Hypergeometric Functions

of

Two Variables

By Tosihusa KIMURA

1 9 7 3

Hypergeometric Functions
of
Two Variables

By Tosihusa KIMURA

The research of the author was partially supported by NSF .

GP 27275 and N00014-67-A-01130019".

: .....

This note is based on my lecture given at University of

Minnesota in the winter quarter of the academic year 1971-72.
The author thanks very much Professor Yasutaka Sibuya who made
a manuscript.

The author thanks also Professors and students who

attended my lecture and gave me many advices.

" :

.., ..

....,

CONrENrS

Introduction
Chapter I:

page

The hypergeometric

F( ~

' )', x)

hypergeom~iric function

functi~n

1.

The hypergeometric series and

2.

Contiguous functions ..

3. :Euler inte.grat representation

4.

'

S~

Hypergeometr!c

6..

Monodromy :r:epresentati.6n

QuiP,ter II:
7.

18 .

diff~t;~~tial eq~tions

26
33

The hypergeometric functions of two variables

Defini tion of the hyp-~rg~omet~!d fu~~tlons'. ~"

variables

. . J ,;

8.

13

represent~_t:J._on

Barnes integral

..:.: :. .

40

.two
40

.. : :.. .

Do~ins _;of convergence, of th~ hypergeom~tric series

of two variables

47
. .. . . :

54 .

9. _.Contiguous functions
representation~. ..
..'

:61

10.

Euler integral

11.

Barnes integral representat.i ons

'71

Systems of partial differential equat;lo_ns

79

,.

' .~

.. .~-

12.

13. Systems of total differential equations

14.

Transforma-.t ion of the differential equations for

Fj into systems of total differential equations

Chapter III:

15.

Monodromy group of the Gauss differential

equation

Euler transform

.1 ~ :

81
87

98

98

16 . Connection formulas and monodromy group of the Gauss

differential equation
111
r

...

' <!

~-1

:: .: . . ::;

Monodromy' representations for the Gauss differential

118
. e"ltul:ti.on. ...

17.

'

: .: ..

. .J.

~hapter IV:

.~

..-

Solutions of systems of par:t,i al differential

equations for hyp_erge.c:>m~tric . functions of two
..

variables

124

1s.

Solutions of the sy~tem for ' F

19.

Connection fonmii~s ahd ~o~bd~~y representati ons

for the syste~ _(12 .1) sa~~~~i~d}?r ,F 1 . , .

136

General solutions of the systems (12.2), (12.3)

" ,,. ' ... ,,.

144

20.

124

.a.nd (12.4) . . >. .: . .: . .

: :

21.

......

. 149

Euler transform in douQle integral

. :: .

2Z .

Characterization of the systems of partial differen.,,~_ ti_~~ .~';1a,~ion~ . .satisf,~ed by , Fl' r 2 , F 3 . an~.., F4 ', 151
..... . .

.. ' . :

_ ~;,.

; ,1

'

Chapter V:

Automorphic functions,
zations

.'23.

Automorphic functiotlis

24.

Reducibility

'~5.

Generaltzat:ton"s '

red~~ibility

and $enerali- .
163
163

175

-::; .(

184

- 1 -

INTRODUCTION
The hypergeometric function

F(CI{ , ,

r, x)

and the hyper-

: gebmetric differential equatiol\ satisfied by F( ~,

sfudied by Euler

and

~, r, x)

were

Gauss. It .Was Gauss who investigated this

.

furiction and its differential equation extensively.

Therefore,

. the ' hypergeometric differential eqtiation is ca lied either the

<' Euler-Gauss equation or

the Gauss. equation. The work of Gauss was

"'J.:'.....supplemented
by Kummer.
.
.
.

NeVertheless . various formulas were not

.....
:..well understood until Riemann clarified the fundamental properties

::

'

. of .tfie hypergeometrl.c function and its equatio~ . from the function-

:..

theoretic point of view.

tial eqtiations which are
.. .

Riemann found a larger class of differen-

cciited Ri~nn.is differenti~l equations.

.

; Bis
study was followed by many mathematicians such as Jacobi,
..

~,

; ;

,.,,Schwarz~
Goursat, Barnes and so on.
..:. ;,
.

'

Above all, the work of

.

..

~chwarz

was remarkable and imi}ort~nt. . Considering the ratio of two Ii.near~

' 1y independent solutions,- he discovered a new.class of automorphic

, .functions
~ -~~=

'

other
:tha~' e lliptic

modular functions.
.

On

the other

hand; following Riemann's point of view, Fuchs established. the

theory of Ii.Dear ordinary differential equations in the complex

domai.n.
The function

(i)
..,: (i,i)
1'

and

F ( oe,

I! , r , x:)

has three important expressions:

power series expressi9n,

Euler integral representation,

- 2 -

Barnes integral

(iii)

representation~

By using the.se expressions> the hypergeomet:ric function has been

..
generalize~

as

functions of two variables

generalization of .the hype~geometric

Appell called these functions . the hyp.e rgeometric func-

function
. .:

In 18,80 Appell introduced four

to .many new functions.

.:._:

tions of two variables.

tial equations.

'

. . 1 .

~ystems

The functions and the

,-.

'

par~i,al... diffe~en-

They satisfy systems of

. .. : .: ...

..

.:.

,.

- -

':

:- ..:. ..-

..

:.

. . ..

..

. .. .

: : .

: .

~ ~-

, . ': : ....; . 'i _: :: .

of differenJ:ial
:

~: ,

.'

P.> i(,x) .

equations. are very similar t:o the fw;ictioq . F( 0(,

.

~es

and

differential .equation.
The systems
of .. differential .
..

eql.tatio~s sa~is~i~d ~Y.

: . .
: , , ' ; . ; :
..

the hypero~

geometri.c functions of two variables are essentially systems

c~letely

~he

integrable t:otal differential equC\tions. .. Recentl}"

foundation of the theory of analytic unctions of several complex

,;

':.:

. :

:-;

'

varf.ables has been completed . Such a theory should be .u tilized to

' ' .., ~.

: . . . .:~, .

'. ~

:: .; . .

. ~ >.-

. .

theo~

establish a general
as

. , . .

..

.: .

: .

eq~tions

of total .4iffe.r ential

as we.11

.partial differential equations in the. complex domain. _ As :a

, ' . :

, ,

. , ; ~,

'!,

matter of fact,

. '

c~lex

' sueh a ,:new

in the

,.

stu~y

..~ .

stag~
,

.- .

~: ';_~ !

": , .

I :_

, :

. .

It seems to

.. .

\'

'

.. .

~ t~t

'

peri.od from Riemann to Fuchso

',

. , , ,,

we are,

quite sil_Dilar to that transitional.

.,

of total dif.f erentJ.al: . .e_qua~~ons , .

domain .has just begu,n.

passing through the

.

'

:. : :

. . ~- . .

. .

. , ' ,,

- 3 .. ,

CHAPTER !.

p, '1 ,

The Hypergeometric Ftinctiqn . F ( c< .~

1.

~he

x)

hypergeometric series and hypergeometric function.

We start

with the power series

.,i (

(l.l)

o( ,

m)( ~ , m)

mcO ( '( , m) ( 1, m ) x

. which is called the hypergeometric s:e ries' where

are complex parameters and the notation

0(

'

(3

and

'(

(a, k) . denotes the factorial

function:

(a, k)

a(a+l) {a+k-1)

(a, O) "" 1
in particular,

. for

for

= 1, 2,

a 1' 0 ;

(1, k) ""k!.

'( 1'- O, -1, -2, .

Suppose that

Then series '(I.1) is always meaningful.

Remark 1.

- Remark 2.

Series ~l. 1) is symmetric with respect to

If orie

of

ri. . a~d .

f , i~ equal to 'zero or a

ci .

and

negative

integer' then series (1.1) is r~duced. t~ ~ polyi:iomial.

Remark 3 .

Since
H(a

k). ~

rca+k)' , for

. . , rCa)

f: 0, -1/>~ 2,

series (1.1) can .be written as . :.

o( ,

p; 0,

re r 2 . 00 . e {Cit +m) r <a +ni) xm

re> r <~> 0 r<r+m) rc1+m) .

.,

.... .-

-1, -2, . . . .

The ratio of the m-th coefficient to the (m-1)-th coefficient

of (1.1) is

~ ...

- 4 -

rn

(oc+m-1) +m-1)
( )'+m-l)m

and we have
(+m-1) (~+m-1) =
( +m-l)m

lim .

m~'

Thus we obtain the following theorem.

.

..

THEOREM 1.1:
one if neither

radius of . convergence

The

nor

C(

is equal to

of the series (1.1) is

0, -1, -2, .

Th~ function defi~ed by (1.1) and its analytic

DEFINITION 1.1:

continuation is called the hypergeometric function and it is denoted

by

,;

F ( ot. ,

Y, x)

In general, the function

F(,

p,

'1, x)

is an infinitely many

valued function with singularities at x = 0, 1 and

Therefore the analytic continuation
C-[1, _oo)

determines a branch of

phic there.

~ 1

at

= 0.

F ( oc,

p.~

-':>' : ~ f '': ;;-..: -::!!.': . ~~ -~ :-~- ' -~~.~~:~ ~~~-=-: .... ~ .

~y

p,

'

'f,x) "which is holomor'

have,
in general,
. .. ., .
.
..
.
.. . - ..

.~

~-

In this section and thereafter we keep

..

the notation
:;

F(ot,

_.

only.

(1.1) into the. domain

However, other branches . of. F.

branch~points

...

of

co

...

... t~. d~?o~~ - --~h~v br_~?~~ ...~e.~ine.~. ~,i,1;_e,~t:~Y

y ~ .~?
:7 .

.1.

). --~ .

. ~'

' ..

' .

. .

'

series (1.1) unless otherwise stated.

. :

A series of the form

. (1. 2)

.:. L

(-l)ma

m=O

(x-l)~;<x-m)

m.

= i(-l)ma
m=O

(x-1)
m

is called a Newton series or a factorial series of the

It is known .that if (1.2) converges ata point
-

x =XO

second .kind. '

and if XO

is not a positive integer, then

..

(a) "the series (1.2) converges for Re x >Re XO'

.5 -

_, ~~~ . tlle series (1 : 2) converges uniformly for

.

and for

I <.

larg(x - x 0 )

7t

--&
, where
2

Re x ) Re x

.f. S

0
is an arbitrary positive

. constant,
~o

i.-. ,:...

the series (1.2) represents a function holomorphic in Re x > ~.e x

0

(c)

. . .

The series (1.1) can be considered as a Newton series with respect

to

1- oc

and

..

p.

1 -

In fact, (1.1) is rewritten as

. . _~ . . i

f,{:-l)m

. m=O

m . '.'. . . .. . . ..

((!,m)x

. ( }'m) (l,m)

(1--1)(1-ot-2) (1-0(.-m)
.
.

and
~
m
2...(-1)

m=O

{ ol ,m)xm

( 1 ,m) {l,m)

Since (1.1) is convergent

r10,
c{

-1, . -2,

and

for

fo~

(1 - ~-1)(1- ... -2) {1-t1-m).

any .~ : , and .

., .

. .

provided . Ix l <. 1 and

the function F is holomorphic with respect to

- : :-

lo(l <eo and

.,

ap I<. ~.

A series of the form

'

(1.3)

. m=O

m!a

xcx+1>. ~ cx-1-m>

is called a factori al series or a factorial - series of the first kind

. .

,-

.!

It is known that if (1.3) converges at x = x 0 ,

(a')

(1.3) converges for

Rex> Re x 0 possibly except at x = O, -1,

' :
.

-2, . . . ., .

then

- 6 -

(b') .
k

(1 :~)

converg~s ~ifo:rmly for. Re. x

> Re
..

7t.

+ S , I x - kl > S ,

1arg(x -x0 )( <; 2-J'. ,

and for

0, -l) -2,

is an arbi~rary positive constant,

&

where

lx-k I>~ ,

(1.3) represents a function meromorp~i~ in the domain

(c')

' ,

>Re x

Re x

. ..

...poss~bly with simple poles at x = 0, -1, -2>

. , The.. series (1.1) ~: omitted by the constant term can be considered

J . Therefore the. function F

as a factorial series . in

function. pieromorphic with respect to .

-..

at

0y

'.

: ... :: : J ..

.:

0 , -1 , -2 ,

possibly with simple poles

Thus we obtain the following theorem.

.~

. . . ,, ,, >'. ~ .. T~OREM 1~2:

is a

;: .

The . function
.

'.

oe,

13, )',

x)

is a function which

. '

.o< , . ~ , l

is holomorphic in

F(

and

x . i .n the domain
,,

. ;. ~: - ,..'. C ' JC C . K . (C ',.;. { 0, ~l, -2,

0
"=

and whose poles are at

Remark 4.

and are all simple .

0 , -1 , -2 ,

As for the convergence of' {1.1) oil

lxl

= 1,

it is

known that (1.1) is

. : ., absolutely

c6nvergent

if . Re {)'-DC - ~)

> O,
0 ~ Re ( - oC - ~ )

> -1

and

divergent if

l ~ Re ( '(- o<. - ~)

Exercise .1.

Prove Theorem 1.2 directly without using the

. . ; .. ! .

properties of Newton series and factorial series.

:_.'

Hint . Consider

the series F( o< " ~ , ', x)/ r

( )")

and, using

the formulas
,.

lim

(a,k)
k~oo (k-1) !ka

=1

or

rc~+k:>

rca) r(k)

-:

(k

-I)>

00 ) ,

,.

.j

.. 7 -

show that

~.

r(o{+m) rc@+m) x

1 r< '> r c~) ro +m) r(l+m)

1

m = O, 1, 2, }

is bounded .
References:

Milne-Thompson,

The calculus of finite differences,

271-321,

and
":

Norlund,

Le~ons

sur les

s~ries

....

d'interpolation, 1926, 99-227.

- 8 -

2.

Contiguous functions.

In this section, we shall start with

the following theorem.

THEOREM 2.1:

and its derivative satisfy the following

relations
.

xF' (oc.,

..

p, > x) + O(F(Ol, ~, , x)

= ocF_(o<.+l, ~, '/, x)

x{l-x)F'(o(,~;r,x)+ ('(--px)F(ot,p,'i,x) = O'-o<)F(~l,~,y,x)

}'(1-x)F' (o<, ~, y ,x)

Remark 1.

