Jugendtraum

(Sections 4.2 and 4.3). For this, we ﬁrst have to compute the values of JðmiÞ and

ðZðmiÞ 2

ZðiÞ Þ for some m (Section 4.2). Some results of Sections 2, 3, 4.1 and 4.2 are taken

from the author’s Ph.D. thesis [4].

integral representation due to Euler, which reads, for jzjo1;

Z 1

1

F ða; b; c; zÞ ¼ xb1 ð1 xÞcb1 ð1 zxÞa dx; ð1Þ

Bðb; c bÞ 0

R1

where Bða; bÞ ¼ 0 xa1 ð1 xÞb1 dx is the Beta-function. The two integrals

converge for ReðcÞ4ReðbÞ40: F ða; b; c; zÞ is the holomorphic solution taking value

1 at 0 to the so-called hypergeometric differential equation

d 2u du

Equða; b; cÞ : zð1 zÞ þ ðc ða þ b þ 1ÞzÞ abu ¼ 0: ð2Þ

dz2 dz

The only singular points of Equða; b; cÞ lie at 0; 1; N and are regular singularities.

The monodromy group of Equða; b; cÞ will be denoted by Dða; b; cÞ: If a; b; cAR;

j1 cj; jc a bj; ja bjAZ1 ,f0g and j1 cj þ jc a bj þ ja bjo1 hold, then

Dða; b; cÞ is a triangle subgroup of SL2 ðRÞ of type given up to permutation by

1 1 1

ðj1cj; jcabj; jabjÞ: If a; b; cAR and j1 cj; jc a bj; ja bjo1; it is classically

known (see [19, V.7, VI.5]) that the quotient of two linearly independent solutions to

the hypergeometric differential equation maps a complex half plane bijectively to the

interior of a triangle whose sides are arcs of circles. Such a map is called a Schwarz

triangle map and the triangle a Schwarz triangle. The real line is mapped to the

boundary of the Schwarz triangle and the points 0; 1; N to the vertices. In particular,

if a; b; cAQ; j1 cj; jc a bj; ja bjAZ1 ,f0g and j1 cj þ jc a bj þ ja bjo1;

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 247

Bð1 þ a c; 1 aÞF ð1 þ a c; 1 þ b c; 2 c; zÞ

Dabc ðzÞ :¼ ð1Þ1þaþbc z1c

Bðb; c bÞF ða; b; c; zÞ

If ca1; we have trivially Dabc ð0Þ ¼ 0: In order to calculate the vertices at the images

of 1 and N; we need to know the analytic continuation of F ða; b; c; zÞ at these points.

The analytic continuation at 1 is given by (cf. [26, 14.11])

GðcÞGðc a bÞ

F ða; b; c; 1Þ ¼ if c a b40

Gðc aÞGðc bÞ

8

< GðaþbÞ 1

GðaÞGðbÞ logð1zÞ if c a b ¼ 0;

F ða; b; c; zÞB

: GðcÞGðaþbcÞ ð1 zÞcab if c a bo0:

GðaÞGðbÞ

a; beZp0 and a beZ (cf. [1, 15.3.7])

GðaÞGðbÞ GðaÞGðb aÞ

F ða; b; c; zÞ ¼ ðzÞa F ða; 1 c þ a; 1 b þ a; z1 Þ

GðcÞ Gðc aÞ

ð3Þ

GðbÞGða bÞ

þ ðzÞb F ðb; 1 c þ b; 1 a þ b; z1 Þ

Gðc bÞ

GðcÞ XN

ða; nÞð1 c þ a; nÞ n

F ða; a; c; zÞ ¼ ðzÞa z

GðaÞGðc aÞ n¼0 ðn!Þ2 ð4Þ

½lnðzÞ þ 2cðn þ 1Þ cða þ nÞ cðc a nÞ;

ARTICLE IN PRESS

248 N. Archinard / Journal of Number Theory 101 (2003) 244–269

0

ðzÞ

where cðzÞ ¼ GGðzÞ is the logarithmic derivative of GðzÞ: Using this and some classical

properties of the Gamma function (e.g. [26, XII]), we calculate

8 sinðbpÞ

>

> ð1Þ1cþaþb sinððcaÞpÞ if c a b40;

<

Dabc ð1Þ ¼ 1 if c a b ¼ 0;

>

>

:

ð1Þ1cþaþb sinððcbÞpÞ

sinðapÞ if c a bo0

and, if a; beZp0 ;

8

>

> 1 if a ¼ b; c aeZ;

<

aþb sinððcbÞpÞ

Dabc ðNÞ ¼ ð1Þ sinððcaÞpÞ if b a40; a beZ;

>

>

:

ð1Þaþb sinðbpÞ

sinðapÞ if b ao0; a beZ:

These results will be used in Section 5 in order to determine explicitly the Schwarz

triangles for the considered tiples ða; b; cÞ; see table (23).

Following the idea of Wolfart [27,28], we write Euler’s integral representation (1)

of the hypergeometric series as

PðzÞ

F ða; b; c; zÞ ¼ ;

Pð0Þ

R1

where ReðcÞ4ReðbÞ40 and PðzÞ ¼ 0 xb1 ð1 xÞcb1 ð1 zxÞa dx: If b; c beZ;

this integral can be replaced up to an algebraic factor by an integral along a double

contour loop g0;1 (or Pochhammer loop, cf. [31, IV.2.2]). PðzÞ can then be viewed up

to an algebraic factor as the integral of a differential form dx y on a nonsingular

algebraic curve and, if dx

y is regular, as a period on the Jacobian variety of this curve.

More precisely, for zAP1C f0; 1; Ng; let Cabc ðN; zÞ be the projective algebraic curve

deﬁned afﬁnely by the equation

yN ¼ xA ð1 xÞB ð1 zxÞC ;

we assume the following hypotheses:

a; b; cAQ Z; c a; c beZ;

which imply

N[A; B; C; A þ B þ C:

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 249

general, Cabc ðN; zÞ is singular at the points ðx; yÞ ¼ ð0; 0Þ; ð1; 0Þ; ð1z; 0Þ; N: We then

let Xabc ðN; zÞ be its desingularization and p : Xabc ðN; zÞ-Cabc ðN; zÞ be the

desingularization morphism. Both curves are deﬁned over QðzÞ: Cabc ðN; zÞ carries

an action of the group mN of Nth roots of unity, which is afﬁnely given by z ðx; yÞ

:¼ ðx; z1 yÞ: This action induces an action of mN on the vector space of differential 1-

forms on Cabc ðN; zÞ: It is a classical fact that a basis of regular differential 1-forms on

Xabc ðN; zÞ is given by the regular pull-backs p on of the eigenforms

The explicit construction of Xabc ðN; zÞ by local blow-ups and the expressions of p

in local coordinates were carried out in [4, 1.4]. With them, the genus was calculated

to be [4, 1.5.3]

nA þ ðN; AÞ

b0 X 1

N

nB þ ðN; BÞ

b1 X 1

N

ð5Þ

nC þ ðN; CÞ

b2 X 1

N

nðA þ B þ CÞ ðN; N A B CÞ

b0 þ b1 þ b2 p 1:

N

The vector space GðXabc ðN; zÞ; O1 Þ of regular differential 1-forms splits into a sum

"0pnoN Vn of subrepresentations for the action of mN : If ðn; NÞ ¼ 1; the dimension

of the isotypical component Vn of character wn : z/zn is equal to

nA nB nC nðA þ B þ CÞ

dn :¼ þ þ ð6Þ

N N N N

dim Vn þ dim Vn ¼ 2:

Remark 1. If djN; let cd : Cabc ðN; zÞ-Cabc ðd; zÞ denote the morphism given by

N

ðx; yÞ/ðx; y d Þ: We have

\

" Vn ¼ p
ðker Gcd Þ:

ðn;NÞ¼1 djN

0pnoN 0pdoN

ARTICLE IN PRESS

250 N. Archinard / Journal of Number Theory 101 (2003) 244–269

In view of Remark 1, the elements of Vn with ðn; NÞ ¼ 1 are said to be new. The

vector subspace GðXabc ðN; zÞ; O1 Þnew :¼ "ðn;NÞ¼1 Vn deﬁnes an abelian subvariety

Tabc ðzÞ of the Jacobian varietyRGðXabc ðN; zÞ; O1 Þ
=iðH1 ðXabc ðN; zÞ; ZÞÞ; where i is the

injection given by ½g/½o/ g o: The dimension of Tabc ðzÞ equals

1 X

dimðGðXabc ðN; zÞ; O1 Þnew Þ ¼ ðdim Vn þ dim Vn Þ ¼ jðNÞ:

2 ðn;NÞ¼1

% when zAQ:

Note that Tabc ðzÞ is deﬁned over QðzÞ and in particular over Q % Its endo-

2pi

morphism algebra contains QðzN Þ; where zN :¼ e N : All these results are proved in [4].

Remark 2. The idea of this construction is due to Wolfart in [28], where the dimension is

given (see also [8]). Abelian varieties associated to Appell–Lauricella hypergeometric series

in r variables are deﬁned by intersection of kernels (cf. Remark 1) in [9] for r ¼ 2 and in

[21] for any rX1: The Beta-function corresponds to the case r ¼ 0: Gross and Rohrlich

[14] deﬁned abelian varieties associated to it as suitable quotients of the Jacobian.

Remark 3. The construction of this section was generalized in [4, 1.4] (see also [3])

for Appell–Lauricella hypergeometric series F ða; b1 ; y; br ; c; l1 ; y; lr Þ in r variables.

In this case, the new part of the Jacobian has dimension ðrþ1ÞjðNÞ 2 and its

endomorphism algebraP contains Qðz N Þ: N denotes here the least common

denominator of c rj¼1 b j ; 1 þ a c; b1 ; y; br :

Remark 4. The abelian variety associated to Bða; bÞ has dimension jðMÞ2 and its

algebra of endomorphisms contains QðzM Þ; where M ¼ lcdða; bÞ (e.g. [4, 1.6]).

periods. For explicit computations, it is useful to know this lattice explicitly (up to

isogeny). The dimension of Tabc ðzÞ being jðNÞ; its ﬁrst homology group has rank

2jðNÞ over Z: We know that the double contour loops g01 and g1 lift to Xabc ðN; zÞ;

zN

thus they lift also to Tabc ðzÞ: The lifts of g01 (respectively those of g1 ) are permuted

zN

under Z½zN and the lifts of g01 are independent over Z½zN to those of g1 : Hence,

zN

the lifts of g01 and g1 together generate a Z½zN -module of rank 2 having ﬁnite index

zN

in H1 ðTabc ðzÞ; ZÞ: Let sn be the automorphism of Z½zN deﬁned by zN /znN and set

8 9

> 0 1 >

>

> Z Z >

>

< =

B C

Labc ðzÞ :¼ @sn ðuÞ on þ sn ðvÞ on A ; u; vAZ½zN ;

>

> g01 g1 >

>

>

: N on basis el: >

;

z

of Vn ; ðn;NÞ¼1

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 251

3. Identities

equation of the universal family of elliptic curves reads [17, II, 3]

d 2 O 1 dO 31J 4

þ þ O ¼ 0: ð7Þ

dJ 2 J dJ 144J 2 ð1 JÞ2

The solutions to (7) are those to the hypergeometric differential equation Equð11 11 4

12; 12; 3Þ

1 3

multiplied by J 6 ð1 JÞ4 : The monodromy group of (7) has type ð2; 3; NÞ: By

Kummer’s method (e.g. [15, Section 2.1.3]), one can compute all solutions to (7)

having the form J a ð1 JÞb F ða; b; c; nðJÞÞ; where nðJÞ is one of the transformations

J J1

fJ; J1; 1 J; 1J

1

; J1 ; J g preserving the set of singular points f0; 1; Ng: We order

below the solutions to (7) by groups of solutions equal to OP;u ; where P is a singular

point and u the characteristic exponent of Equð11 11 4

12; 12; 3Þ at P:

8

1 3

>

> F ð11 11 4

> J 6 ð1 JÞ4

> 12; 12; 3; JÞ

<

1 1

O0;0 ðJÞ ¼ J 6 ð1 JÞ4 5 5 4

F ð12 ; 12; 3; JÞ

>

>

>

> 1 1

: 6

J ð1 JÞ6 F ð12 5 11 4 J

; 12; 3; J1Þ

8

1 3

>

> J 6 ð1 JÞ4 F ð12 7 7 2

; 12; 3; JÞ

>

>

<

1 1

O0;1c ðJÞ ¼ J 6 ð1 JÞ4 F ð 1 ; 1 ; 2; JÞ

>

> 12 12 3

>

> 1 1

: 1 7 2 J

J 6 ð1 JÞ6 F ð12 ; 12; 3; J1Þ

8

1 3

>

> J 6 ð1 JÞ4 F ð11 11 3

>

>

< 12; 12; 2; 1 JÞ

1 3

O1;0 ðJÞ ¼ J 6 ð1 JÞ4 F ð 7 ; 7 ; 3; 1 JÞ

>

> 12 12 2

ð8Þ

>

> 3

: 34 7 11 3 J1

J ð1 JÞ4 F ð12 ; 12; 2; J Þ

8

1 1

>

> J 6 ð1 JÞ4 5 5 1

F ð12 ; 12; 2; 1 JÞ

>

>

<

1 1

O1;cab ðJÞ ¼ J 6 ð1 JÞ4 F ð 1 ; 1 ; 1; 1 JÞ

>

> 12 12 2

>

> 1

: 14 1 5 1 J1

J ð1 JÞ4 F ð12 ; 12; 2; J Þ

8 1 1

>

> J 4 ð1 JÞ4 F ð12 1 5

; 12; 1; J1Þ

>

>

>

>

> 3

< J 34 ð1 JÞ4 F ð 7 ; 11; 1; 1Þ

12 12 J

ON;a ðJÞ ¼ ON;b ðJÞ ¼

>

>

1 1

1 7 1

>

> J 6 ð1 JÞ6 F ð12 ; 12; 1; 1J Þ

>

>

>

: 1 1

J 6 ð1 JÞ6 F ð12 5 11 1

; 12; 1; 1J Þ:

ARTICLE IN PRESS

252 N. Archinard / Journal of Number Theory 101 (2003) 244–269

independent periods on the universal elliptic curve

27J

EJ : y2 ¼ 4x3 ðx þ 1Þ;

J 1

where J is the modular invariant normalized by JðiÞ ¼ 1: Since the parameter space

H=SL2 ðZÞ of isomorphy classes of elliptic curves over C is isomorphic to C under J;

for each tAH such that JðtÞ ¼ J; there is an isomorphism jt : Et -EJ making the

following diagram commute:

where Et : ỹ2 ¼ 4x̃3 g2 ðtÞx̃ g3 ðtÞ; gk being the Eisenstein series of weight 2k and

YL the Weierstrass Y-function associated to a lattice L: For such a t; there exists

mðtÞAC such that LJ ¼ mðtÞLt : Checking that jt is given by ðx; yÞ ¼

3

ðgg23 ðtÞ g2 ðtÞ 2

ðtÞ x̃; ðg3 ðtÞÞ ỹÞ; we ﬁnd

sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

g3 ðtÞ 1 1 1 1

mðtÞ ¼ ¼ JðtÞ6 ð1 JðtÞÞ4 DðtÞ12 ;

g2 ðtÞ 27

where D ¼ g32 27g23 is the discriminant function. The functions mðtÞ and tmðtÞ are

C-linearly independent and multivalued functions of J: Thus, their branches form

local systems of solutions to (7). Hence, for each solution to (7) of the form J a ð1

JÞb F ða; b; c; nðJÞÞ; there exist A; BAC such that

1 1 1

F ða; b; c; nðJÞÞ ¼ ðAt þ BÞJðtÞ6a ð1 JðtÞÞ4b DðtÞ12 : ð9Þ

deﬁne CnðJÞ;t to be the connected component of ftAH; jnðJðtÞÞjo1g which contains

t0 : Such a domain is simply connected and invariant under the action by fractional

transformation of the elements of SL2 ðZÞ which ﬁx t0 : In the following charts, gt0

will denote such a modular transformation. On such a domain, the constants and the

branches of the roots can therefore be determined. In view of the relation D ¼

1

ð2pÞ12 Z24 ; where Z is the Dedekind Eta function, D12 can be replaced on CnðJÞ;t by Z2

up to a complex constant.