+ 't ~'t-ot.-p)F (o<., p,'r ,x) =

p, 1+l ,x),

We can derive two formulas from the first and the

' and

second by interchanging
Proof:

(1-) (t-~)F (ix,

(See Remark 1 in 1.)

Let us put
_ ( 0( , m) ( p, m)
Am - . ( '(, m) (1, m)

to write

F as
F ( o<. ,

(2.1)

00

p, "(,

x)

= ~A

~U\
m

m=O

Clearly we have
(2 .2) .

xF I

o<._, ~,

r,

CIO

x)

=>

mA xm

m=O

Next from the definition of the hypergeometric series we derive

F(o<+l,

~, 1,

(c<+l,m)(~,m) m
x) = ;:O (o ,m) (1,m) x .

.,M...,

Hence
(2 .3)

F(ot +l,

fi, '(,

In a similar manner, we obtain

x)

~ o<+m (o<,m)(~,m) m
~O-;;:- ( (r,m}.(1,m) x

- 9 -

(2.4).
From (2.1) and (2.2) we derive

r, x) + oCF(oe, p, r, x)

xF' (, a,
I"

This formula together with (2.3)

the

ytelds

E (mh<)Am'x_m.

m=O

first:

r~l~'iion

of the

theorem.
.

'

Now let us prove the second relation.

x(l-x)F 1 (0<,

:. . ~

~, '{, x) = .(1-x) ~

eo
m
, - mA x .--.
m

,......

'.

~ E mA xTA.i-~ i . ~

mA xm
m:::0 . m

m=O .

From (2.2) we derive

xm+l
m=O ., m

m=O . m

. o-. . '

.., .., ...

m
'
(m-l)Am-1x
~

.:

m=l

By tpe .defini ti.on .we hav.e

A
m-1

(,m~l)(8,m-1)

:. ,. i

. .

and we make the c onvention

90

A_

= 0.

::1_L_ (m-l)Am_1x

m=l

Then
~ : :

= :L

(m-l)A

in=O

m-l

x m'

and we get

x(l-x)F' (o<,

~. l'; X) = ~F ~~t:~~~~:!"lijYAm~.;
.

On

.'

',

'

.. . : .; .

,. .

',

'

the other hand,

to . .
:
..
. ' ..a .
.\
m+l
('t-c(-~x)F'(0< ', p ~ {,'x) '= 2: <.r-d..)Amxm- L, ~Am;._
,
. m=O . .'
.
m=O .. . ...
.

. . . J

_.,

. .. . .

L.
DC)

..

((-ol)A xm ~ .

m=O .

. m

. .... , .

(m = 1,2., )

(o<+m-1) (jl+m~l) . m

( '( ,m-1) ( l,m-1)

,;

()"+m-l)m

00

,E
m=O

pA . x_m.
'

. : :. .

1n.~~ :._ .,

i : .

- 10 - .

Therefore

.... : 1 - 7'

_ _

_;

. : ' .

_ .!!. .(

( t o< flx)F(oe, p, 1, x) - ~O

_ _
o{

tHr+m-l)m
) . m
:(O(+m-l)(p+m-l) Amx .

Bence
~(l~x)F_' ~oc, ,,. 1 ~ x) '. (~ .-"5-tP,.,~>,F ~cl,.~, r,, x)
.

...

...

... . .

..

... .

. .........!..:_ :

(2~~)

This result together with

;.,

yields the second relation of the

f.'.

theorem~

'

-~

_.

~' . ~

.-.

We can prove .the other relations in the same way.

.

Remark 2.

be Written as

The identities .in Theorem 2.1 can

follows:

~-<+l ! (x'~( o<, ,:: ~>) ~ aCF (l(~l, ~, l', i<) ~

x -Vi<t+l (1-x) r-:<~e ~ {xr-.c(l-x) -r+at+p~(.{ , ~ ,

-~1

,-; ' '

' : _,,_..

:'
'

r.
o

'

'

' '

. .

rr.~:;:
.,

'

,. .,

x))

r,
.

' "'r ~--

{)'-ol)F(aC-1,.~,y,x),

f{i_-x) f-,o<-p+l

! ((1 ~x) -J++t!F

(o< ,

~ d', x~

a Cr~~)(y.-p.)F(,~, -~, )'~l~ -~),

--r+2 d ( r-i

x . >>dlc . ~ ~ F(~~ P.. ~

:: . ~

'

: .

_:

. 'J'

r'

<-.~ .

::.

or decreasing one

th~ ' formulas in Theorem 2.1 there

appear functions
,O:btairn~d
r,?m,
. E'(o( ~'.~. , ..r, x)
...
.
.
.. .
~

~)} ~~ (J-1)~(,, ~ .~ 1,-1, .x)

"; On the right-ha'nd members. of

.-

'.

.~

of ; ihe p~~ameters :by unity .

by increasing

- 11 -

. DEFINITION 2 .1:

,- .,

,'

,:

,,

!(

The six functions obtained from .

, : I

~ -:

.;

~,

,l.

-~.

:-

~ ' r 'x)

o( ,

by increasing or de creasing one of the parameters by- unity":

. .

:;;

F(oc ~~+l, '/,

F(o<, ~, )'+1, x) and

F ( o<~ ,

xL

..

p-1 ,

F(o(, ~, t-1, x:) . are called contiguous to

F(o<,~,r,x>~
.}_

.. ,. . THEOREM 2.2:

0 , x) ,

: .

The function

. .

_;;

F (. > ~, 'f, x)

. '

' .

and its two contigu: :.;

~ . ;.

'.: :

: : ? : : : ; ~ ' ..::

' ,

ous functions are connected by a homogeneous linear relation whose

.

coeffic'i:e nts a.t e polynonii.~is in . o(

(2.5)

ot (1-x)F (o<+l, ~,

_",

'

'I, x) + ( -20l + (o<. - ~ )x)F(, ~, 'r, x)

../ <Y-ol)F(o<-1~p,r,x>

t.. \ '

(2.6)

( Y - O{) ( ~

- ~ ) xF ( o( , ~

, )'

= o

- ~(t-1)(1-x)F(oc,p, Yl~ x) ~

x) .

For example,

from the first two iden-

The number of slich relations between -: F and -its contiguous

functions is

~)

= 15.

On

the other' hand, (2. 5) and (2. 6} are

regarded as linear difference equations with respect to

F 1 (oc, ~, r, x)

from . the corresponding . two ide~tities ~f Theorem 2 .1.

in order to derive (2.5), eliminate F'

.;

..' ~

0.

In order to ' prove these formulas, eliminate

tities.

+1 , x)

+ '( (r-1-(2}' ~ol~~~i)x)F(o<, ~,r,

.\ .

.:.

, ~, '( and x. For example,

and

:\ .

respectively ..
COROLLARY:

F (O{ , ~, 't, x),

,i c)f -three parameters

oe ,

r~

considered as a function of one

satisfies a linear difference

equation of the second order whose coefficients are poly"nomials.

- 12 ..

Notice that the coefficients in (2.5), are linear in

!

ol

: .

A difference equation of a form

(a 0 z+h 0 )w(~l) + <aiz+hi)w(z) +(a2z-l-h2)':1(z-l)

.

or

'

~c

=o

in a more usual form

(2. 1)' - -(~~ztb

is

~:

~alled

0 )w(~2) +(a 1 z+b1 )w(z+l) + (a2 z+b2 )w(z)

a hPergeometric difference equati9n .

.

.. i_eo~~:iri~:.- diffe~e:~ce
.:.

:.

: . :1'

. :Batchelder:

;~~~ion, . see

,-

:'.'. .,; ,. '~ . : ~

'.' .

...

=0

As to t,:he .hyper;.
. . . I : : . ,

Introduction to the theory of linear

~ifference

.. equations .
. ;

. ,:,
.. . :,

..

F(oe+p,~+q,
t+r, x),
i:
.
. .

A function
", .;

where

every function associated

to . ~(O( ~- ~,
.:.
1,.
'.
~

func.tions in . ~ - -' ~ - ' 1" :an,~

.~

. '...

r, x)

..
ca.n be expressible as

and F' whose coeffic.ients are rational

a linear combination. of .F
.

are integers,

FfO(;p,'t,x) . . lt is known that

is often called associated to

p, . q, r

x. T~ree

.:

,. !

functio,n s associ.a ted to

are connectec1, by a linear relation whose coefficients

. }. ;

-:r.'

:. :-: .:

. .

in ' ~'

are :polynomials

"

~ ~

and

x.

. 2
Exercise _L
~~ppose . in , (~)) _tpa~ _ ~ a I 0 .. and ai
0 2
. . ; .: _: ."". ._-: _;.: r: r
:1
.. .. - ... :, '
..._
Then show that difference eq.u ation (2.7) . admits a soltion
~. . .

.1~,_

. .

,...,

:1 . . .. .

.:

;~. ~:. ~:

w(z)

:. .. . _.

-... _ ;

_:

"':.. .

. . .-,

= eczF(z+d,p,'f,x)

"' :. )._:~~ < ,: ::.: :. __; : i: ~-.

. : -:~
';. .
c, d, ~, 't and x are constants.

,
. i

. .~ .

where

assume that

lx l

Hint.'. Make

(It is not

< 1.) .
'1

~ ;:_:

- 4a 0a 2 :f

transformations of forms

z =

g+ d,

J:l~CE;?ssa:ry . to.

- 13 -

:~'. .

;;;. l.

Euler integral representation.

There are many functions which

.,.

.~ :' -~

/ can

be defined by definite integrals~

:

.is given by

i .:

For. example, the r-function

. :"' . :.:.. .-_. .,, :' .

'

.rcx: ~ J.
r+IG e-uux-l ..du .' ..
0

. ..

,.

.
'

.. .

. ~be hypergeomet~lc fundt.i on .is also defined by a defiriite integral.

If \xi< 1

'IHEOREM 3.1:.

and

>

Re> Re~

then we have

O,

s:

(3.1) . .F'c ~; r, xl . ~ r<1SMr-p > ~~.:1c1..\l> r-~~t'(~~x\i> ~" 4u, .

-
..
...
.....: .. .
where the. integration is taken along the segment . 0 ' u
- ~

. branches of many-valued functions : ~l--l;~ '

are determined by. ..

..

arg(l-u) = 0

arg u. = 0,
. . I

. . . .

... : .

and

. .

'

1,

and ' c1..:~u)-oc

(i-tiJ 1 _,_l

1. :

'

.:

(arg(l-x:u) I

and

~ ~

'.

respectively.

Proof:
... ,. .....
_:

<1

for

~ fix~d

F.or'

O & , ~ 1 >

Ix

Ix'

such:' that

< 1~

' w~ hav~

l x~ ( 'f fx I

: -oc ,'.:,.:,._.: ',. :~) ; .' . ~:. .'

(1-xu)
is expanded

':

I .

and hence the function

. ..

- into the series

":'.

(i-~u)-~

:,.

_ (~0c) (~~ ~l) '; .- {-~ ~m+~1 (-~u)m =

m=O

m.

~ ; ..

- .

~ ( ',m) :tiixm

. ' ... m=O . (1,m) . . . . .

which . converges unifor~ly in . u . far . 0 ~ ~ u ~ l. .. .Therefore we have

~-1(,1: -u. )y-~-1(..l '-xu)etdu

-= J u

( . I . (

. 0

s=:

il

~.

-.

0 m=O

( ,m)

(1,m)

. _

f 1 ~;..l(l )-~-1 ~

J.

O,

-u

.-

m=O

(o(,m) um md

(l ,m)

xmu~+m-1(1-u ) t-:~-ldu
:

' Interchanging

..

. .

: '.; \ .;~ ' . . ..

the order of the integration and the summation, we

".

- - 14 ..;,

.....
= ""'.'

. ... .. :.

m=O

"

. .

:.. ..

: '

~+m-1( ._ . )Y-~-1 d..

0 u
. 1 u
u

'(Ci ,m) m
(l ,in} x

..

.,

Applying the definition o( the: Beta;,:furi.ction:

.B(r,s) =

. . :L''"

r~(~~))

: . . . :

=.JO u.r~:~(l~u_l~-l- ~u,.. . {Re~ .s

Res~O),

we .obtain .
I -

.rm+m> r.~r-~) = rc.f) re r-p)

.!!.. . (cc ,m) m

(1,m) x
.

:0

. f,(t+in)

rC ~,_>

. ...
'"

.:

2::

(qe ,m) (@ ,m) m

m=O ( :r-,m) (l,m) ~

Therefore Theorem 3.1 is proved.

.; .

:;,)

~.
. .

. . Remark

'

:,: ;:_

.. ~

:. i "

Tile -Euler, integral .

. .

1 ,_1

r<1>

f(~)

rcr-p>

.::.: '. . . . .

r-~-1
-et
( 1 -xu)
.. . .
. ''": .' . ,. .

: . .. . . . . '. o,,'

is. well defined for

' .

. -

(1-u)

x E C - (1, eo)

Re }' ') Re{!

if

-.-

du

..

,.

> 0.

This

'implies
that the hypergeometric ,;eries is analytically continuable
. '
.
.
::..
..
.. .
.

into the domain

obtaine~

'

C - [ 1, oo)

F { oC , ~ , 'I , x)

The branch o f

is called the principal branch.

THEOREM
3.2:
: :. :..
...

'

thus
......-e

Th.e hypergeometric function has the following

...... ,. .,

three formulas of .transformation:

F(

oe, ~, 1, x)

= (l~x)-~F(ot, 1-~
=

x/(x-1))

_{lx)-~F( ,:~ ,, , _, "(; x/{x~l))

.

.:

,_.

,.

= (1-x)f-ol"'.~F( )'-oc,
which are valid for

r,

,
.

1- p, "I, x) ,

larg{l-x)' <:: 7t '._(o~ .-, C.: [1, oo)),

formulas are valid for the principal branch.

t

: .

: ; ' "

~ '1

i.e~ these

- 15 -

Proof:

t.;ehe
~~

It should be noted that by the mapping. y ... x/(x-1),.

domain . . C.,. [l, oo)

.

is mapped into _;tself in the .. one-to:-one

-

i;J~nner.

Make the change of variable u := l~v. .T hen l.,.u = v,

;;1;w.-.-.. .
x
~i};7xu = 1-x+xv = (1-x) (1- x-l v), du -dv . . Hence we obtain
~if:.-:,
.. .
. . . ;.. . . 1 . . . '
.

;r.,,~- - F( , p, "/,.x) .- = .r(~~~t~-~)

J u,_ (1-u)r-~-l<~.~xu)-O( .du

1

J:
0

rcJfn-~)

ci-x>-"

(1-v)~ -1v 1~~~1u-x)-"c1-xv/(x-l)) -"(-dv)

rc~~m-Plc ;,r-p-16-V>~-1c1~x;,/c,;~1))"""civ
,. .

=
~fthis

._,
.;_,:';

..- :

.i .

(l-x)-'F(oe,r-p,r,~~(x-1)).
..

proves the first identity

~f~ ~heo~em

,_,.u

3.2 . .,_ The changes of

~;;~~riables

,.

u
; ._:

= v/(l-x+xv)

and

::~

'.

= (l~y)/(1-xv)

..

i~-~ill yield the second and third identities respectively.

~:~:/.~

.. COROLLARY:
oy r.J.

If

t -

0 , -1 , -2 , ... ,

'.l~:~_olynomial multiplied by

"?}\':::.

Proof:

or ( -

o(.

then

is zero or a negative integer

F ( ci<

, ~,

'f, x)

is reduced to a

(1-x)r-o<- ~.

This corollary follows di.r ectly from the third identity

' ::

/{of
Theorem
..

:; .'

3 2 .

Consider the series

S_;~'$ it was previously mentioned , the series (3. 2) is absolutely con..

, :~.:;~: .

: ;;~ergent if

Re(l'- Ol-~)

>O

and . y /: 0, -1, -2, .

This implies

. - 16 -

that lim F(o<,

x..,.l

p . Y>.x)
. .

_~~i~;s and

.
.
p; "1 ;. l) = lim.
F (o< , p', f

F(c(,

x-+l

,x)

. ..

= 2:.

. . m=O .

(c~.m)(~.m)
y ,m)(l ,m)

;'

. ' '

Re ( '( - o< - ~ ) :> . 0

if

:f 0, -1 > -2 >

and . Y

':.

'

. It is also .known that (3.2)converges uniformly in every

compact set in the. domain

': : p.) .> 0 ,

.. Re ( .

,, .',

.. . .

FC

1:his means that

..

~,.

':

0(. ,

,",

~,

t>

>

is holo~orphic' in

. . in th.i s domain>- ()~ the other hand, .for .. x ' = 1> ; Rep
~

: .~:

.. .

'

'

~e( f-ol.. - ~)

>

0,

and

> O,

the Euler integral is well defined and equal to

.

f0 ~ . . (1-u>.
1 . . .

r<t >

r.<~>r<1-~)

~-1

r-~-p-1

du.

Furthermore,

. " .-

for

Re ~?Re ~) 0,

. ..

.Re ( 1- 0(
'

....

~)

> 0..
.

:' '

This integral is .equa~ . to

.....

r< 1>rcr-0( - 6 >

. . r(~-o{) r<~-~)

Hence
F(o<,

fo, r'

1 . _ r<r>rcr--~>

>-

rcr~>r<x-~>

By virtue of the theore'.'l of identity in the function theory, this

relation holds f o r
.

ReCi-~)

>
' ,,

''{--# O, -1,

0,
:;,

; .-..

we obtained the following theorem~ ;-. .. .

THEOREM 3. 3:

We have

-2, . Thus

- 17 -

F(a(. As rs 1)

Cl:

~.

is holomorphic in the domai~:

,~

(cf

,m>CI ,m)

m=O (~ ,;m)(l,m)

..Re(y- .

ot -

p> .> .o., ,

.

'

.~

and equal to

rcoc)f(r-oe -{!)

r<r-> rcr"- ~>

Exercise 1.
r

. . : /.,

Prove that

d9

2.2 =2"JC

O yl -k

Sl.ll

F( 1 1 l k2 ) .
2'2>;

Y I O, -1, -:-2,

. ..

- 18 -

Barnes integral representation~.

4.

Th~t::e

~ (o, ~,

integral representation of

Barnes integral representation.

i,s

anot:.he~

type of

which is called the

'(, x)

We shall make a general and

introductory remark on the construction of such a representation .

..
Let f (x) be a function define4 by a power series
m .
ax

DO

f (x)

= 2:.

. m=O

Suppose that

ther~ '~.xists .

a function g(s) . s atisfying the follow-

ing conditions:
(a)

g(s)

is meromorphic in

(b)

g(s)

is holomorphic at

(c)

g(m)"

lsl<co,
s

= 0,

1, 2,

= am

Let us put
h(s)

= g(s)

1t:

sin( 7Cs) (-x)

Condition (b) implies that the points s = O, 1, 2

most simple poles of h(s)

~nd

for

s=m

2~.

is

are at

condition (c) implies that

. Res h(s) =axm

m
Therefore, if C'

'

= 0,

1; 2, . . . . .

a loop which surrounds the poles

s = 0, 1,

N in the positive sense and only those poles, then we

have
1
2'1ti

fc,

N.

h(s)ds

= 2::

m=O

a x

m.

If c denotes the inverse loop of C' ,

we have

- 19 -

- 2

,!i

lc

a x
m

h{s)ds

.'\~ter ;.this . preliminary rem.:l'rk' let us go back to the hyper~i.

~':

:~eometric series

(o(,m)('fl,m)xm
F(oc, ~,...,,a , x) ""' ~
~
( v
) (1
m=O
o. ,m
,m)

("'
11

J.
T

a
1

2 .

As. it was previously mentioned, ,this series can be written in

r -function

terms of the
F(o<,

as follows:
C(lf)

tO

~' r,x)

~ C(<ll)r(~)

m ..
r(?f+m)r(l+m} x ..
r

,I

' "I

'

y is zero or a negative integer . ,

nor

'

, . <

~,

.....

nor

oc.

rc~+m)r(@+m)

I.

if neither

.In such a case we can naturally take as

g(s)

the function

g(s)
This function

g(s)

(a) is satisfied.

.If

o( ,

p:f

is meromorphic in

(s I< oo and hence condition

Poles. of
-..1

c::

are at

.
"" . -ct-1
. ,

S =

O, -1 , ~2,

or a positive integer.

g(s)
)

p ,, - p -1,
'

. ..

then none of these points is zero

This means . that condition (b) is satisfied.

Clearly
- .(o<,m)(f ,m)
g()
m .-. ( )', m) ( , m)

This implies that condition . (c)

for

m = 0, 1, 2,

fs. 'satisfied.

- 20 .,.

.. .

Note that the function

7t:

h(S,) ;~ ~(s) siq(x:s) ~-x)

r<r>

.._'

.:

C(o<+s)r(~+s)

;c

)s

(-

= r(o<)r(p) r(r+s)r(l+s) sin(7C8)

has its poles just at

R~call

::

, ....

..

'

'.': .: ''7t
" . .';; . .. '. .

:>-.: ;::.: ::. _, .'.

r(s) r(l-s)

sin(1CS)
' ; . .

, :

- '}. .( ...

' :

'.

:::

.. ,_

: ..

' ,

'" I

: ,. . . , ..

'

..

'....

'::'

. ir

rC-s) r (l+s) =----.,....

sin(-11:-s)
,. r. '. :. : r _;,_

. . .

_;

From this formula we can derive

'

:'"

l ~function

a well-known formula for the

1 .. :' ;_ .....,.

,;.,; _\

""f

...

fJ ,,..-p : 1 ). _~ -~ . ; 0., 1, 2,

oC - ~, . .. . ~ ; ./ ,.

s "." .- - ,. -

'.1'

'

sin .( n:s) '

;' ; 1-

";.

;"1t

== -

. ... .

;.'.l

..:.

and hence
,..... ..,.

...

,..

..

) ,,

\.

~ -~

::. '

rc1+.~ )~i.n (~s)

: _;

".'",. r<-s,~

.~ ~ ... :

:,, .,

.. , .. :

_,: .
.

-.

Therefore we have
_; ( :

: ;-

' :_.

-h(s)

'

~--

..

. ;

. ;

: : .

. '.

THEOREM4.l: Suppose that

for

tx\

'I , x)

starts from

~ -

- ' . .... .,. . '

oe,

'

.. . .

~.:.

p, rt:o,

-1, -2,

Then

(-x)s ds
f(t_+s) r(~+s).: .sin{ 7tS) :
.
,.

. 2.1C~- r(~) r(~J JB

7t:

lciirg. ..(1-x)
I < -1 t "' :,where :'th~:
patli" 6f...integration
.
. . '_;; .-...

1,

~ioo

:.

+i~

and goes to

:_~;I ii;\:::;_\_; \\:~~.. ;~~!

..

;'

-~

. : , .

' ~~

.. . . .. .

_;

rc),} . :f C(ct~-+s> C{f +s)

= _ _L

'

ro+s)

Now we shall prove the following theorem

F ( ot., ~,

rcc<+s) r(@+s) r~-s) (-x)s

r(r)

rl(>r<~)

~,

in

. ~l)e . cq~lex, , s-plane, run- - ~

-

'

' ol

- 21 -

ning along the imaginary axis, but curving around in such a way that
the poles
.. poles

= 0,

1, 2,

. lie to the right of

= -oc, .-oc-1, ;--p. - ~-1,

. --1..
..

. 0:

B, . while the

lie to the left of

B.

2.

. . . .

-i6t>

.. Proof,,:;, _, 'I\l~ ,.Pr9Q~ .;. ~s. di;yi~ed.. intq ._.~o~ ~ teps. ~ -~ .. :

I)

If

o<,

p, '( -:/:

O, -1, -2, ,

then the integr,al

( . r.c~+.s,l(~~;s) r<-s>: (""x)s ds

is holomorphic in

for .

I arg (-x) l

be
It is sufficient:
x

for

I arg (-x) I

t()

< n.-.

: . .:

an' arbitrarily
~

..

small positive number

shaw that this integral :coi:iverges uniformly in

~ '7C. -~

f,.

In . ot,her word~' ~or ~ny '

e >;:o, : .:we .

- 22 -

shall find a

>0

Ir. .

we have

ds

. i.to

such that

I< c.

.-

is independent of

J-1"'

-.it 0

and

" ds.

1<

and that

To . do this, we must estimate the ,.,i ,ntegrand.

Recall that the asymptotic behavior of the

r-function

r(s)

is given by Stirling's formula

r(s) "- J2 7C. e ~SSS-~
which is valid as

1t - S

tends to infinity in the sector

I arg s I "'

This formula implies

log

r<s)

=s

logs - s -

~logs

+ 0(1)

and hence
log r(a+s)

=s

logs - s + (a-~) logs + 0(1)

. ..

In order to estimate the integrand on the upper part of B,

let us put s

= it,

where

is ptjsitive and sufficiently large.

Then

Observing that
log(it)

log t

i \' 7t,

we obtain

:
1
.
. 1
1
log r(~.+s) = --21tt + (Reo(. '."' 2 )log~ - : 21timoc. +PI+ 0(1)
1

where _ PI
obtain

-21tt + (Re o< -

1 .
)log t + PI
2

0(1) ,

denotes a purely imaginary quantity.

Similarly, we

- 23 -

1
= -2Ttt

log

re ~+s)

log

r (~+s) = - 1 Ttt +

l)
- 2 .log t +PI+ O(l.) ,

+(Re~

(Re '( -

1
z)
l,og

PI

0 (l)

and
log r(-s)

"on

-it(log t - i ~'JC)

.. ~ ic t

- ~ log t

+ it -

PI

log(-it)

+ 0 (1)

0 (1)

the other hand,

log(x)

= it(logtxt +

= s log(x)

.. - t arg(-~)

From these

formulas~

iarg(-x))

+ PI

we can derive

log f'(.o<+s) r<@+s) C(-s)(-x)s = Re(o<.+. A - 0V-l)log t

ror+s)

,..

- t(lt+ arg(-x).)

PI+ ~(l)

and hence

as

tends to. +oo,

where s = it.

In the same way we obtain

C{ot+s) C(@+s) r(-s) (-x)sl = O(tRe(oe+~-l-1) e -t(n:-arg(-x)

I
as

r<r+s)

tends to

. .

in case when

+oo

s.

= -it.

I arg (-x) f "

By virtue of. the hypothesis:

'1t arg ( -x)

. . .

S >0

7C -

S , we get

Therefore the quantity

. tRe (~+~-1-1) e- t (i:arg(-~))

?.decays exponentially as

tends to

:easily concluded that, for any . E.

> 0,

+oo.

From this fact

there exists a

it is

>o

such

24

that

If~

". I ~
ds

< .

IJ

and

-.it

dS

II

<

-uo

This completes the proof of I).

., .

F(,

txl

for

<

1, larg(-x}l<1t. .-

Proof:
large.
of
CN

Lrc;!:~ rn::.i~(~~) (~)

~ 1. x) = 2~i ;~gm")
.

Let

.: .

in the strip between

ds

..

..

N be a ppsiti.v.e integer which is sufficiently

Consider the closed path

th~n '

Im

- - . .. -

~s ' a semi-circle defined by

BN+CN, . . where

s = N+ %

s = (N+ %)e

is the arc

Im s :: - (N+ \)

and

BN

16 ' .
,

and

\6 l 1! ~ 7C.

As itwas explained at the beginning of this section, we have

By utilizing the result I), we can show .that the first integral
.

approaches the Barnes integral

:...L.
,21t'i

r(a'l
r(O(+s) r(15+s)
r<' >rc13> JB .
.r<t+s)
.

as

tends to

r(~sl (~

)S d '
x

+bO Therefore all we have to do is to show that

- 25 -

_tJ:1;e : s.e~ond,

~ntegral . tends ~():_ zero

J~~rg ( '"'.x)

.a s

N. tends : to .t +oo:- ~ ~ if:.: '( x

I < 1,

.~~

I < 1t.

This step is left to the readers . as Exercise 1.

. ;.::- 26 -

' -? ~ ,r Hypergeometric ,differential

equation~ We shall derive

a dif-

feJ:e11tial equation satisfiedby the hypergeometric function.

Let

us consider the differential operator:

.A1.

xd/dx.

a:

It is easily seen that, for a power series

00

.L:

a xm ,

mO m

we have

Therefore, we obtain
. ~~Jl+r-l)F{oc.,

p, )', x)

= ~ m(m+y-1) ( ,m) <D .m) xm

m=O
( 1 ,m) (l ,m)

(ol ,m) <f ,m)

~
L.J

v ,m-1) ( ,m-1) x

m=l . '

t . Col (l,~)(
,m+1> Cf .m+1> x m+l
,m)

. m=O

=x

2-i

m=O

(ol+m)(A+m)(O{,m)(~ ,m)xm
r
( 1 ,m) ( 'm)

= x (~+o<.) (~+fo )F (o{, P, '(, x).