3.1. Solutions at J ¼ 1

For these solutions, we choose t0 ¼ i: In the cases with c ¼ 32; we have 14 b ¼ 12: The

left-hand side of (9) being holomorphic at t0 ¼ i; the right-hand side must also be so. As J

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 253

1

takes value 1 at i with multiplicity 2, ð1 JðtÞÞ2 has a pole of order 1 at i: Consequently,

At þ B must have a zero of order 1 at i: This means that B ¼ iA must hold.

In the cases with c ¼ 12; a relation between the constants can be found by applying

the modular transformation t/ 1t to the identity. The invariance of the domain

and of the function J under this transformation together with the property Zð1tÞ ¼

pﬃﬃﬃﬃﬃﬃﬃﬃ

itZðtÞ imply the relation B ¼ iA:

In each case, the value of the constant A is then determined by inserting t ¼ i in

the identity. The branches of the roots are real positive for real positive arguments.

We obtain the following results.

c ¼ 12

For tAC1J;i and tACJ1 respectively, we have

J ;i

1 1 1 t þ i ZðtÞ 2

F ; ; ; 1 JðtÞ ¼

12 12 2 2i ZðiÞ

5 5 1 tþi 1 ZðtÞ 2

F ; ; ; 1 JðtÞ ¼ JðtÞ3 ð10Þ

12 12 2 2i ZðiÞ

1 5 1 JðtÞ 1 tþi 1 ZðtÞ 2

F ; ; ; ¼ JðtÞ 12 :

12 12 2 JðtÞ 2i ZðiÞ

1 5 1

Note 1. The identity for F ð12 ; 12; 2; zÞ was known by Beukers and Wolfart [6].

c ¼ 32

For tAC1J;i and tACJ1 respectively, we have

J ;i

7 7 3 1 ZðtÞ 2 1

F ; ; ; 1 JðtÞ ¼ resi ðð1 JÞ2 Þ1 ðt iÞð1 JðtÞÞ2

12 12 2 ZðiÞ

2

11 11 3 1

2 1

1

2 3 ZðtÞ

1

F ; ; ; 1 JðtÞ ¼ resi ðð1 JÞ Þ ðt iÞð1 JðtÞÞ JðtÞ ð11Þ

12 12 2 ZðiÞ

7 11 3 JðtÞ 1 1 1 7 ZðtÞ 2

2 1 2

F ; ; ; ¼ resi ðð1 JÞ Þ ðt iÞð1 JðtÞÞ JðtÞ12 :

12 12 2 JðtÞ ZðiÞ

ARTICLE IN PRESS

254 N. Archinard / Journal of Number Theory 101 (2003) 244–269

3.2. Solutions at J ¼ 0

2pi

Here, we work out identities in the neighbourhoods of r :¼ e 3 and resp. r: % It

goes very similarly as for the solutions at J ¼ 1:

The cases with c ¼ 43 correspond to a ¼ 16: Since the function J takes value 0 with

1

multiplicity 3 at any point congruent to r modulo SL2 ðZÞ; J 3 has a pole of order 1

at any such point. It follows the relations B ¼ rA at t0 ¼ r and B ¼ rA % at t0 ¼

%

r:

In the cases with c ¼ 23; we apply a modular transformation gt0 ; which ﬁxes t0 : This

transformation together with the obtained relation are given in the following table.

The constant A is then determined by setting t ¼ t0 in the identity. The branches of

the roots are real positive for real positive arguments.

The results, which are summarized in the above table, imply the following

identities.

c ¼ 23

8

pﬃﬃr% ð

< t ZðtÞ 2

Þ if tACJ;r

1 1 2 i 3 ZðrÞ

F ; ; ; JðtÞ ¼

12 12 3 : tþr

pﬃﬃ ð

ZðtÞ 2

Þ if tACJ;r%

i 3 ZðrÞ %

8

1

> pﬃﬃr% ð1 JðtÞÞ2 ð

< t ZðtÞ 2

11 11 4 i 3 ZðrÞÞ if tACJ;r

F ; ; ; JðtÞ ¼ ð12Þ

12 12 3 >

: tþr

1

ZðtÞ 2

pﬃﬃ ð1 JðtÞÞ 2 ð Þ if tACJ;r%

i 3 %

ZðrÞ

8 1

>

> tr% ZðtÞ 2

1 7 2 JðtÞ

< ipﬃﬃ3 ð1 JðtÞÞ12 ðZðrÞÞ if tAC J

J1;r

F ; ; ; ¼

12 12 3 JðtÞ 1 >

>

1

ZðtÞ 2

: tþr

pﬃﬃ ð1 JðtÞÞ12 ð

%

ZðrÞ Þ if tAC J :

J1;r%

i 3

1 7 2

Note 2. The identity for F ð12 ; 12; 3; zÞ on C J was already known by Beukers and

J1;r

Wolfart [6].

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 255

c ¼ 43

8

1 1

>

< resr ðJ 3 Þ1 ðt rÞJðtÞ3 ð ZðtÞ Þ2 if tACJ;r

5 5 4 ZðrÞ

F ; ; ; JðtÞ ¼

12 12 3 >

: resr ðJ 13 Þ1 ðt þ rÞJðtÞ

ZðtÞ 2

1

3

% %%

ZðrÞ ð

Þ if tACJ;r%

11 11 4

F ; ; ; JðtÞ

12 12 3

8

1 1 1

>

< resr ðJ 3 Þ1 ðt rÞJðtÞ3 ð1 JðtÞÞ2 ð ZðtÞ Þ2 if tACJ;r

ZðrÞ

¼ ð13Þ

> 1

: resr ðJ 3 Þ1 ðt þ rÞJðtÞ3

1 1

% % ð1 JðtÞÞ2 ðZðZðtÞ 2

%

rÞ Þ if tACJ;r%

5 11 4 JðtÞ

F ; ; ;

12 12 3 JðtÞ 1

8

1 1 5

>

> 1 ZðtÞ 2

< resr ðJ 3 Þ ðt rÞJðtÞ 3 ð1 JðtÞÞ12 ðZðrÞÞ if tAC J ;r

J1

¼

>

>

1 1

1 5

ZðtÞ 2

%

: resr% ðJ 3 Þ ðt þ rÞJðtÞ 3 ð1 JðtÞÞ12 ð

%

ZðrÞ Þ if tAC J ;r :

% J1

3.3. Solutions at J ¼ N

1

The invariance under t/t þ 1 of J and Z implies A ¼ 0: The function J 6a

1

ð1 JÞ4b Z2 has a pole of order 16 a þ 14 b 12

1

¼ 0 at inﬁnity with constant

1 1

term equal to 172812 ; because a þ b ¼ 0 in each case. This implies B ¼ 172812 ¼

1

124 : We summarize these informations in a table and give the consequent identities.

c¼1

1 5 1 pﬃﬃﬃﬃﬃ 1

8tAC1 : F ; ; 1; ¼ 4 12JðtÞ12 ZðtÞ2

J ;N 12 12 JðtÞ

ð14Þ

7 11 1 pﬃﬃﬃﬃﬃ 7 1

F ; ; 1; ¼ 4 12JðtÞ12 ð1 JðtÞÞ2 ZðtÞ2

12 12 JðtÞ

1 7 1 pﬃﬃﬃﬃﬃ 1

8tAC 1 : F ; ; 1; ¼ 4 12ð1 JðtÞÞ12 ZðtÞ2

1J ;N 12 12 1 JðtÞ

ð15Þ

5 11 1 pﬃﬃﬃﬃﬃ 1 5

F ; ; 1; ¼ 4 12JðtÞ3 ð1 JðtÞÞ12 ZðtÞ2 :

12 12 1 JðtÞ

ARTICLE IN PRESS

256 N. Archinard / Journal of Number Theory 101 (2003) 244–269

Remark 5. We curiously note that in the identities with nðJÞ ¼ J1 resp. 1J

1

; the

exponent of JðtÞ resp. 1 JðtÞ is equal to the smallest parameter of the series.