This means that
y = F ( o<. ,
.

p, r,

x)

is a solution of the differential equation

~GJ+ 1-l)y - x(4'-+o<.)(,J.+~)y

THEOREM 5 1:

equation

= 0.

satisfies the differential

- 27 -

(5 .1) '

d2y/d + ( f+

(5.2)

o<+~~11-r

) 'dy/dx+ x(:~l) Y = o.

The equation (5 .1) is derived by rewriting the ''differ~ntial

equation

,J.(~+ 1 ~l)y - x(,J+ot) (,J.+ p)y

=0

Dividing (5.1) by '. x(l-x)~

, we obtain (5.Z)~
.
. .
.
DEFINITION ,5 .1,: . .T.h e differential
equation (5 .1) or (5 .2) is
.... .. :.. ___.. ..;. .
_,:

called the hypergeome'tric (or Gauss or Euler-Gauss) differential

"eqliation.'
This equation has been subj'ects of many mathematical works.
Furthermore, a con~iderable part

bf

th~se toiorks were done along

:.:the . developmerit . of ~ 'the clas'sica.1 ana1y~is

the

and' sumin~d up to

.i

general theory of linear ordinary differential equations in the

complex domain.
..

The hypergeometric differential equation is susceptible of

';,

several changes of variacles, independent or dependent.

... .
consider a change of variable
y
,~here

the

f.

11

hype ~geome~ric
:

equation

:r: .

and
.

= x f ( 1 ~x) O" z

First

are complex constants.

By ,this transformation, .

differential eq.u ati()n . (5. 2) is t .a ken to an

- 28 -

(5. 3)

~'

:t

~ [ 2f

2
2
d_ z/dx
:

'

'

'..' '

+ 2 ~+o<.;_q+ _1 ,- '

: ,

r ] ~~/dx, . '.'

',

+rP<r+r-1) + ~<6""+()(+~-t) + f<~+oi.+~+1-cr)+6"<r +r>+o<i

. l

(x-1)2

. .

x(x-1)

In fact, we .have

. dy/dx

= xf (1-x)~[dz/dx+(+ + x~l) z]

and

'

2
2
d y/dx

= xf (1-x{[d 2 .z/dx 2 +
(.:.f<t-1) . 2f~-

+\
. . ... _. .

~- .-: .

.2(~ + x~l) dz/dx .

-+
. ,

+ x(x-1)
,

. .

~(ir-1~)

_
(x 1)

..

]_ '

. ,

,..... .

..

..
.t

Inserting these into (5 . 2), we obtain (5 . 3).

f =O

or

1-)' , .

<>' = 0

or

- 0(

(5.3) becomes a _hypergeome~r_iC:,

~ .~

In particular, if we

dif(e~.ent~al equation again . ., :In other'

words, the .t ransformations

(5.4)

(5.!))

....Y

(1-x)r--~z

t-r

~: ~c. : ~ .

and

(S.6)

= X 1-~ (1-x) r-ot-A

. ~z .
1

take (5.3) respectively to

+(r-ct)
(a'-~) z = 0 ,
x(x-1)

(5 .4')

d 2z/dx2 +( + r--fl+l)dz/dx
x
x-1

(5 .5 I)

d 2 z/dx 2 + (Z-'( + o<.+fl+l.:. ~ ) dz/dx + (o{+l-() C@+l-c)') z -= 0

\. x
x -1
x (x -1)

and

ta~

- 29 -

..
'"

..these three equations have solutions

~

1,

F(1-oC, r- . ~, -. t, x),

(5 .4")

F(o( +1-1,

(5. 5")
. '

p+1-t ,' 2-;r'

, x)

~,~d

. F (1- ot 1- ~ , 2-

(5.6")
respectively.

r , x)

Therefore the equation (5.2) has the following four

solutions:
.

..

= F ( .t)( , p, "t ., x) ,
y . = (1 ~~a'-OL-~F ((}' ~ 0( ;, o- ~ , ((, x)
y ~ x 1 ~rF.(~+i- r:, ~ ~t~ r ,2~r. , '~
y

1-r
y = x

. . .
.
. ,.
. : F(l- ot , l~fo, 2-r-, x).

Y-ct-~

(1-x)
.

'~

.'.

..;

: . .. ..

Not.e that, as we showed in Theorem 3. 2, the . hypergeometric function

: 'F ( ot.. ,

~,

r,

x)

has three di ff er~nt expressions:

, F(ot., ~,"I,:

-o< .

."

x) = (1-x) . F( o<.,

"

r-p, "(,

= (1-x)-~F( "(-o<,
r-()(-A

p, >

= ( 1 "'.'X) . , : r F (, y - o(

, '(. -

"

x/(x-1))
x/(x-1))
. .

f, , '( ,

x) .

Let us consider .next the .change of the independent variable

sJ-x.
~his transformation takes the points

::.;~~[~ = 1, O,

60

= 0,

1;

oo

to the points

respectively; and the tran.s formed equation is

F( o<. - ~ ~ , .Cl<+~ +1.- '1

which has a solution

..

.. :

.~

has a solution

~o , -:-

'

5).
.

>
.

~,o(+~+l-r,1-x).

F(ot.,

'

Hence the equation (5. 21

.~

If we make the change of

variable
' y

where

f=0

:5 f(l- 5)~ z .

:It

6- = 0

and

or

or

we obtain the fol-

1- "( ,

.. 1

lowing four solutions of (5.2):

.

-~,-

('

.:: .1

~, "'-+~+i-, i-x)
. . . . . . -~- . '
-.
F.( ol+i-{,"p+1-1, ot+~+1-r,

F( o(,

l 'i .. '

x -

. r--A

rF(1-, "'(-(L

(1-x)

.'

',

~:

;,

'

r+1-ot~~,

l"".x),

l~~,. r+l~:ci._~~, 1-x) .

xl-r (l-x)r-o<-(3F(l-,
'

: . . .
1-x),

,;

'

.~

~. . .

'

;(

'

.'

Finally consider . the combination of changes. of

..

'

......
~,

:'.

~..

'

~he

independent

'.

variable and the dependent variable:

'J:GI( . .

11s , y = ) z .

x ...

Then the equation (5.2) is taken into an equation

(5 .8)

which has a solution

....z = F(O( ,.ol.+l-1 , o<+l- ~,

5)

Moreover the ~quation (!L 8) has three other solutions

(t- ~ )1.-0l-~F(l- ~,

(3--<

sl'

'(.-~, ol.+1-(J, ~)

.. .

F(~+1~r,~,~+1-oe,~), .

~-oe

r-o<.-13

c1-s> :c ':. :mc1~o<,

i-oq ~+1-0(,

Going back to the original variables, we

.

solutions of (5.2)f '

, . J::

obt~ir\

5>.

the ;following four

- 31 - ..

-o<

'
.
F(ot.,o<.+1-)',o<+l-:~, l/x)"

x~-a- (1-x)r-0(-~F(l-~' ~-~, c(+l-~-, l/x),

x-~((3+1-~, ~, ~+1 ..o(, l/x),

;,

xo(-((l~x)~-:c(-pF(l-o<) r~~, ~+1-o(; l/x)

Among linear 'fractional transformations:
"C
'.>

. :;.:;~:;r .
.">:=~ '

ax+ b
cx+d'

)'# Illy the following six transformations:

:'ts.. 9)

-:::

s = s= 1-x ,
x,

{0,1,60}

''map the set

. , . ~, ' I '

-~

~ = l/x,

s= 1/ ~1.~~), ..; ~ = . (x.-1~/x, , .J = x/ (x-1)

onto itself.

These six transformations

form
:
.;
.

.,

:,,;.

group under the composition of functions.

;~y

two transformations

1-'x

and .

.1:

This group is generated

S:0:. l/x:

:. )""

To; each of1 these six

:>:>"'

.~

:: transformations of the independerit variable c::ombined, wit.h :a : suitable

f~ I '

',:

;i.:.:.

,'~hange . of the dependent variable there corresponds a hypergeometric

:1..

i'.~ifferential

,-_ .;

;~quations,

equation.

each of which has four expressions of solutions .

.THEOREM 5 2 :

;1.hen
~!{~~(

Thus we obtain six hypergeometric differential

Assume that

1 , 'I-.- (3, o<.- ~

are not integers.

the hypergeometric differer:itial equation. (5 .1) has twenty four

:~ ...~>-:;

i'~xpressions of solutions which are written in the form

~-:&~:~;~;

a.,
1

oL.,
1

~i

arid

ri

are linear in

and

(,

is one of the six transformations. (5 . 9).

The twenty four expressions of .solutions are called Kummer's

~'fui~_lutions.

We supposed that

'(:f 0, -1, -2,

for

F(ot,

p,

't, x).

- 32 -

1, t-o<.-p, o<-

The condition that

;.

are not integers guarantees

that the third parameter in eae;h of the twenty four-expressions

. .

satisfies the corresponding condition.

It can be shown .that any one . of Kummer's . solutions is essen-

tially equal to one of the. following six solutions:

F ( o< ,
{

p, 'I, x)

p+1- ~ , 2-t, x)

xl-rF (O(+l ,

F(c(, ~,+p+1~r, 1-x) , ,

{
.

'

(1-x)J'.;;~~F( 'I-:

r-p, "t+l-o<~/3, 1-x) ,

-cC .

. .

l', ot.+l:Y_, 0(+1-~, 1/x) ,

x -pF<p. ~+1', ~+1-ot, l/x) '.

, { _x
.

Ex'erci"E;e L Ffnd all solutions of Kummer.

.. .

' : '

.:.

.'

.,

. . . ..: '

'

- 33 -

~f:~:'. 6. Monodromfrepresertt~tiori~

The

~'ifj~'.-; : .

h}7pe~geo~et~!-c

fcinction satis-

i~:;. d.~s "the hyp~rgeometric differential .equation:

f;}:~ .

! .

'...,~..'~.-.~" c6.1>

.d 2yt~ 2 +(I+o<+p+i~r )dy/~ + ~~ . 1 ~ : o

,r
--

.:.

;~i

:::T~~.

_ .. .

.x-1
..

. ..

x(x-1)
...

..

coefficients afe simple rat~onal fi.inctions whose. poles are just

. .

............

:;:fat

..

x ...

The point at infinity i.s a zero of: the .coef-

and

'

'):.ficients, but solution$ of (~ .1) are not necessaril}I'. ,holomorphic

.

~,,~;-:

..

... .

'

\'.::at x = oo ., ..Therefor~
~:'._',

'

"

.. :

H
nelghbqrhood
of .x
.:.- ::
...;,
.; ...
.
:.

'

. .'~

'

w~

must st'l,ldy the

: , .. I . -

= oo

equatio~ .

(6 .1) in t .h e . ..

To . ;do
this, we make the change of in- .
: .
.

t dependent: variable
.. ,, x ."" ..~/;

~:::.

J.~~~f1:~

we shall . investig~te . the . tran~~QJ:med

~quation.

S=. 0.

around

< The
equation (6 .1)
is t' _aken
to
(
.
. ; . .,

... 2:

!~~:'.

..

s <s -1>

..
;

o<p

\'.;!,) clearly

S~

y -

0.
.. .

;-; ..

..

is a pole of the coefficients of this equation.

;~~'ift~~~. ~. .\ ~;; ~' ; ...~~: .~

.
t . ~
we have found that the hypergeometric differential

~i\;i~hus

.
.
.}}((6.1) has three singularities at
?~~f:M:
:;_'-'':

'\;f::~omp lex

plane

};({,;i,J.,. .. .Let

x0

~~;~;~., .

{co}

x = 0, 1,

00

or the Riemann sphere

be any point

/: 0 ' 1, oo .

equation

in the extended
S.

The general theory of

,,:.~.;~/.{:.;".

~i{JJnear

;(HJ~?:~

~.::::N -"

differential equations guarantees that every solution of

= :. .

[~1:~6.l) is

. .

holomorphic at

and that every solution of (6.1) can

1r~~::~:::c:1: ::n:in::: a:o~g ::r::::e~::~::::t::: :::o:::::::

!~~~1:;C:

- 34 -

x = O, 1

of (6.1) are only at

ti0} = : - { 0,

S - { 0, 1,

and

...,;

is no~ simply connected, solutions of

(6.1) are not necessarily single-valued.

1

and

oo .

This means that

are in general branch-points of solutions.

..

fs one of

Since the domain D. =

of!) .

the

x = 0,

This fact

reasons 'why the group .. theory is useful in the study

of linear differential

equations~

Let us keep XO . fixed, and consider the set of all loops

starting and termina'ting at
can . be .repre.sented by

XO . in the domain

'a.continuous

map . u

from

D.

Such

a loop

the unit interval

.

11 into D satisfying u(O) = u(l) = x 0 ~ Let us put

I [0,

={u; u: I _.,.D,. continuous and

u(O) . =u(l) =x0 }.

We shall introduce an equivalertce-reiation "" in oC' by defining
u .._ v

if and only if there exists a mapping F

from

ix I

into

D such that
(1)

F . is continuous,

(2)

F (t, O) = u(t)

for

t E I,

(3)

F(t, 1)

= v(t)

for

t E. I'

(4)

F(O,

and
:

= F (1,

s)

'

The condition (4) implies that

s EI

for

s) =XO

f (t)

is a loop belonging to .L:.

tions (2) and (3) mean that

Therefore we can say that
continuously to

in

D.

= F(t,

s)

for any fixed

On the other hand, the condi-

fo = u
u "" v

s e I.

and

fl

=v

if and only if

respectively.
u

can be deforroed1

-. 35 -

We shall state five propositio"ns concerning the relation ,.._,

Proofs of these propositions will be left to readers as Exercise.

PROPOSITION 6.1:

The relation

Let U:S denote by

[u}

and by

1t'

xo>

the equivalence-class containing

="I~

becomes a group.

loops in t.L:.

~learly

uv

{u(2t)

.. .

v(2t:..: l)

'1f

by ,,...., )
under which

(D, x ),
0

To do this, let

Define the product uv

uv(t)

of ..c-

(the quotient set

We shall next define a multiplication in

1rl(~, x )

the set of all equivalence-classes: i.e.

(D, x )
0

1tl (D,

is an equivalence-relation .

and

be two

in this order by

for

~-

1i,

for

\ ~

~.

is .a loop in ,' . which is composed of

and

in this order.
PROPOSITION 6.2:

u' v' -

uv

if u ..-u'

and

v ....,v'.