(Thanks to R. Pink for this observation.)

Remark 6. Similar identities can be worked out when the hypergeometric differential

equation is the Picard–Fuchs differential equation of some family of elliptic

curves. We give an example for a hypergeometric series whose monodromy group is

of type ðN; N; NÞ; i.e. is isomorphic to the principal congruence subgroup Gð2Þ of

level 2. For t in the connected component of ftAH; jlðtÞjo1g which contains N;

we have

1 1 1 1 1 1

F ; ; 1; lðtÞ ¼ 22 34 pg2 ðtÞ4 ðl2 ðtÞ lðtÞ þ 1Þ4 :

2 2

classicaly known (see for example [23]). Local relations follow also directly from

identities (14) and (15). Indeed, we have

1 5 1 4

8tAC1 : F ; ; 1; ¼ E2 ðtÞ and

J ;N 12 12 JðtÞ

6

1 7 1

8tAC 1 :F ; ; 1; ¼ E3 ðtÞ;

1J ;N 12 12 1 JðtÞ

This section gives some consequences of the identities proved in Section 3. Of most

importance for the determination of the exceptional sets is Section 4.1, which

determines, in each case with co1; an inﬁnite subset of the exceptional set. In

Section 4.2 some values of JðmiÞ and ðZðmiÞ 2

ZðiÞ Þ and some algebraic evaluations of

% are calculated. These evaluations provide examples for

F ð 1 ; 5 ; 1; zÞ at some zAQ

12 12 2

Kroneckers Jugendtraum. Section 4.3 gives some relations between values of

transcendental functions (Proposition 4.5) and also some transcendental evaluations

7 11 3

of F ð12; 12; 2; zÞ at algebraic points.

For the determination of inﬁnite subsets of the exceptional sets, we will use the two

following classical facts on modular functions.

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 257

AtþB

Zð

D Þ2 ; then the Þ

(F1) Let A; B; DAZ with AD40 and for tAH; set jðtÞ ¼ ð ZðtÞ

function j is integral over Z½1728J (see [18, 12.2, Theorem 2]).

% (see [18, 5.2, Theorem 4]).

(F2) If tAH is imaginary quadratic, then JðtÞAQ

c ¼ 12

Let nAf1 J; J1 J g: If tAQðiÞ-fwAH; jnðJÞðwÞjo1g; then by (F2), we have

nðJÞðtÞAQ:% Moreover, t can be written as AiþB with AD40 because tAQðiÞ-H: By

D

% By the identities (10), we have

(F2), this implies ðZðtÞÞ2 AQ:

ZðiÞ

% and

Lemma 4.1. For tAQðiÞ-fwAH; j1 JðwÞjo1g; we have 1 JðtÞAQ

1 1 1 % 5 5 1 %

F ; ; ; 1 JðtÞ AQ and F ; ; ; 1 JðtÞ AQ:

12 12 2 12 12 2

JðtÞ1 %

Lemma 4.2. For tAQðiÞ-fwAH; jJðwÞ1

JðwÞ jo1g; we have JðtÞ AQ and

1 5 1 JðtÞ 1 %

F ; ; ; AQ:

12 12 2 JðtÞ

Proof. By the above explanations, it is clear that these results hold for each t in

QðiÞ-CnðJÞ;i : Further, for any t0 congruent to i mod SL2 ðZÞ; there exists an

identity like (9) valid in the neighbourhood of t0 : This identity will defer from

that of (10) only up to an algebraic factor which comes out by the transformation

properties of the Z-function. We conclude by applying the same reasoning to that

identity. &

c ¼ 23

pﬃﬃﬃ

It goes very similarly as above. If tAQði 3Þ-fwAH; jnðJÞðwÞjo1g;

then nðJÞðtÞAQ % by (F1) and (F2). The algebraicity at the

% and ð ZðtÞ Þ2 AQ

ZðrÞ

pﬃﬃﬃ

points of tAQði 3Þ-ðCnðJÞ;r ,CnðJÞ;r% Þ follows then directly from the identities

(12). The algebraicity of the values at the other points is obtained as in the

above proof.

ARTICLE IN PRESS

258 N. Archinard / Journal of Number Theory 101 (2003) 244–269

pﬃﬃﬃ

% and

Lemma 4.3. For tAQði 3Þ-fwAH; jJðwÞjo1g; we have JðtÞAQ

1 1 2 % 7 7 2 %

F ; ; ; JðtÞ AQ and F ; ; ; JðtÞ AQ:

12 12 3 12 12 3

pﬃﬃﬃ JðwÞ

Lemma 4.4. For tAQði 3Þ-fwAH; jJðwÞ1 jo1g; we have JðtÞ % and

JðtÞ1AQ

1 7 2 JðtÞ %

F ; ; ; AQ:

12 12 3 JðtÞ 1

1 5 1

4.2. Some algebraic evaluations of F ð12 ; 12; 2; zÞ

ZðiÞ Þ for some m: These values

were computed in [4, 2.6.2] with the help of Mathematica in using some classical

results on modular equations (to be found in [25]).

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 259

1 5 1

The following algebraic evaluations of F ð12; 12; 2; zÞ were calculated by insert-

ing the values given in the previous charts into the identity for this series given

in (10).

For t ¼ 2i we get,

1

1 5 1 33 72 3 114

F ; ; ; ¼

12 12 2 113 22

t ¼ 3i;

pﬃﬃﬃ !

1 5 1 23 72 ð22 11093 3 19 31 3Þ 2 pﬃﬃﬃ 1

F ; ; ; 3

¼ ð3 7 þ 22 5 3Þ4

12 12 2 ð11 23Þ 3

t ¼ 4i;

pﬃﬃﬃ !

1 5 1 33 72 112 ð83967 24 3 19 59 2Þ 5 pﬃﬃﬃ 1

F ; ; ; 3

¼ 3 ð7 13 þ 22 3 5 2Þ4

12 12 2 ð23 47Þ 2

ARTICLE IN PRESS

260 N. Archinard / Journal of Number Theory 101 (2003) 244–269

t ¼ 5i;

pﬃﬃﬃ !

1 5 1 24 33 72 ð23 5 5237 22067 3 23 47 83 5Þ

F ; ; ;

12 12 2 ð11 59 71Þ3

3 pﬃﬃﬃ 1

¼ ð7 23 þ 23 3 5 5Þ4

5

t ¼ 6i;

pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

pﬃﬃﬃ pﬃﬃﬃ !

pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

1 5 1 ð2 þ ð2 þ 3 þ 9 þ 6 3Þ8 Þ3 27ð2 þ 3 þ 9 þ 6 3Þ8

F ; ; ; pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

pﬃﬃﬃ

12 12 2 ð2 þ ð2 þ 3 þ 9 þ 6 3Þ8 Þ3

0 sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

pﬃﬃﬃ14

1

7 p ﬃﬃ

ﬃ 7 3A

¼ 2 @3 7 11 þ 22 5 7 3 þ 23 3 5 2 3 þ

2 3 2

t ¼ 7i;

pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

pﬃﬃﬃ pﬃﬃﬃ !

pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

1 5 1 ðð2 þ 7 þ 7 þ 4 7Þ12 210 Þ3 218 33 ð2 þ 7 þ 7 þ 4 7Þ12

F ; ; ; pﬃﬃﬃ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

pﬃﬃﬃ

12 12 2 ðð2 þ 7 þ 7 þ 4 7Þ12 210 Þ3

sﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ!1

4

22 p ﬃﬃ

ﬃ 8

7 43 þ 23 3 5 7 þ 22 3 5 7 3 þ pﬃﬃﬃ

7 7

t ¼ 8i;

! 7 1

1 5 1 ð27 U þ 1Þ3 25 33 U 34 ð27 U þ 1Þ4

F ; ; ; ¼ 95 pﬃﬃﬃ 1 1 ; where

12 12 2 ð27 U þ 1Þ3 224 ð1 þ 2Þ2 U 6

pﬃﬃﬃ

U ¼ 24 257 2441 þ 22 32 11 17923 2

qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

pﬃﬃﬃ

þ 32 22 3 221047 937823 þ 11 17 139 67672987 2

! pﬃﬃﬃ 24 1

1 5 1 ðS24 þ 231 Þ3 260 33 S24 11ð1 þ 5ÞðS þ 231 Þ4

F ; ; ; ¼ ; where

12 12 2 ðS24 þ 231 Þ3 21 1

2 4 34 5S4

qﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

pﬃﬃﬃ pﬃﬃﬃ

S :¼ 1 þ 5 þ 10 þ 6 5:

Remark 8. The evaluation at t ¼ 2i and 3i were given by Beukers and Wolfart [6]

and Flach [13], respectively. Joyce and Zucker [16] produced the evaluations at

t ¼ 4i and 5i by a different method. The evaluations at t ¼ 6i; 7i; 8i; 10i are new.

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 261

7 11 3

F ð12; 12; 2; zÞ

The comparison between the work [16] of Joyce and Zucker and some of our

results provides some amusing relations.

Proposition 4.5. Let J be the modular invariant, G be the classical Gamma function

and Z be the Dedekind Eta-function. Then we have

1 23 ip2

resi ðð1 JÞ2 Þ ¼ pﬃﬃﬃ 1 4 ð16Þ

3Gð4Þ

and

Gð14Þ2

ZðiÞ2 ¼ 3

: ð17Þ

22 p2

7 4

7 11 3 1323 ð11Þ4 1 ð18Þ

F ; ; ; ¼ G

12 12 2 1331 336p2 4

3

1 5 1 JðtÞ 1 2p2 1

F ; ; ; ¼ ðt þ iÞJðtÞ12 ZðtÞ2 : ð19Þ

12 12 2 JðtÞ iGð14Þ2

Jð2iÞ1 7 11 3 JðtÞ1

We use 13231331 ¼ Jð2iÞ to compare (18) to our identity (11) for F ð12; 12; 2; JðtÞ Þ

evaluated at t ¼ 2i: This yields relation (16). The comparison of (19) to the identity

1 5 1

for F ð12; 12; 2; zÞ given in (10) gives relation (17). &

Note 3. We thank the referee for noting that relation (16) could be obtained in a

standard way along modular forms. Indeed, a direct calculation shows that the

1

residue of ð1 JÞ2 at i is the quotient of a period by a quasi-period on EJ with

J ¼ 1: In particular, this is a transcendental number by Wüstholz’s result (W2)

quoted in Section 5. Further, this corresponds via (16) to the algebraic independence

of p and Gð14Þ proved by Chudnovsky.

Using relation (16) together with the values of JðmiÞ and ðZðmiÞ 2

ZðiÞ Þ calculated in

7 11 3

Section 4.2, we compute as examples the evaluations of F ð12; 12; 2; zÞ at t ¼ 3i; 4i; 5i:

We obtain, for t ¼ 3i;

ARTICLE IN PRESS

262 N. Archinard / Journal of Number Theory 101 (2003) 244–269

pﬃﬃﬃ !

7 11 3 23 72 ð22 11093 3 19 31 3Þ

F ; ; ;

12 12 2 ð11 23Þ3

! pﬃﬃﬃ 1

Gð14Þ4 ð3 47 361359611 þ 22 73 401 77899 3Þ4

¼ p ﬃﬃ

ﬃ

p2 23 7 11 6

t ¼ 4i;

pﬃﬃﬃ !

7 11 3 33 72 112 ð83967 24 3 19 59 2Þ

F ; ; ;

12 12 2 ð23 47Þ3

! pﬃﬃﬃ 1

Gð14Þ4 ð733 3074274799 þ 22 3 11 99746711 2Þ4

¼

p2 25 7 11

t ¼ 5i;

pﬃﬃﬃ ! !

7 11 3 24 33 72 ð23 5 5237 22067 3 23 47 83 5Þ Gð14Þ4

F ; ; ; ¼

12 12 2 ð11 59 71Þ3 p2

pﬃﬃﬃ 1

ð23 43 147600623 4715541403 þ 23 32 73 190829 108837819119 5Þ4

5

:

23 3 54 7 11 19

For the triples ða; b; cÞ considered in Section 3 for which ca1; we give a necessary

condition for zAC to lie in Eða; b; cÞ: Here will be used the abelian varieties

constructed in Section 2 and the mapping properties of the quotient of two linearly

independent solutions to Equða; b; cÞ given in Section 1. The key tool to get the

condition on z is the following consequence of Wüstholz’s Analytic Subgroup

Theorem [30, Hauptsatz].

% For each j ¼ 1; y; m; set n j ¼ dim A j and

simple abelian varieties defined over Q:

ð jÞ ð jÞ ð jÞ ð jÞ

choose 0ag j AH1 ðA j ; ZÞ: Let o1 ; y; on j and Z1 ; y; Zn j be bases of differential

forms on A j of first and second kind, respectively. Then the vector space generated over

Q% by the numbers

Z Z

ð jÞ ð jÞ

ok ; Zk ; 1XjXm; 1XkXn j ;

gj gj

Pm

has dimension 2 j¼1 n j:

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 263

R

% For i ¼ 1; 2; let oi be a

(W1) Let A1 ; A2 be two abelian varieties deﬁned over Q: gi

R

o1

period of the ﬁrst kind on Ai : If Rg1 AQ; % then there exist simple abelian

o2

g2

subvarieties A0i

of Ai ; i ¼ 1; 2; such that A01 is isogenous to A02 :

R

(W2) Let A1 ; A2 be two abelian varieties deﬁned over Q: % Let

g1 o1 be a period of

R

ﬁrst kind on A1 and g o2 be a period of second kind on A2 ; then

2

R

o1

Rg1 eQ: %

g o2

2

The following table gives the denominators of the integral representation of the

considered hypergeometric series in dependence of the value of c:

ð20Þ

Conditions (5) of Section 2 for the regularity of differential forms show that the

numerators of the hypergeometric series with c ¼ 43 and 32 are of the ﬁrst kind when

za0: One can verify that the denominators are of second kind (see the holomorphy

conditions in [4, 1.6.3]). Hence, (W2) implies the transcendence of the quotients for

which za0 and we have

7 7 3 11 11 3 7 11 3

E ; ; ¼E ; ; ¼E ; ; ¼ f0g ð21Þ

12 12 2 12 12 2 12 12 2

5 5 4 11 11 4 5 11 4

E ; ; ¼E ; ; ¼E ; ; ¼ f0g: ð22Þ

12 12 3 12 12 3 12 12 3

Note 4. These are in fact special issues of a more general result. Indeed, if a; b; cAQ

satisfy a þ 14c4b40; ao1 and c41; then Eða; b; cÞ ¼ f0g: The conditions on

a; b; c ensure the period in the numerator to be of the ﬁrst kind, whereas the

denominator be of second kind. Hence, the result follows from (W2). We wish to

thank the referee for pointing this out.

Remark 9. In the cases with c ¼ 1; the denominator has a nonzero residue, hence is

of the third kind. We do not know what the nature of the exceptional set is in that

case.