By virtue of Proposition 6.2, we can define a multiplication

. [u](vJ

PROPOSITION 6. 3:
e : I

e u ""u

and

= [uv)

Le.t q.s denote by

-..o,

e(t)

the constant map

for

u e ""- u for every u

I.

in .C .

)lence
[e}[u]

= [u][e]

[u]

This proposition means that . [e]

is

a unit-element in

- 36 ..

' .

t. "

..,,

'7l:l (D_~ x 0 )

wi~~ - r~sp~ct . tp tJ:ie .mltt:f:i.~J.f:9a.t}_p~--'.~e.fJp.ed_;: :~bov~

-1
PROPOSITION 6.4: Let us denote by u
. t .be;,_ i9ye_r se Joop of

d~f ined
'

hy
.

-1

Then
u -1u,.., e

Hence

uu -1 ,.., e

and

..

= [u] [u -l]

[u -l) [u]
.

,,

This proposition ID;eans

~

,'

'

,,.

, .

= tel.

t~at

' ..'

, I

[u]-l

= [u -l].
[uJ ([vJ[w]).

PROPOSITION 6.5: ([uJ[v] )[wj =

:

'>

These five. _propo_~i,',t~ons prov~ that : ~ (D, x ) . is a group.

1
0
.This group is called the fundamental group of D with the bas~.
. .i . ...

'.

: '. 'J'

'

< ...

Let us go back t<;> .. the . hypeJ:"geometric . differential equa t;Lon

.\

. (6 .1).

~ .

i.

,-.

Let . ; <r

.:

. :

'

. arid ."' t

'

"

. :undam~~~al __ system of
.
.'
. ..
..
.
. .
\

'

be ..linearly independent-(~Oll:l.~-iops pf . (6r. l)

which are defined in a neighbo~h~od, U ..()f x .

0

..

.so],.ti,ons
.
..
.

it_l .~ u.

f '.:_ ::ana:~yt,~ca~~y

aloqg

~ rJ: and conti~u~

::,f _aq~,,
..

This means that

"

AS~U!lJ.e . that

u.

Then. '

and 'f' will be changer into other soluti~ns defined in U,

'f'

and

are not
nece~sa~i.~y-_, ,_.~ingle-valued.
. .
.
. .

the analytic continuations of

Since

9>

and .

'f

'

Let

since

'f'

..and .

'

and

along

u . respectively. :

are linear, ,inqep_en~,ent, the 'Splutions

are ef~r~ssJ'bl.e as .. ~ipear f()rms i~ ; , ~- - and -_

be

f :.

, ! _ :

9'
,; ;

and

- _37 , -

<J . ,-.:. a Cf -t: b 'f' " ,

>'1'

C(f + d,Y. ,

a:

which
can be writt;en as
. : ...

~~~:

where

'

'

two~by-two

is a

["]

matrix, and

are column vectors.

-.:

.' \11
~

'

'

"';

h~o.:~e~,is ~~~.~- ' :.~ ,. ~,~d, 'f':.' are . linea.~ly independent,

that ft .. and ~ are also linearly independent. In. ~~ct,

From the

it. follows

~ j'. . +c t i= O,

, i f there were a linear relation

would hold for

'J': becaus~ ~f'. the

i. ~ -. ; ~nd

r~lations

for analytic functions.

:

A t GL(2, t)).

... .

"

. :~

p~ruo/-nence of functional

: \.y .

Hence

.,.~ .

'.

. !

then this relation

must be non-singulai' :.
'

I~ is \Well know~ _that. '; GL(2, C)

(the set of

. all non-singular two~by~two ;roa.tricie'S ,with c.omplex entries) is a

'

group under the usual multiplication of matrice.s.

follows from the general theory

u ' ....,.. u and if

continuations of

'="1~'

.;;,

't'. .

-,
~

and

'",

of

analytic continuations tha,t , if

denote respective 1y the ana 1ytic

<f ahd t along

_u ,

This means that the.matrix

:~ather than

u.

Moreover i t

then .'.

.j' = <f. .an~

A ":j_5 ' d~termLned by . [u.l ;

The matrix A .'. does not d~pend ' on

the ~h~ice

of

- 38 -

representatives of . (uJ.

f :
Let

Thus we ccan d,efine a map

be another loop in aC.

analytically along

v7

denote by

f ([v]) 6

Con~ider

GL(2, C).

'f'

and

continued

{!] ,

goes to

.';.:

If

f ([u]) = A
<f and f are

then

[!]
where - B

GL (2; It),

7tl (D, xo) -

the product uv

and

the a~c_-.lytic continuations of

<'f

and

'f

along
~. :

uv

we

.. .

c - r._<(u ~ v]) = p ([u](v]).

;

and

. _a1ong . v,
~~

We have, _then,
;

..:'since
' ;

respectiyet,y.

't

are the al.'!alytic continuations of

"f and

-'I'

and since

Ti} QA[~]

.'

have

and 8= f ([v]) ,

'"' '[JJ :'~t!J'~.~s[:l

. :. ' ''

and hence

AB ,

or

f ([uJ[v])
. Therefore .
group

In other words,

,. ._,: . :2 .
on

f ([u]) f ([y]) .

is a h?momorphism of the group . 1tl (D, . x )

0

GL(2, C).

1t1(D, xO)

C.

f,

The representation

into the

is , a representation qf

is called the monodromy

39 -

.! ,epresentation ?.~ ..(6.1) with.. respe_<::t t.o .a fundamental system

~ 'f,

t} .

Tl1e image

'w ith . respect to

f (7C 1 (D,

{' ;: 'f'

x 0 ))

is cal led the monodromy group

}:

The following . theorem will be proved in Section 17 (Chapter

:L :Il).
THEOREM 6 .1: - Suppose that none of

ls an integer.

~ ,

0( ,

Y- o(

arid 't -

Then there exists a fundamental system; 'of (6.i)

such that the monodt:()my group with respe.ct to this system is gener-

ated by
1

and

1-e -2Jc:lill . ]

[ '. " e -l7ti~+/i~i)

0

and there exists a fundamental system of (6 ~ 1) such that-th~ :

:monod.romy group with respect to this system is generated by

:1

e -2n::ID<. (1-e -2'1ti(J-'3)

.
. )
.

and

. e

,.

-21ti (ot +~ - )')

..
.

..

- .f.O

CHAPTER II
.

: ' ' .i :,

.:The :Hypergeometric Functions of Tw~ ' Varlabl~s ;-.

7.

.. ! .

'

. .1

~l

.. . ~

; :

Definition of the hypergeometric functions .. of two

variables ~'

In

the prec._e ding _c hap,t er, we. sta;te,d.. s~veraLproper~_ies of the : hyper~j

J tt r~. : . ~ ."'

.1

tial
equation.
'
:.1

: '.

~-:

:'j ;:.

F ( ot,

geometric function

~hall
.
I '

We

, J', x)

differen~

and the hypergeometric

now
proceed
to
'
.
., -

~he

function ~

hypergeometric

of . twq
va,riables. w~ .. st.~rt .with .two.. .hyper.geometric s-eri:es1..;. '
f ..' .: ~<! ., ..:..;~ ~-. ~ .j . ~ :,
. ' ..."' .,.

..

and

F(

'.,. (l(,

f3,

,' I.' y

f. :

(pt.'. n)

m=O. ()'',

CJ'

I,

n)

n .

f> (1-, n)

~ , ~ ..,_....

Consider t~~ . ~r~~u~t ;_ 9 . ~hese .~o-. functions:

!

(7 .1)

F( o(

~ d, .

x)F <--

"!~" t" ;) ; I: f~':n~ >~Hf ::;2t~i'~ixmyo

. , . ..

::

'

. .

m, n=O . '

::

Thi.s is '. a double p~we~ series in : x, ,y.,

however~-

can

'~ot: .b ,~ '~e,garded r~ally

. l ..

'

T1his ~~~~~~~~~ of

as a function of

since this is a product of a function of x

: '
'

and

two ~ariables,

and a function of

y.

To obtain functions which are really regarded as functions of two

variables, we shall modify the coefficients of the general
this double power series .
f

To do this,
( 0( ,

l~t

m) ( c< ' , n) ,

us consider the three products

( ~ , m) (

n) ,

.( 'I

, m) (

'(' , n)

and assign to these products the three quantities

~erms

of

Y.' i

- 41 ;..

(' , m+n) ,

( ,./1 . , mt-ri)
,
,. ::,, : .~ c' 'l. m+n)
' ' .'

respec t:ive ly, .,i.e .

m) (

( o( ,

( ~ , m) (
. .I

'

' ' , n) -t--+ ( o< , mtn) ,

j3' ,

n) ~ (
.

m+~) ,

(3 ,

'

..

. .~

..

,r'.

( y , m) ( ' ' , n) ~ ( 'I , mtn)

' .

: '

'

..

' J ;

.If we replace just one of the three produe:ts

'

,l

',:'

~'.-:' ~

: ,;

~Y.

':

the corresponding
: :

..

'.:i \

A; 4 :

o .,

' <

'.~

'

quantity, we obtain from the double series (7.1) the following

,.
three double power series:

f .. (Cr, m) .( 1', n)
o< , m) ( o<

: m,n=O .
:.

n) ( ~ , m+n)
m n
(1, m) (1, n)x Y '

(7 .2-3)

(3 , m) ( {!> 1 , n) m n
, m+n) C.1 , m.) (1, n)
x Y
.
~- '...J,_ -. ~

( ot.. m) ( oC. ' , n) (

.m,n=O

~ . i .;. :~~ .

'

;:

'

'

, ''

,'}

."'.~

-:-

Replacing just two of the three products by the' corresponding

quantities> we obtain the following three double power series:

.< _
.,.l ~
.'
r .; ( ~ J1rtrif(~ ,~rt), j ~
*~~n
1
m,n=O. ( y ,m) ( " ,n) (l,m) (1,n) .

t
.

(7.3-1)

..

.I '.

_ ; _

,. _

(_

(7.3,-2)

(7. 3-3)

....

~ (o< \ m) ! oc '.n) ( (.3 ,m+n) m n

~
( ,.m+n)(l,m)(1,o) xy
m,n=O . .0,
., .

The double power series

>

~:, ~:

.. 42 -

OQ

"

(7 .4)

~O (

( Q(.

m+n)(

n, m+n)

t, m+n) (1, m) (1, n)

xm n
.Y

is obtained from (7.1) by replacing all three products by the

corresponding quantities.
Among the seven series, (7.2-1) and (7.2-2) are essentially

the same, while (7.3-2) and

by the symmetry on

(7.~-3)

the parameters.

are also essentially the same,

~

and

o<.

On the other hand,

rearranging the double series (7.4) into the series of homogeneous

polynomials, we obtain F( ,

~,

( ,m+n) ( (3 ,m+n)

"/ t x+y).
.

xmyn .,.

m, n=O ( a', m+n) (1, m) (1, n)

In fact,

L L
~

k=O

(o< ,rirl-n) ( (.Lm+n)

m+-n=k ( ;r ,m+n) (l ,m) (l ,n)

xmyn

60 .

:L (o<, k) ( ~ , k) 2:,

11;1.

(1, k)
m f:f
k=O <i,k)(l,k) m+n=k(l,m)(l,n)x Y"'

Observing
(l,k)

(l,m) (1,n)

(mk)

if

m+n

= k.

we get

(l 2 k)
mn
m-1-nk (1, m) (1, n) x y

(x+y)k .

Therefore

Thus we arrived at four power seires to which Appel gave the following notations:

- 43

The power series

. ' DEFINITION 7 .1:

F (ot, ~, (3', y, x, y)
1

F ( fJ.
2
',. .

A'
'I"'

'
(
. F 3 o<,c<.

F4

'u'

m,n= 0

"',x,y)=
~ . (o<,m+n)(~,m)( f3',n) m n
0
~ ( ) ' ,m)('1',n (l,m).(l,n) x Y
.m,n=O
!

,p, ~ '"'' x, Y

..

' , 1-A' .., ' '1

(o< ,m+n)(@ ,m) ( #' ,n) m n

( }(, m+n)(l, m)(l, n) x Y

= -2,_

1 '

' x, y

~
(c<,m)(o<',n)(a.m><' .n) m n
... ~O (y,m+n)(l,m)(i,n)
x Y
m,n=
.
~

(O(,m+n)({3,m+n)
mn
( <1
y'I , n,
' (1 , m) ( 1 , n )
x y

= "--" ( v m)
m,n=0
0 ,

~te cal~ed the hYJ>erg~~fi:letric series of two variables and the . func~~ons
,C~f..

defined by these series are called the hypergeometric functions

two variables.

We assume hereafter that

'.

r + Q,

"'( >

-1, -2,

... ,.,, -.

:..~ :

. Remark 1. . F

:Hl,teger.

F1

is a polynomial if

is also a polynomial if

:~~~ative integers~

:~{9.~ -1, .. 2,

,,. .

, 1.~,:::r,

,; ~;;'.@ r. and

..\J\

.~

j:

~._/

')

= O,

2
3

is zero or a negative

tX

and

..is . a polynomi~l if

0(

is a po~ynomial if

-1, -2,

F4

~ '

are zero or

or

/3

and

( c:<. . and " ~ ')

is a polynomial if

~ ')

or
tJt

:{:'~\ c: O, -1, -2,

~~....f:-'.

Remark 2.
F3(c;(,

!i.1m

o<,

p,

is symmetric. in

'.~i:~;(is symmetric in 0(

p,

(3

and in

c<. ,

p':

(3', :i,~, -y) .~ F3(~, ''~c<. , ~', 1,x,y)

F4 (<:<.',

o<. ,

~ F

3 to<, ~,~)ct', "1',x,y)

3 (p,

p',ot, al',

o--,x,y).

"t, ~, x, y) = F 4 (~,ol, 1, '( 1 , x, y).

or

- 44 -

U_s ing the

Remark 3.
2, 3

and

..

'

F2 ..

';

"

x y
'

..

rcr >C< r' > ~

rcoe.+m+nl CC(?+m) re~ '+n)
r (oC.) re rn r< (3 I) kn=o r<r+m) r ((j' '+n) r<1+m) r(l+n)

.; ' :. ....., :

= 1,

r<O(+m+n)r(@+m)r(~'i-n) m n

..

. . , . , . . i .,_ . ,.

> .. ......

r .(o() C(~ ')

r<t> f(

:, ~

r(~_) r<(3') m~=O

',

xm n
y

:: : .'. ., :: . . . :. ,, - ...; .: . ..

r<t+m+n) C(l+m) r(l+n)

x y

'

') ~

=.. r(o<)r (I~-)

b
I"' m,nc:O

, Remark 4.

. .. . . . ..'

. ~ r (t:>t +m) r (d.. '+n) r (~ +m) c(t? '+n) ;fu

r (;()

. F4

-~

r.cO() r (~) r (~I) ~=O r< r+m+n) r(l+m) r(l+n)

"".
-. '

we .can write F j .. (j

respectively as

4)

r()

~l

r "."'function,

C(o<.+m+n)r<,s+m+n)
m n
r<r+m)r(t'+n)r(l+m)r(l+n) xy . , ., ,

Rearranging the hypergeometric series into iterated ;:

'

'

.;

. .

'

series, we' obtain, for example,

,I

'

.= i:o ( 'J

~ (ol,n)(@ 1 ,nJ
;n) (1 ,n) .F.(

In fact, observing
(a, m+n)

'. =

o(

.
n
'47n, ~, +n' x ) y . <

'.

a(a+l) (a+m-l)(a+m) (a+m+n-1) -

we get

(a,

m)(a+m, ;n)

t.

(ot ,mtn) (@ .m) ( ~ 1,n)xm n . ;. .

(ol ,m) (@ , m) (ot+m,n) (@ ',n)xmyn
m,n=O <or,m+n)(l,m)(l,n) Y
m,n=O (,m)(l,m) ( +m,n)(l,n)

!'

~ {cl ,m) ( t3 ,m~ 1,

(3' v+
) m:'.
....i(y,m)(l,m F,c(.+m,
,Q
m, y x .~
~o

..

- 45

similarly, we get the second identity.

: F (o<.

. :.:. 2

ar y

' r ' 1~

11

r'

x,

For other Fj

we have

y) .. ~ (c<,m)(f,m)F( +m .R'
...... . ( v m) (
m)
0(
' r
' l
mO /1 '
,

.,,

(oe,n)(~'.n)

. F (et . ' A, A', v, x, y) .. ~ (O(,m)(~ ,m}F(o<. . r

a
~o ( 1 ,m) ( ,m)
, p ,
. 3
, . , r I"

(o(' ,n)

oO (

x, y) ..
:: 1: .

2_

(ot ,m)(

<f' ,n) F .(

,n) ( ,n)

d..

(3 ,m) F(o(+m,

mu:O ( (1,m) (l ,m)

, ,

. . ..

~O ( , , n) { l , n) F ( " +n , ~ ,

y)xm

r.

t, x) y

r+m, y) xm

p , y +n,

~+m, "1',

x)yn

y)xm

~ (cl ,n)(s ,n)

n
(
, ) 1 ) F ( d. +n, A +n , 1 , x) y
.
n=O . ,n
,n
r

L-

.....
. Remark 5.

If we give one of x and

y the value zero,

or if

:,~ertain

parameters take the value zero, then the hypergeometric

~ eries

Fj

are reduced to hypergeometric series of one variable.

:ror example:
Fl (

' , ~- , p' , 'lf, x,

0) ""' E' 1 (cl. ,

. ..., F(,

F1 (o<.,

p,

~, ~,

0, y)

Remark 6.

~,

= F 1 (o<., 0,
= F ( '- ,

:.1

(3 ,

0, y, x, y)

't, x)

~,

p'' '(,

'({,. x, y)
y) .

If certain parameters take special values, the

hypergeometric series

F.
J

are reduced to known functions.

For

- 46 ..

example:
F' (o<.,
1

~, ~,

o\, x, y)

F2 (

~, ~'

~,

Exercise l.

o( ,

p' ,

= (l-x)-(3(1-y)-~

x, y)

= (1-x-y) - .

Prove that

r 2 ( o(. ,

~'

I t

~, ~

> X>

Y)

(1 -

X -

Y)

-.

and .

x F 2 (1, 1, ~', 2, ~' , x, y) + y F 2 (1, ~ , 1,

(
c

..

log 1 (1 -

-1
lC, -

y)

p, 2, x, y)

s;

Domains of convergence of the hypergeometric series of two

variables .

..~i

: '

In this section, we shall study domains of convergence of the

i-: .

and

four double series

To do this, we begin with

s everal remarks concerning double power series.

First of all, consider a series
.:

(8 .1)

;:

As it is

~ell

known, the series (8.1) converges to a certain value

....
j "

(or the series (8.1) has a sum

if for

..

p
a-2:,'a
<e
m= 0 m ..

for every

if and only

such that we

> M.

oq

' N

This definition implies that if

2:

m=0
sequence

2:.

m=O

> 0 there exists a. posi.tive integer

:: have

ol

00

Ja l

60

\ mlm=O

is convergent, then the

. am

is bounded.

Let us consider next a double series:

44

L:

m~n=O

m,n

;. The following definition might seem to be very natural:

"The series (8.2) converges to

if . and only if for any
M and

>0

(or

(8.2) has a sum

a),

tJ,lere ex:ist two positive i ntegers

N such that

laL, a J<.
I O~m~p , O~n =':z m, n

for

>M

and

> N."

definition, however, does not necessarily implies that the

- 48 -

double sequence

{ a m,n j. is bounded.

a.

m,n

. i

~f ~:

For example, if we define ,

for

for

for . n

-,
0

. .. .

. .

..

. .

_.

>1 ,

00

.....;- a m n 0 in the sense of. the de~inition given above.

L.-..
m,nO
'
It is evident that { a m,nll is not b_o,..._unded. Therefore, we shall
then

use the concept of absolute convergence for double power series,

instead of the definition given above .
Let .
Q

CIO

L,_

a
xmy
m,n ..
m,n-0 . . .. ,. . .

(8.3)

A' the set of all points

be a double power series, and denote by

(x0 , y 0 ) where (8~3) is absolutely convergent.

interior of A.. It is clear that if
sequence :{ am,nx~y~}

Let

D be the
....,

then the doub ii'. :

. ..

.. .

is

bound~d~

.;.

.;,,

We shall prove the following

p _r oposition:

PROPOSITION 8.1:
and that

<

Suppose that \ x l

< 1. Then,

if

and

m n}
{am,nxOyO

are positiv(

.;~:I

is bounded, the double

power s eries (8.3) is absolutely and uniformly convergent for

Proof:

..,

By the hypothesis, there exists a positive constant K :

..

such that
(m, n

= 0, 1, 2,

) .

 49 ..

.,
'

. Since

I = llam,n xom Yon I lx/xo

Im :1 y/yo
1n m+n
. cf: K 9

t
m,n=O

K em+n

is convergent,

and uniformly convergent for

2._

m n

x y

is absolutely

e lxof, \ y l "' e1 Yo:I .

U of. (x~, y 0 ) which is co'ntained -in A.

(x , y )

V of

...and uniformly convergent in

. ::

' ..

. .

(x 0 , y 0 }.E

From this

there exists a

such that series (8.3) is absolutely

,;,

(x , y ) E' D, ' there exists a

0
0

fact it follows that for any point

. neighborhood

m, n=O
. . . . m,n ..

. \ x\ ~-

Observe that for any point

neighborhood

.,,

mn
Iam,nx Y.

.,

V.

The domain

D is called the domain

:"

. .

Let us go back to the hypergeometric series of two variables

and prove the following theorem.

THEOREM 8.1:
.

Then

Fj

S~ppose

that

(j .:: 1, 2 f 3, 4)

c<. ,

~,

o( ' ,

are not polln_?mials

F
F

The do1Min of convergence of

'(8 .4)

i~

x, y

2 ""

f (x~y);

jxj+

IYI < l}~

= { (x , y) ; l x I < 1 , I y l < 1 ~ )
{(x,y);

{i";;,\ +.jfif < 1}

We shall use the well known formula:

(a, k)' "" _1'--li
m
(k-l)!ka
k .... ~

r<a)

oor
'(8.4')