For each triple ða; b; cÞ with co1; we have N ¼ 12 and the dimension formula (6)

shows that all the Vn with ðn; NÞ ¼ 1 are 1-dimensional. Let Vn be generated by on :

ARTICLE IN PRESS

264 N. Archinard / Journal of Number Theory 101 (2003) 244–269

The construction in Section 2 associates to each ða; b; cÞ and each zAP1C f0; 1; Ng

an abelian variety Tabc ðzÞ of dimension jð12Þ ¼ 4 which is isogenous to a quotient

C4 =Labc ðzÞ; where

80 1 9

>

< Z Z >

=

B C

Labc ðzÞ :¼ @sn ðuÞ on þ sn ðvÞ on A ; u; vAZ½z12 :

>

: g01 g1 >

;

N

z n¼1;5;7;11

R R

Since g01 on and on are linearly independent solutions to Equða; b; cÞ (see [15,

g1

zN R R

2.4.5]), their quotient Dn ðzÞ :¼ g on = g on is a Schwarz triangle function. One

1 01

zN

veriﬁes that D1 ¼ Dabc : Hence, D :¼ D1 maps the upper half plane into the interior of

a hyperbolic triangle with angles p2; p3; N: The vertices can be calculated thanks to the

results of Section 1. The triangle depends only on the transformation nðJÞ for which

J a ð1 JÞb F ða; b; c; nðJÞÞ is solution in (7). The following chart gives the vertices and

the corresponding angles of the triangle in dependence of nðJÞ:

ð23Þ

Each transformation nðJÞ maps either the left-hand side L :¼ ftAH; jtj41;

12oReðtÞo0g or the right-hand side R :¼ ftAH; jtj41; 0oReðtÞo12g of the

standard fundamental domain for the action of SL2 ðZÞ on H by fractional linear

transformations. Moreover, the composition D3nðJÞ is a conformal mapping of L

resp. R to the interior DðHÞ of the hyperbolic triangle. By the uniqueness part

of the Riemann mapping theorem, there exists a unique inverse conformal mapping

m to D3nðJÞ:

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 265

The following table gives the map m for each nðJÞ and some relations between the

Schwarz triangle functions Dn :

Remark 10. Using the multiplication theorem [26, 12 15] for the Gamma function

1 n1 1 1

GðzÞG z þ y G zþ ¼ ð2pÞ2ðn1Þ n2nz GðnzÞ;

n n

1 5 1 1 1

B ; ¼2 34 B ; ;

12 12 2 4

1 7 5 1 1

B ; ¼ 26 B ; :

12 12 2 6

By table (20), the above numbers are the denominators of the hypergeometric

series with co1: Since we care about these denominators only up to algebraic

factors, we can consider the period Bð12; 14Þ when c ¼ 12 and Bð12; 16Þ when c ¼ 23:

c ¼ 12

By Remark 4, Section 2, Bð12; 14Þ is a period on an abelian variety TL;0 of dimension

1 (hence simple) whose endormorphism ring contains Z½z4 : Thus, TL;0 is isogenous

to the torus C=LL;0 ; where LL;0 ¼ Z"tZ with tAQðiÞ: On the other hand, Tabc ðzÞ ¼:

TL;z is isogenous to the quotient of C4 by the lattice

% then TL;0 and TL;z are deﬁned over Q %

2

and (W1) implies the existence of an isogeny of TL ð0Þ to a simple factor of TL ðzÞ: In

%

particular, there exits a Q-linear homomorphism F : C-C4 such that FðLL;0 ÞCLL;z :

Writing Fð1Þ and FðtÞ as elements of LL;z ; the condition FðtÞ ¼ tFð1Þ implies

ARTICLE IN PRESS

266 N. Archinard / Journal of Number Theory 101 (2003) 244–269

ðm3DÞðzÞAQðiÞ: Using z ¼ ðnðJÞ3m3DÞðzÞ; we get

% 1 1 1 % ) (tAQðiÞ-H s:t: z ¼ ð1 JÞðtÞ

zAQ; F ; ; ; z AQ

12 12 2

% 5 5 1 % ) (tAQðiÞ-H s:t: z ¼ ð1 JÞðtÞ

zAQ; F ; ; ; z AQ

12 12 2

% 1 5 1 % J 1

zAQ; F ; ; ; z AQ ) (tAQðiÞ-H s:t: z ¼ ðtÞ:

12 12 2 J

c ¼ 23

It goes very similarly as for the previous case. Bð12; 16Þ is a period on an abelian

variety TR;0 of dimension 1 whose endomorphism ring contains Z½z6 : Hence, TR;0 is

pﬃﬃﬃ

isogenous to the torus C=LR;0 ; where LR;0 ¼ Z"tZ with tAQði 3Þ: TR;z is

isogenous to the quotient of C4 by the lattice

% then TR;0 and TR;z are deﬁned over Q % and (W1)

3

implies the existence of an isogeny of TR ð0Þ to a simple factor of TR ðzÞ: As above,

pﬃﬃﬃ

this implies a condition on DðzÞ; which is here DðzÞAQði 3Þ: By the previous chart,

pﬃﬃﬃ

the maps m are here rational functions with coefﬁcients in Qði 3Þ; hence we have

pﬃﬃﬃ

ðm3DÞðzÞAQði 3Þ: The function nðJÞ3m3D being the identity on H; we have shown

1 1 2 pﬃﬃﬃ

%

zAQ; F % ) (tAQði 3Þ-H

; ; ; z AQ s:t: z ¼ JðtÞ

12 12 3

7 7 2 pﬃﬃﬃ

%

zAQ; F % ) (tAQði 3Þ-H

; ; ; z AQ s:t: z ¼ JðtÞ

12 12 3

1 7 2 pﬃﬃﬃ J

%

zAQ; F % ) (tAQði 3Þ-H

; ; ; z AQ s:t: z ¼ ðtÞ:

12 12 3 J 1

Bringing together Lemmata 4.1, 4.2, 4.3, 4.4, 5.2, 5.3 and considering (21), (22), we

can state

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 267

% of the

hypergeometric series F ða; b; c; zÞ with ca1 appearing in Kummer’s solutions (8) to

the Picard–Fuchs differential equation

d 2 O 1 dO 31J 4

þ þ O¼0

dJ 2 J dJ 144J 2 ðJ 1Þ2

of the universal family of elliptic curves (cf. Section 3) are given as follows:

7 7 3 11 11 3 7 11 3

E ; ; ¼E ; ; ¼E ; ; ¼ f0g

12 12 2 12 12 2 12 12 2

5 5 4 11 11 4 5 11 4

E ; ; ¼E ; ; ¼E ; ; ¼ f0g

12 12 3 12 12 3 12 12 3

)

1 1 1

Eð12; 12; 2Þ;

%

¼ fzAQ-Dð0; 1Þ; (tAQðiÞ-H s:t: z ¼ 1 JðtÞg

5 5 1

Eð12; 12; 2Þ

! "

1 JðtÞ 1

E ; 5 1

; ¼ zA %

Q-Dð0; 1Þ; (tAQðiÞ-H s:t: z ¼

12 12 2 JðtÞ

)

1 1 2

Eð12; 12; 3Þ pﬃﬃﬃ

%

¼ fzAQ-Dð0; 1Þ; (tAQði 3Þ-H s:t: z ¼ JðtÞg

7 7 2

Eð12; 12; 3Þ

! "

1 7 2 pﬃﬃﬃ JðtÞ

%

E 12; 12; 3 ¼ zAQ-Dð0; 1Þ; (tAQði 3Þ-H s:t: z ¼ ;

JðtÞ 1

1 5 1 1 7 2

Note 5. The determinations of Eð12; 12; 2Þ and Eð12 ; 12; 3Þ were known by Beukers and

Wolfart [6].

Remark 11. The nontrivial exceptional sets appearing in Theorem 5.4 are dense in

%

Q-Dð0; 1Þ; because J induces a continuos bijection H=SL2 ðZÞ-C: In particular,

they are inﬁnite.

Remark 12. It follows from the triviality of the exceptional sets in regard to identities

1

(11) resp. (13) and to the facts (F1) and (F2) of Section 4 that the residue of ð1 JÞ2

1 2pi

resp. J 3 at any point congruent to i resp. e 3 modulo SL2 ðZÞ is a transcendental

number. See also Note 3 in Section 4.3.

ARTICLE IN PRESS

268 N. Archinard / Journal of Number Theory 101 (2003) 244–269

Acknowledgments

This research was funded by a grant from the Swiss National Science Foundation.

The author is grateful to McGill University for its hospitality while this article was

written and wish to thank the referees for their instructive remarks. Many thanks to

Paula B. Cohen for proposing some additional references. Special thanks to Henri

Darmon and the fantastic number theory group in Montreal.

References

[1] M. Abramowitz, I.A. Stegun (Eds.), Handbook of Mathematical Functions, National Bureau of

Standards, USA, 1964.

[2] Y. André, G-fonctions et transcendance, J. Reine Angew. Math. 476 (1996) 95–125.

[3] N. Archinard, Hypergeometric abelian varieties, Can. J. Math., to appear.

[4] N. Archinard, Abelian varieties and identities for hypergeometric series, Ph.D. Thesis, ETH Zürich,

2000, http://e-collection.ethbib.ethz.ch.

[5] F. Beukers, Algebraic values of G-functions, J. Reine Angew. Math. 434 (1993) 45–65.

[6] F. Beukers, J. Wolfart, Algebraic values of hypergeometric functions, in: A. Baker (Ed.), New

Advances in Transcendence Theory, Cambridge University Press, Cambridge, 1986.

[7] P.B. Cohen, Perspectives de l’approximation diophantienne et de la transcendance, Ramanujan

Journal, to appear.

[8] P. Cohen, J. Wolfart, Modular embeddings for some non-arithmetic Fuchsian groups, Acta Arith. 56

(1990) 93–110.

[9] P.B. Cohen, J. Wolfart, Fonctions hypergéométriques en plusieurs variables et espaces des modules

de variétés abéliennes, Ann. Scient. Éc. Norm. Sup. (4) 26 (1993) 665–690.

[10] P.B. Cohen, G. Wüstholz, Applications of the André–Oort Conjecture to transcendence, in:

G. Wüstholz (Ed.), A Panorama in Number Theory, Cambridge University Press, Cambridge, 2001.

[11] P.-A. Desrousseaux, Valeurs exceptionelles de fonctions hypergéométriques d’Appell, Ramanujan

Journal, submitted for publication.

[12] S. Edixhoven, A. Yafaev, Subvarieties of Shimura varieties, Ann. Math., to appear.

[13] M. Flach, Periods and special values of the hypergeometric series, Math. Proc. Camb. Phil. Soc. 106

(1989) 389–401.

[14] H. Gross (with an appendix by D. Rohrlich), On the periods of abelian integrals and a formula of

Chowla and Selberg, Invent. Math. 45 (1978) 193–211.

[15] K. Iwasaki, H. Kimura, S. Shimomura, M. Yoshida, From Gauss to Painlevé, A Modern Theory of

Special Functions, Aspects of Mathematics, Vieweg, Braunschweig, 1991.

[16] G.S. Joyce, I.J. Zucker, Special values of the hypergeometric series, Math. Proc. Camb. Phil. Soc. 109

(1991) 257–261.

[17] F. Klein (ausgearbeitet und vervollständigt von R. Fricke), Theorie der elliptischen Modulfunctionen,

B.G. Teubner, Leipzig, 1890.

[18] S. Lang, Elliptic Functions, Springer, Berlin, 1987.

[19] Z. Nehari, Conformal Mapping, McGraw–Hill Book Company, Inc., New York, 1952.

[20] H.A. Schwarz, Ueber diejenigen Fälle, in welchen die Gaussische hypergeometrische Reihe eine

algebraische Function ihres vierten Elementes darstellt, J. Reine Angew. Math. 75 (1873) 292–335.

[21] H. Shiga, J. Wolfart, Criteria for complex multiplication and transcendence properties of

automorphic functions, J. Reine Angew. Math. 463 (1995) 1–25.

[22] C.L. Siegel, Ueber einige Anwendungen diophantischer Approximationen, in: Ges. Werke, Bd 1,

Springer, Berlin, Heidelberg, New York, 1966, pp. 209–266.

[23] P.F. Stiller, Classical automorphic forms and hypergeometric functions, J. Number Theory 28 (1988)

219–232.

ARTICLE IN PRESS

N. Archinard / Journal of Number Theory 101 (2003) 244–269 269

[24] K. Takeuchi, Arithmetic triangle groups, J. Math. Soc. Japan 29 (1977) 91–106.

[25] H. Weber, Lehrbuch der Algebra, Vieweg und Sohn, Braunschweig, 1908.

[26] E.T. Whittaker, G.N. Watson, A Course of Modern Analysis, Cambridge University Press,

Cambridge, 1915.

[27] J. Wolfart, Fonctions hypergéometriques, arguments exceptionels et groupes de monodromie, Publ.

Math. Univ. P. et M. Curie, Problèmes Diophantiens 79 (1985–86) 1–24.

[28] J. Wolfart, Werte hypergeometrischer Funktionen, Invent. Math. 92 (1988) 187–216.

[29] J. Wolfart, G. Wüstholz, Der Ueberlagerungsradius gewisser algebraischer Kurven und die Werte der

Betafunktion an rationalen Stellen, Math. Ann. 273 (1985) 1–15.

[30] G. Wüstholz, Algebraische Punkte auf analytischen Untergruppen algebraischer Gruppen, Ann.

Math. 129 (1989) 501–517.

[31] M. Yoshida, Hypergeometric Functions, My Love, Aspects of Mathematics, Vieweg, Braunschweig,

1997.

- General Math DLL for SHS - (more dll at depedtambayanph.blogspot.com) Q1, Week 10.xlsxUploaded byMelanio
- Arithmetic Geometry.pdfUploaded byPablo andres7123
- Ecrin 4.10 Update NotesUploaded byKrishna Kumar
- Introduction to Functional Programming Using Haskell - Richard BirdUploaded bySaptarshi S Mohanta
- Hash FunctionUploaded bywinitzki
- Cfx9970 Chapter16 EnUploaded byYavuz Keleş
- Hack Reactor MakerSquare Telegraph Academy Questions and SolutionsUploaded byAnonymous XB8cNE7dz8
- Introduction to the Theory of Groups - G. PolitesUploaded byAlexander Lopez Hernandez
- Business15 Exponential FunctionsUploaded bymarchelo_chelo
- Real Functions - Prelim 16362871361872Uploaded bykit42
- ConMcohomology1.pdfUploaded byDieter Luisenhoff
- precalculus practice 2Uploaded byapi-248877347
- Arrival of the FittestUploaded byAngel Smith
- Tangential Algebras and Tropical Potential Theory.pdfUploaded byHenry Pablo
- 1.3 Properties of Parent Functions 1.4 Domain RangeUploaded byBradley Singh
- Kilpatrick Algebra GameUploaded byJack Oranle
- Representation Benefits of Deep Feedforward NetworksUploaded byRichieQC
- Review Double IntegrationUploaded byJoana Rose Dela Cruz
- 20378737766Uploaded bygr1bble
- individualized instruction projectUploaded byapi-278804002
- MA1200 Chapter 2 Sets and FunctionsUploaded byWai Ho Choi
- 10.1.1.68.9906Uploaded byHORUS_KHAN
- Real Functions - Chemisty162763Uploaded bykit42
- MATH1009 Test 4 SolutionsUploaded byfatdoge
- assgnUploaded byZhichao Wang
- Exploiting Symmetry in Complex Network AnalysisUploaded byMessiasBruno
- lecture1.pdfUploaded byAwais Chaudhry
- hw3_solnUploaded byIlham Saputra
- Presentation 4Uploaded byPAVITHRAN SRIDHAR
- Elements of CalcculusUploaded byMark Williams