~~~

=. { (x, y) ; l x I < 1, j y t < 1 } ,

=
Proof:

f3 '. , ?f , 6 , ,; o, -1 , - 2 ,

(a, k) ,..... (k-1) ! ka

r<a)

as

k ~

" .

- 50 -

Case F :
1

Put

'

:.~ '

( 0(

Aoi,n

,m+n) ( ~ ,m) ( ~ ',n)

'J , m+n) (1, m) (1, n) :

It follows from (8.4') that

'

.. '.

i:<~)
:
~

m, n

(n-1) !nP'

.rep')

r<~)
. !.

m, n ~ 0o

The ref ore

.... m.i;i
. .-'.:
. -

. .

whe~e

(m+n-1~ ! (_mtn)'i .m!n!

rCt)
. '11.-1 ~-1 ~ -1
r(c".)r<p>r<~') (mtn) .m .n
.

I A. . I~ K(~~)1-ri' m(Sl-1 /i~l

.

1.

C<t>

oo,, . and hence

Am,n as

~ (~~-1) ! (m+nt< (m-l)!mP

m,n
as

! '. 1

. .

K . is a positive constant

~' = ~e( ~ ')

and

and

1 .. , '::.-' . ' .

for
:.

,., .~ :.,:
: . . . . . ..
.

'

c(

m, n .. 0, 1, 2,

l .. R~ (

..

'

.: , ,.

otJ ,

This shows that

.

.f

(8.5)

. , .

is absolutely convergent for

Ix I< 1, ;IYI ~1:

there exists a positive constant

1{'

K' such that

. p -1 {3 '-1

: '. .oe -r
(m + n) 1 . 1 -qi 1

. , ...

On the other hand,

4'

n 1

. .

I Am,n I

for . m, n = 0, 1, 2,

the double
.

Therefore, the series (8.5)

is not abs~lutely Conver.g ent . if;
that the domain

IYl<

. : .. :

of convergence of. F1 is

li.
Case F 2 :

'xi') 1.

Put
. . ..

or

IYI) 1.

This means .

{Cx,y); Ix I< 1 and

- 51 -

( Cll , m+n) ( ~ , m) ( ~

w,n

n)

~,m)(. ,n)(l,m)(l,n)
1

.ff...,follows from (8.4 1 )

.:y .~ .

<~-l)!n/J'

~ (m+n-1) ! (m+nyi< (m-1) !m~

A .

r()

m,n

r<~') '

re~~

f(ir)C(l')

r<'>r<p>rcp')

C(ir}

C<t')

{m-l)!m1 (n-l)!na''.m!,n!

(nrl- )o1.-1 ,_r fl'-i' Cmtn)!

n
' m
n
m!n!

and hence

for

m, n "" 0 , 1, 2 ,

a. pos~tiye number

tr

_,

where

is .a 'positive

constant. , Take

greater than

Then we have

Therefore

~e

1Am,n xm Yn I '

K 4 -r (m+n)2cr~oe1.;.l (m+n)

. m!n!

l xi1ml YI n

get
~

'

2=_
m,n=O

.
'
' ' 00
26'+-i. -1
k
nxmynl ci!, K4-IJ"' ~ k . . (lxl+) yt) . .

'

lAm
'

k=O

:i,~ follows from this that the d~main

{<x,y); Ix t+ IY I< 1}

is

. ~'.ontained in t~e domain of cortyergen~e .~ F 2 ~ . To show that this

. .
.
'~. ,\~
~Qmain is. exactly the dc;>tru;lin
co~vergen,ce of F , it is sufficient
2
'

'

'

'

of
I

- 52 -

to prove that the double sequence

lxl

+ IYI >1.

.This will be left to the readers as an exercise.

This case can be treated in the same way as

Case F :
3
of

fA
xm yn l is not bounded if
l m,n
1

F p :_ Hence the proof of the theorem in this tas~ "wiil

'

the case

_.be left :

"

to the readers as an exercise.

~

F :
4

Set
=

m,n

( o<. , ml-n) ( ~ , m+n)

( ~ , m) ( 'J' , n) (1, m)(l, n)

In the same way as in ,previous cases, we have

,..... r(;y)r( t')
m,n
P(o<) re~)

(m+n)ot.+~-2ml-~ 0 1-1' (<:r~~!)

Hence

I Am,n I
where

)2
' K(m+n) o(1+P1 -2 m1-Nl n1-li ( (m+n):
.
m., ,
..

n.,

is a positive constant.

'

'

, .

Take a positive number

that

o ~ max { 1

- (J 1 , 1 - .o

l }.

Then

Using these

estin~tes,

we OQtain

+ ~ +2d"-2 .(

xmyn' ~ Kl ~a- (m+n) 1

A
\ m,n
.
O{

'

Cons ider the estimates

(~n),.

m. n.

)2

lxlmlYtn

so

'

- 53 - .

, and

. .1.

~( L

(:!n2/
\xl~lYl~ 0 )
m+n=k

... .JiXf + JG"f >k>2

.Therefore we obtain
00

2. IAm n xmynl
m,n=O

-~ ~

-& K4

'

o(l+f.l.+2<>-2
'"l
c{IXI

k=O

The series on the right-hand member co~verges if

This means that the set .

{ (x>y);

the domain of convergence

o~ . F

. seen that the double sequence

Jfi\ + .JTYl

> 1.

~+ [\Yl .<

4.

+JfYj)

1}

2k

JTXT +. JTYT <

1.

is contained in

On the other hand it is easily

xm n l
A
{ m,n Y J

is not bounded if

Thus we have
.

'

conv~rgenc~ of ~4 :;

.the domain of

i (x >Y?,; JTiT + .rrYT._< 1 J.

As in the.case .of one yaria~le, the hypeigeometric series of .

two variables : m;e Newton series with .r:espe.ct . to . th~. parameters
of. ' ,

'( '.

Thus we obtain the following theorem: ..

and

THEOREM 8.2:

and factorial series with respect to

't

C(

and

The hypergeometric
functions
of two var iables
.
.
!
.

. are meromorphic with re spect to the parameters . a nd _their poles a r e

at

'({ .= 0,

Exercise 1.

~1,,

-2, ,

and

V .'

Q. '

1 ' :- 2 ,

Complete the proof of Theorem 8.1. .

- 54 -

...

9.

Contiguous functions.

by

are obtained
one.

The contiguous functions of

F(

o( ,

p, '{, xf

increasing or decreasing one of the parameters by

In the same way, we can define the contiguous functions of

the hypergeometric functions of two variables.

DEFINITION 9.1 :
functions

The functions obtained from the hypergeometric

(j = 1, 2, 3, 4) by increasing or decreasi~g one of the

Fj

parameters involveq in

The function Fl

by one are called contiguous to

Fj

has eight contiguous functions:

x, y)

F (o(~~+l, ~, '(,
1

x, y)

F (0(+1,~, ~',

F 1 ( , ~ , ~ '+1, !> x, y)
'
F (~,~,~,~+1, x, y)

The function
~

F
:

Fj.

F (<>C.-l,p,
1

'

~'' r,

x, y)

'

~'' if, x, y)
Fl(~ ,13, (3'-1, i, x, '1> '
F (ot,~-1,

..

Fl(o{,~, (3''

1-1, x, y)

has ten contiguous functions. etc.

'

It was proved that any contiguous func tion of

F ( o<. , ~, , x)

is expressible as a linear combination of F and its first derivative .

We shall prove an analogous theorem for the function

Fl(o(' ~' ~''.1' x, y).

THEOREM 9 .1:

We have

F 1 (oe+l, ~'
.

~Fl(o<,p.f-~, (3 1 , 1,x,y) = ~F 1 +xdF /dX,

~'F 1 (, ~; (3'+1, ~,x,y)

. (~-l)F 1 (o<, ~,

!1' ,

1-1,x,y)

(J-ot)F 1 (oc-1, p, p', '1, x ,y)

(3'F +ydF /Jy,

1
1

(!-l)F +xdF /ax+y;)F /ay,

1
1
1

(1-iX-px- p'y)F +x(1-x)~F /dx + y(l-y)aF /()y

1
1

'i

- 55 -

_')t-P-13' )F1 (, p-1, ~', '/ ,x_,y )=~t-{3- fo.' -x)~ 1~ (1-x).JF 1 /ax~ (_1 -y) JF 1 /a y,
'

..
. (t-~- ~)Fl(, p,(3' -1, )",x ,y)= (1-p- ~' -!Xy)F +y (1-x);)F /dx+y (1-y )aF l /;:) y,
1
.

(1-o<) (1-~ -pi )Fl (oe._,p'

I ,

r"+l ,x ,y)= (1~o<.:-p- fl' )YF 1+1(1-x)aFl /~ x+ (1-y )a F 1 /a y.

Sketch of the Proof:

Set
00

---.

=2._

m~ n

m,na::O

m n

xy

where
A
m,n

o< ,' m+n) { ~ , m) ( ~ ' , n}

, m+n) (1, m) (1 , n)

Clearly we have

,
. (;iQ

__....
m n
y 'a F' I a y = L_. n A
x y
m n
1
. m,na=O
'

Observe that
CQ

/.\
A'
y)
F 1 ot +1 ' I"' ' 1- ' 'I ' x '
(

= '--'
~
m,n=O
00

F {o( ,~+1, (3', ;)', x, y):: ~

1

m,n==O

(9 .1)

Fl ( o< , (3, (3'+1, '1, x ,- y) =

2:

m,n=O

+: +

nA

m, n x

{2 + m A
~

(?'+n
t.l'

1..

m,n

m n
Y

xm n

'

m n

A_m,n x y

flt)

Fl( CJ(., A'

1-

(3', y-1,

x, y)

-=

2:. . 1+m+
n - 1
- 1.

. m,n=0

A. . xm.yn
m,n

Using these identities, we obtain the first .four identities of

Theorem 9.1.

- 56 -

we

sh~ll next , prove the , fifth identi.ty.

'

: , .fl-~ Ot. - ~, .
~- '
, . . . . . . . ..

p.'

; . . ' . . .

DO

. .

Ob~e~e" that

oC-1

, , m

'6. ., x ~ y). =.. m,n=O

. 2,~ ()(.+mi-n-l . A
. m_,n ~x. y

On the other hand,

. ~ : ;. '

.. .

:(

00

m n
~
'
Am-1 n x y
m,nm:Q .
'

wher~

,. ( o< ,uri-n-1) ( ~ ,m-1) ( ~' ,n)

( ~ ,nri-n-9 (1,m-1) (l,n)

m-1,n
..

( r+m+n-l)m
A
(o< +mtn-1). (f.> +m-1) m,n

and

A-1,n O
Hence

Similarly

2 ""F I
Y " 1 C)y ""

..~ .

( 1+m+n-1) (n-l)n A . xm n
~ ( o<+m+-n-1){ A'+nl) m,n . :,Y.
m,n=O
1-

Inserting these formulas into the right-hand member of

'the

fi'fth.. ; ~

identity of Theorem 9.1, we can verify this identity by somewhat

- 57 -

teiigthy computation which will be omitted. 'he othe1 three

i.dentities can be proved in the same manner

Comp ~ete the proof of Theorem 9 ~ L

Exercise 1.

The eight identities given by Theorem 9.1 mean that every

_contiguous function ,of . -~l

of

~l

can_. be ~xpressed

as

a ~inear combination

and its first derivatives whose coefficients are rational

and y

in x, y, o<, ~; - ~

Eliminating the first derivatives

from any three identities, we obtain a linear homogeneous

between three contigtious functions and

'
It should be. noted

F
1

that there are cases when the first derivatives can be

from two ide[ltities.
.- relations between two
9.2~

THEOREM

elimin~ted

-In..: such cas.e s, we obtain ti.near homoge'n eous

~ontiguous

functions and . F ,

F
",

1
:

...

Three contiguous functions and

_c ontiguous functions and

Fl'.

or two

' '

.: '

'

.:

; ,:

,! '

...

:..,

o(, .

p,

' :

, : ...

For example, we obtain

Fl (Q(+l, ~, ~, 't ,_x, y) -~F 1 (ot. ,~*:l, (3', , x, y)

.

- .(3 'Fl (

! .

' , ~, ~ '+1, 0 , x, y) - (Cle - p- p'?F1 _~ ~ , r, f3' ~}, x, y)

( - C{ )Fl ( ol , ~

' ; ~ -1, f.;' , 'I , x, y)

.
,.+, )' (1-x)F1~0( :> ~, p, '{, x,)) :;. o .

, ~' ,

.)

Exercise 2.

1+1, x, y) - Fl (
:

'

Complete the proof of Theorem 9.2.

(3 , '(, x,

Eliminate the first derivatives from three or two

:, 'identities of Theorem 9.1.

'

are connected :by a line.a r _homogeneous

.

<.relation whose coefficients are polynomials in

Proof:

r~lation

= 0 .,

-~I

Observe that on the right-hand :membe'rs . of n(9."l) the indices

and

are in the numerators of the coefficien,ts ..'.of power series .

On the other hand, .on the right-hand members.- .o f the foll.owing

formulas:
. . '' ;.

"

,..

P..
,
F 1 ( a(.~1 , r ' ~

00

' 'I ' x ' Y ...

o<.-1

L-

m,n=O

ot.+mrn ..;.1

.: .: . ,

,l

..

A. . m n
~. n: x Y

~
/31 A
m n
a+m-i m nx y
m,n=O I"
,

Fi.(o<., ~-1, (3, ,x~y)

-,.

(9.2)
Fl (0(., ~ , f>'.-1, r ,- x, y) =

'

00

2:.
m,n=O

/3' -1

f.>' + n - 1 m, n

xm n

Y '

. .!.:-:::.

...

(Ol, ~, ~, 1+1,x,y)
1

. . .. .. ,F.

_;

the indices

60

't
m n
A
x y
Y+m+n
m,n
m,n=O

m and

..f\: .

are in the denominators of the coefficients

of power series.

This fact was one of reasons why the. last four

.

..

formulas of Theorem 9.1 are more complicated than the first four.
.

..

" ,

The same fact may be observed also for

and

F ,
4

and we

can prove the following theorem:

THEOREM 9.3:

The contiguous functions of

I

with

o(+l,

oc.'+l,

~+l,

..

F2 , . F3 and F 4
are expressible as

'+l, 't-1 or

~-1

linear combinations of the original functions

Fj "and their

...

derivatives

iJFj/ ~x,

'dFj/ ~y.

For example, since

60

2:

. m,n=O

we obtain

c<+mt-n A
C(

m n

m,n x Y

<

m~

- 59 -

..

a.,

o<.F2(oi..+1, ,...
a.,

x:

'!i',

v,

,..

11

y)

.JF
.....
2 + x r;I"F 2 I"\" x + y q.,.F 2 I

1:11

... y

Eliminating the derivatives from the recurrence formulas of

betw~en

Theorem 9.3, we obtain relations

~

..

'...

the original functions.

'

contiguous functions and

..

For example, we have

;~1.L ot.F 2 ~ _0(+1, ~' - ~, 't, l,'t<x_~-~) ~.} .f. 2 C~-'~ -~l, ~- ', ,~_>_. a-'_ , x, _y)
-~'F (~, ~ p'+l, , ~',x,y)- (oe;,.p-~')F 1 Co<._,_p, p',

otF 3 (at+l,

ce.' '. ~' ~ .' )'~ x,

1', :JC y)

otF4 (~+1,~.' - ~~
.

'

..

ce> ~,~-',(. JC: .'f)

~ ~.F4 (_o<,~+1,

';

..

' .

. '

'

~,x,y~ O

t,

..

,.'I, x,

.. - ~

'

'

'

y)

Y_,.t, x, y)

.. . (o(. -(3) f 4 (ct, '3~ y, t' ~ x,_y) . ~- 0

, '

y) .- ~F 3 (o{, oc' ,~+.l, (3'

(ct~ p> F3 (oL,

~.

'.

. . - . "'-

..

'

..

.,, ,

For other contiguous functio~s of . F , . F3 ~ " and .F :....w!iic~ are

2
4
!Ir. ~~~ain~q

ing

by d~~li~as,.ng . o(

't . o~ . . 4 '

. ,

.tJ.t, :f3 _, or

._. ~ ' ; ?Y qp~, or -.: l;>y' increas-

~y one :

.,

. ...

..

..

.:'::' (9
3) " .
.
'

'

the situation is not so simple.

. :

'

;,

'.

'

,.

~ II

'

< :', '

~:-~I

_'I

. : .".

3Fj

~.

F j . .d_on,n~e _ .s '. exp.ri~eqt

.. , ,

lineairement au moyen de cette fonction

partielles

... ..

. . ! . .

'i

~.

Appell and Kampe de Feriet claimed

.
'
. j .. '
.. '
.. ' .
in their book that
'.
,.
(1.:
"tes f onctions contigues A une fonction

~{"[::,.~,

..

F.
J

et de ses derivees

I .ay."
..
'

Their statement is certainly tX'l:le for those contiguous

functions
. :.
.,i. '

in Theorem 9.3.

However, their statement is not necessarily true

:1

. .~ >~

- 60 -

The author does not

for the conti.g1.1Ql1S functions (9.3) .

k~.ow
'

whether the recurrence formulas for those contiguous functions were

:

: , .

i:. ,

already disc.overed.
.

.. . .. :PROBLEM 1 :

functions
:

'

..

~ecurrence

Find the

formulas for the contiguous

(9~3).

' :

'

'

'

In fact, the author

;- '" 'This 'p roblem mlght not be very difficult.

fou~d

the :follo~i.ng
~

thre~; f~~~la~ fo~ F2 :

, ;

: ..

..

..,.

.(at-t)(o(- r')(C(;..1- i'+l)F (oc.-l, ~, ~', 'l/,

2

~+1){. <0( ~ 1)(oc' - {'f+ (~ :. '~. ~x+ (Cit-> p'y}F2

.. (cl -'t-

I '

t'',x,y)

. -

{ c ' - 2'') cat-

(f-

:.

~.

~+1) (1-x) + c2o{'.

...

~...!

I~ \i

:. t

~ -:~

:-.

-'..

.:

- a'')

l ..

'

. - ..

. .

......

piy}x aF2/ d x

?;,, r ::.::, i , -;1_ , ;

'

."

t>px}yaF 2/'dy

._ "." {<o(r><oc:-1-1'+.1 )(1-y}+(2oe ~..1~

. ..... + ;.r2o< :r.~ ~ l')(l-x-y)xy ' ~ 2 F;i'/'ax ()~ /

..

~: ~ ~)F 2 (oc; ~-1, ~'," "' ' ~',x,y)= c-r-e- ocx>Fi~xci-~> aF~}ax
,1
' ' - xy if. /;J f ,
. 2
,:

(. yU ' - ,..
At

)F2 ( OC ' rP.,' ... I""-. I -1,

'V
t

,~I , 'X' ,

y )" ., . ( .'t i"- ~' - ct y) F 2

xy

: ,:f,

'

. '

~F~'/~x+ y(l-y) dF1_/ ;i~/ .

.. '

It seems to the author that the . other recurrence formulas as

. . ..
:. :;.: .c. ~~~:-.
:1

: .,, ; . "

,;',

. . :

~1

::.

...- .

~ ....

. ,;

-~

. . i f.'

, ..~ ..;

~ ):

,; .\..,.,: j

. :' . .i

. !: ~:

..., ..- ~

well as the abo~e formulas wiil be derived by careful computations

~;

... i) j .. :..1J_J ::.. ... _:, -~

~ .. ~ j

:. ~ ~.

together with inspections and that each of the contiguous functions

. ~;" .;.;;1.": :::y . .. '. .~.:.

:.. : ' '

r.

."I .. .. . ]'..~~.'~ - >.-~::.:

'1::

{: .{::; '\~ ;~:; _:,~.:-:. J : ~ !' . ~

.'

". ~

::_.

:.

(9.3) is expressible as linear combinations of the original

.
. ..function

. :. : :... .::

Fj,

-:. .-.. :.. . .

~:

its f .i rst

. .

-~

. :. I'.:.~ .-: ~.~:.:.

derivativ~s

second derivative

. ~ Fj I

;:. :: .1 . : " ..'

ax a.y .
4

~~B~EM 2'{ i Fi~d ~~y~~~~~:i.~

~-:

:-,.:.> :.

~:. b

'dFj/ dX,

:::!.i :

>

-~

deriving the recurrence formulas.

'

a,~~~a; ~~~- -~~~. ~~~ed

.:

(or intrinsic) method of

- 61 ..

~ib.

Euler integral representa'tions.

.-.-

The hypergeometric functions

:g.f two variables, except for .F ,

have Euler integral represen-

./~ations which are double .~ntegrals but very similar to that of

":;~_t.1'' ~ ~ , i , x >.

THEOREM 10.1:

We have
..
' .
'

fJ u
.
u, v, 1-u-:v~O

re~>

..