- Davidoff, Sarnak, Valette. Elementary Number Theory, Group Theory and Ramanujan Graphs (CUP, 2003)(ISBN 0521824265)(154s)_MAcUploaded byRenee Bravo
- Tarasov v, Varchenko a - Identities Between Q-Hypergeometric and Hyper Geometric Integrals of Different Dimensions - Adv. in Math. 191 (2005), 29-45Uploaded byRenee Bravo
- Qi F, Chen C P - A Complete Monotonicity Property of the Gamma Function - J. Math. Anal. Appl. 296 (2004), 603-607Uploaded byRenee Bravo
- Narukawa a - The Modular Properties and the Integral Representations of the Multiple Elliptic Gamma Functions - Avd. in Math. 189 (2004), 247-267Uploaded byRenee Bravo
- Andrews G E - Ramanujan_s Lost Notebook VIII - The Entire Rogers-Ramanujan Function - Adv. in Math. 191 (2005), 393-407Uploaded byRenee Bravo
- Van Assche W, Vanden Berghe G, Van Der Jeugt J - K. Srinavasa Rao and His Work - J. Comput. and Appl. Math. 160 (2003) 1-8Uploaded byRenee Bravo
- Karpenkov O N - On Tori Triangulations Associated With Two-Dimensional Continued Fractions of Cubic Irrationalities - Functional Anal. and Its Appl. 38 (2004), No.2, 102-110Uploaded byRenee Bravo
- Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Applications - Adv. in Math. 187 (2004), 53-97Uploaded byRenee Bravo
- Bremner a, Tzanakis N - Lucas Sequences Whose 12th or 9th Term is a Square - J. Number Theory 107 (2004), 215-227Uploaded byRenee Bravo
- than R - Some Properties of an Extended Wigner Function - J. Comput. and Appl. Math. 160 (2003) 259-263Uploaded byRenee Bravo
- Friedman E, Ruijsenaars S - Shintani-Barnes Zeta and Gamma Functions - Adv. in Math. 187 (2004), 362-395Uploaded byRenee Bravo
- Avdeeva M O - On the Statistics of Partial Quotients of Finite Continued Fractions - Functional Anal. and Its Appl. 38 (2004), No.2, 79-87Uploaded byRenee Bravo
- Waldschmidt M - Algebraic Values of Analytic Functions - J Comput. and Appl. Math. 160 (2003) 323-333Uploaded byRenee Bravo
- Vanden Berghe G, De Meyer H - A Special Identity Between Three 2F1(a,b;c;4) Hyper Geometric Series - J. Comput. and Appl. Math. 160 (2003) 315-321Uploaded byRenee Bravo
- Andrews G E - Ramanujan_s Lost Notebook IX - The Partial Theta Function as an Entire Function - Adv. in Math. 191 (2005), 408-422Uploaded byRenee Bravo
- Yesilyurt H - Four Identities Related to Third Order Mock Theta Functions in Ramanujan_s Lost Notebook - Adv. in Math. 190 (2005), 278-299Uploaded byRenee Bravo
- Ono K & Penniston D - Congruences for 3F2 Hyper Geometric Functions Over Finite Fields - Illinois J. Math. 46 (2002) No.3 Fall 679-684Uploaded byRenee Bravo
- Tóth L, Bukor J - On the alternating series... - J. Math. Anal. Appl. 282 (2003) 21-25Uploaded byRenee Bravo
- Ahlgren.. comUploaded byRenee Bravo
- Sills a v - Identities of the Rogers-Ramanujan-Slater Type - Inter. J. Number Theory (2006)Uploaded byRenee Bravo
- Nayandeep Deka Baruah, Nipen Saikia - Some General Theorems on the Explicit Evaluations of Ramanujan's Cubic Continued Fraction - J. Comput. and Appl. Math. 160 (2003) 37-51Uploaded byRenee Bravo
- Ole Wamaar S - The Generalized Borwein Conjecture. II. Refined Q-trinomial Coefficients - Discrete Math. 272 (2003) 215-258Uploaded byRenee Bravo
- Yee a J - A rial Proof of Andrews' Partition Functions Realted to Schur's Partition Theorem - Proc. Amer. Math. Soc. 130 (2002) No.8 2229-2235Uploaded byRenee Bravo
- Yang Y - Partitions Into Primes - Trans. Amer. Math. Soc. 352 (2000) No.6 2581-2600Uploaded byRenee Bravo
- Schlosser M - A Non Terminating 8phi7 Summation for the Root System Cr - J. Comput. and Appl. Math. 160 (2003) 283-296Uploaded byRenee Bravo
- Rivoal T - Selberg's Integral and Linear Forms in Zeta Values - J. Comput. and Appl. Math. 160 (2003) 265-270Uploaded byRenee Bravo
- Yang X - Approximations for Constant e and Their Applications - J. Math. Anal. & Appl. 262 (2001) 651-659Uploaded byRenee Bravo
- Yang T - Taylor Expansion of Eisenstein Series - Trans. Amer. Math. Soc. 355 (2003) No.7 2663-2674Uploaded byRenee Bravo
- Milne S C - New Infinite Families of Exact Sums of Squares Formulas, Jacobi Elliptic Functions, And Ramanujan's Tau Function - Proc. Natl. Acad. Sci. USA 93 (1996) 15004-15008Uploaded byRenee Bravo
- Yee a J - rial Proofs of Generating Function Identities for F-Partitions - J. Comb. Theory Ser.a 102(2003) No.1 217-228Uploaded byRenee Bravo

- Igor V. Dolgachev-Introduction to Algebraic Geometry (2010).pdfUploaded byJosmir S. Rodríguez K.
- PolyLinkagesUploaded bygkovacsds
- Harvard University Qualifying ExamUploaded byJonel Pagalilauan
- Number Theory IIIUploaded byAnu Olutimilehin
- Ramon Urquijo Kempf NessUploaded byWalter Andrés Páez Gaviria
- Introduction to Tropical Geometry SturmfelsUploaded byodracir_04
- BS Syllabus 02-10-2018Uploaded bymustafaphy
- Universitext Karen E. Smith Pekka Kekäläinen Lauri Kahanpää William Traves an Invitation to Algebraic Geometry Springer 2000Uploaded byNikolai
- Algebraic Geometry - A Concise DictionaryUploaded byAnonymous QlJjisdlLI
- CA Arturo Fernandez AbsUploaded byAndres Lipa Carrizales
- Algebraic Geometry - A Concise Dictionary (gnv64).pdfUploaded bygarrisonian96
- Algebraic Geometry - GathmannUploaded byNiflheim
- [Caterina Consani, Matilde Marcolli] Noncommutativ(B-ok.xyz)Uploaded bygog
- Gathmann Algebraic GeometryUploaded byant314159265
- ttUploaded byAdrian Camacho
- Algebraic GeometryUploaded byXaed Hazhme
- Hartshorne - Open ProblemsUploaded bycompul123
- Abelian VarietiesUploaded byVincent Lin
- uUploaded byRuiPaiva
- AGUploaded bykiscl
- (Book Draft_) Wolfram Decker, Frank-Olaf Schreyer, Oliver Labs-Varieties, Gröbner Bases, And Algebraic Curves (Up to Chapter 6.3) (2012)Uploaded byJosmir S. Rodríguez K.
- annals-v172-n2-p08-p.pdfUploaded byESTALIN
- The Hodge-Arakelov Theory of Elliptic CurvesUploaded bycharlyshaka1
- Algebraic Geometry,Joe HarrisUploaded bySwapnil Bishnu
- algebraicgeometry-shatzUploaded byKevin Shi
- A Survey of Residuated LatticesUploaded byNathan Lawless
- Algebraic GeometryUploaded byBoobiez
- (Graduate Studies in Mathematics 128) J. M. -Tensors_ Geometry and Applications-American Mathematical Society (2012)Uploaded byElíGomaraGil
- computational algebraic geometry markwigUploaded bygersan2008
- AGLN grpthryUploaded byAakashParan