.~ r(~)rc~) ro-~- ~)
1~

'

'

' ;'j... :,

:,:' if

F 2 ( a(

. . .. '

Re (

' .. J . i;

i-~-p'~1

(1-u-v)

Re ( 'I - ~ - (3 ')

~ .'' .
",.j, I '~

::. :1

(1-xu-yv)

dudv

: '.1,f. . Re ( ~ ) '> 0, . Re ( ~ ') "> 0,

:Jno. 2)

--1 ~-1

fl ) >

0'

p , ~ ' , r ' 't

>

0;

x ' y)

I ,

.! :.

";

: .: ;

Re ( ~ I )

')

0,

Re ( l -

p)

...

13' ) >. o;

Re ( 1 ' -

'> 0,

..

'.

:'.{10. 3) .. F3 (o<. , ' ~ ~ , f> ' , y ; x , y)

i.:.
..i'..

'~,fo>r< ~~ rc~-~-n
...
:

J'

.. . .. '

u,v,1:-u-v_O
. . '

. :. ; . ... .

<: ~ .

. . r .

Proof:

:> O,

~:

The main idea

.. _.

~ - ~ .. ~ ' )

Re (

. . :

. -

') 0 ~ ~e ( ~ 1 )

Re ( p )
~.

...

.H u!-lvP'- (1-u~v) -~~,~

~ ~

: ..

..; -

-~

-r

-~ c'1-~) ~"'dudv

> O
.. .,

i'

of the proof of this theorem is similar

; : : (j~ :

:. ~~ tha~
y::\.

~:i ..

-l Ct-xu)

: .

'. ..; ! ;. .. .!

.: . '

, .... '

of the corresponding theorem (Theorem 3 .1 of Section 3,

.

..

"-! :

.,

.' .

., .

.,C~pter.t) ' in the_: case o~ one variable'. :we shall Sh()W ~n~y .. formal
t.

:~~flculations.

.~ j

" ;

'.

First of all, remark that

"

' L

'.

.:: , .

..

...

:. ,

- 62 -

(1 -yv)

-o<'

(c:i< . n)

00

n.'

,n:::Q

(1-xu-yv)

-o<.

~ (oc, k) . ' . ... k

k'
(xu+yv)

= L.

k --0

00

2:

m,n=O .

. y :.v

~ (o<.', k)
k,

......

k =O

"" _!51_ m n m . ~~
, , x y u v ,:
m+n=km.n. . . ' ' 1.
~

: ,:

( oc , mtn) m n m n
m!n!
x y u v

Furthermore let us note that ,, we have

...
'.
', ' i - .
. ,, .

p-1

(1-u)

q-1

~u == .

rep> reg>

.r

r(p+q)

: "

'

> o,

if Re(p)
,

Re(q)

..

. and
!

> 0,

1.

: ...... . : .

The second formula will be derived by a repeated application of

.. .:

first formula.

t' I;

.'

'

\'

. Now let us return to the Euler integral

?::
the,~
=6

....
_

represe~tations .

Consider
I

f
J
u~-lv(3'- 1 c1-u-v) 1-~-11- 1 c1-x~-yv)-ricdud~.
u,v,1-u-viO

= .

Replacing the last factor

power series representation,

fr

..

..

~ -~

(1 - xu - yv)

we

,,

by the corresponding

obtain

121 -1
111
-1. VP
(1-u-v) ' (-,4.. - _., .- 1 'z:oo -(O( ',-m+n)-

;ll-= . . J
u.,
u , v , 1-u-v ~O

(1 m)(l n)
m, ns:O '
'

/
xmnmn
y u v dudv
l

\;
. :1{

Interchanging the order of the integration and the summation, we

m+n) xm n
.
11 L~ (l(,
m) (1 n)
Y
m,n=O
'
'
u,v,

If

P+m-1

-u-v~O .

~'+n-l(l

-u-v

1-~- ~-id d
.

u v

ge~

- 63 -

rcp+m>r<p'+n)r(1-~- ,1') ~m n '

rct+m+n)
y

(oc. m+n)

m,n=0

(l,m)(l,n)

, ;

C<A>r<~'>r<-~-~>
F1<, rA.. ~ ' , '( , x' y > .
r< I{)

...

Notice that we used here the identity .. f'Ca+k) = (a, k)T(a)' ..

Similarly we have

u ~-1v ~ -1
(1-u) r-~-1
. (lv) 1

'uSS 1

U'-v~l

.. . ,

lJl

. . ...

~- -1 (1-xu-yv)
.
- du dv . :

<'- ,m+-n>

xm n

. . . ,.

'2

r o - @>

-~'>r.

, . .i

. u

~-1 ~ ~1"(. l

u,v,1-uv~O

-u-v

'

. .

. '..

. '

11 ,

re r- f r

rca+n)

. I( ~+n)

. f.\ .' li'

' "

.- ss

...-

>'

~-

2 ~ , ,... , ,.. , a , , x , Y

and

.. .

'

)1;..(3-~i _l ~
..

1 .
1
' 1 1 .
du.
v~:~ ~ . (1-y) 't .-(3 - dv
.. . 0
. .
.
.. :. .. :

a.-... .~.

rc~+m>
. r()'+m) .

.. T<6>r<~'>rc1-a>r<1
re ~ > r <t >

'

'II'

m,n=O (1,m)(l,.n )

m, n=O
.

aO
(
)
j' 1 u. 1
, ... m,n=O
Lcie.
,m+n xmyn . u'" m- . (l ';'U)
(1,m)(l,n) . . 0
.. ,
O<I

~-1 ~'-l(l
. )1-(3-l(l. ). 1'~~-1.::. . (l)(,m+n). m nm n d.-~.d

. -u
-v
~ (l m) (l n)x y u v u v

u v

2.

..

...

.
,o( ,m)(oe.'.n) m nm nd d
(l,m) (l,n) x y u v u v

m,n...O

- 64 -

~ ( <X , m) <ex '. n) m n C( #+m)

= L- .
x y
(1 ,m) (l ,n)

r ( #'+n >r ( ~ - p-

m,n=0

- rep) re f)r ([(

l t)

fl - f)

F ("'

r< i+m+n)
.,,

~ > "'

r '

ar

I'

II

y)

v
'

@' )

>

'

This completes the prcof of Theorem 10.1.

The function

does not have any simple integral represen4

tation
such
as (10.1), (10.2) and (10.3). Thi~, however, does not
.
.
'.

mean that

does not

h~ve

any double integral representation.

For example, the following formula is kn.o wn:

r<a-> re 1 >

F 4 (c<, ~, ~, i , x(l-y), y(l-x)) "'r(.i)r(~)r(~--:<)r(~'-/3))(

f.
u cJ.-lv~ -l (l-u)i-o1-\1-v) '-p-l (1-xut-tt(l-yv l-t-t(l-xu-yv /+J'-oc-~-ldudvi.
Jo o

1 1

Re ( ol.) > 0,

where

Re (

) '> 0,

Picar d discovered that

F
1

Re ( '4 -

o<. )

> 0, Re ( i ' - (3 ) ) 0.

admits another integral represen-

tation which is not a double integ~al, but a ~impleintegral.

THEOREM 10.2: . We have

(10. 4)

F 1 ( ct. ' ~ , ~ I

if

Re ( . )

't

r <1 > J.
r<~) r<i -o.:) 0
>

0,

Re ( -

x J y)
1

ei )

-1

(1-u)

r-o: -1
-P
- P du
. (1-xu) (1-yu)

> 0.

The proof of this theorem is very similar to that of Theorem

10.1.

- 65 -

From (10.4) we can conclude that

tinuable and holomorphic for
(10.3) implies that

F
1

is analytically con-

[C - [l, C)O)] x [C - [l, oo)].

is holomorphic for

Similarly

[- [l;oo)] x [C- (l,oo)].

On the other hand> we can prove the following theorem by using (10.4).
We have

THEOREM 10. 3:

(10.5)

Fl (OC, ~, ~', 'lf, x, y)

(1-x)-p(l-y)-/J' F. (
1

.. (1-x)

.. (1-y)

-ct

F (ct, J- ~- ~,
1

Proof:

p', 'l, x/(x-1),

r (Q(,~,Y-(3-8',r,

..., (1-x) ~-g( -p (1-y) -

(1-x)

-o< > p, p', '(, x/(x-1), y/(y-1))

~IF 1 ( t - 0( '

-ta (1-y) l--13' F

(~-y)/(x-1))

(y-x)/(y-1),y/(y-l})
r- p- pI

1 (r-oc,~,

fl' , "( , x,

'

(y-x) I (y-1) )

~-~-{J',, {x-y)/(x-1),y).

This theorem can be proved by changing variable

respectively by
u

= 1-v,

u "" v/[ (1-x)+vxJ,

(l-v)/(1-vx)

Exercise 1.

and

u ... v/[ (1-y)+vy] . ,

= (l-v)/(1-vy) .

Prove Theorems 10.2 and 10.3.

It is easily verified that the following six changes of

variables:

1-u'

u'

1-u'

zo

1-v'

u ... v'

v . 1-v'

'
,

v *" 1-v'
v ""'u'
v

'

lv'

v ... v'

1-u'

,
,

lu'

- 66 -

map the square . 0 " u, v

l . onto itself and do not change the

form of the integral (10 .2).

By using these transformations we c a ri"'

:;~~

. _pr.ove the :following theorem.

:
10.4:
,. --THEOREM
.

(10. 6 >

We have .

' ,

r , r' , x, y>
(1 ...x)-"'F ("', r-p, f, "I, 'I', x/(x-lL y/(1-x))
2

r 2 <,

~, 13 ' ,

(l-y)~F (C(~

p, 1'-p',r, 't',x/(l~y),y/(y-1))

(1-x..:y):.-F (0c, y ;..13, y ! 2

- (1-y) -~If 2 (: > .

r .- p' ,-p, i'

. (l-x} ~4 ~'-2 ( .2 P', 'j.

. -Cl(

~,

'/, 't!, x/(x+y-1.) , y/(x+y-1 ) )

>:'( ''y /(yl)-;

x/ (1-y))

P; '/ 1, '(. > y'/(1.:.x) > X/ (K-1))

I

t '

..

. ., :. . ..

..

- .(1-x-.y) . F (ot, -(3 . > r-_ ~> .'t ,1,y/(x+y ... l),x/(x+y-l)).
2

It can

be,_, shq~n

that there .is no change of variable's of the

form:
u

A"u I + B"v I + c"

. .v ' . = Au' + Bv ' + C

'.u ' + B 'v ,. + C

= AAu'+
Bv' + C

which maps the

triangie u; v~ . 1-u-v ~ 0 into itself and which

does not change the f~rm' of the. int:egral (10.3). The~efore, it is

difficult to find any rtransformati on .formulas" for

3 '. which

are
.. ;

similar to (10.5) and (10 . 6).

Let us nex't find some relations between contiguous functions.

To do this, consider the . representation (10.4) of

F
1

Raisi ng cc

by one we obtain

rc'+nr<v>
re y- ~-1>
a

p, P', "(,

F1 (((+1,
x, Y>
1
.
.
ot
f
-2
A
u (1-u)
(1-xu) t' (l ~ yv) - ~ t du

J
0

- 67 -

On the other hand, lowering ~

::'., r c'. >rci-1>

r <t -

o( -

>

Fi <oe., ~, ~
'""

by one we get

, t-1, x, Y>

1
.
~-1
t--2
- I
u
(1-u)
(1-xu) f. (1-yu) /3 du

If we put

we have
>

r (IX ) r (1 - ~) F
r( )

( t:<

(.f '

' r'

/.t I

IJ '

~ ' X' y

f 1 u

.
(u) du '

1
f(ij.+l) r((--l) F (-'+1 f'. a.' "
.
)
""
[
<~ >
i .... ., ,... , ,~ , 11 , x , y
o u(l-u)-lU(u) du

':}. ;

-Io

.: . .:

'

f(ot) CCX-ot.-1) F (,...,

r <J -1 >

V' '

a' " 1

, ... '

j.J

'

u-

, x' Y

'
U(u)du+

f 1 (1-u)-1' U(u)du
0

J,(0 1(1-u):-lU(u)J.u

and hence
'i:

olF 1 (ot+l, ~ ~ ~, 'J, x, y) ~ (~ --l)F 1 (O{,

- ( ~ -1) Fl ( o< ,

p,

(10.7)

U(u)

to obtain

-0:..i.J U(u)

Re ( o<. )

> 1,

Re ( ~ - ot) '> l.

lim U(u) -= 0 ,
u,+0

~, y )

r(x+l)

U' (u) = [. o<-~+o(+l- +-AL+

u
1-u
1-x~. 1-yu

Assume that

i,

p, ~ ' , ;r -1, _x , y)

In this computation, we used the. identity

Let us differentiate

~',

Then

. lim U(u) 0
U' 1-0

= 0

= x r<x).

- 68 -

Therefore, integrating both hand members of (10.7) from

to

1,

we get
1

(ci -l)J

1
u- U(u)du+

(ot.+1-~)J.

1
(l-u)- U(u)du

(1 . .

+ px Ji

r1
.1
p'y Jo
(1-yu) - U(u)du

(1-xu)- U(u)du+

'"" 0 .

Thia. means that

<t-ot)F (o(-1, ~,
1
-(b'-l)F

(oe.,p,

p', r, x, y) + (ot-l)F1 (ol, p,

~',

~,

r-1,x,y)+pxF1 (o<,p+l,

f3'yF (o(,p,~'.+l,J,x,y)-=

y, x, y)

~',J,x,y)

0 .

Finally we shall present the following application.of Euler

intE>gral representations .
the kernels of integrals

In the proof of Theorem 10.1, we expanded

(1-xu-yv)

-oc

and

O(

(1-xu)

the corresponding double power series in u

and

(1-yv)

v.

-ot'

into

If we expand

'

these kernels into series of different types, . we shall obtain some

other expressions of Fp F2, F3

We

in fonns of certain series.

shall present such an expression for

Fl.

In (10.1), change

variables by
u ... s(l-t),

0 ~ t, s

This maps the square

~

o.

If s .. O,

the side:

u ... 0,

10.1.)

s = O,

= 0.

then
0

~-

' 1

v "" st

'

0 arid

.
u, v, 1-u-v

into the triangle

'

v ; 0

.'.

for 0 ~

'

Hence

t ' 1.

of t:he square is mapped onto a point':

Otherwise. the mapping is one-to:--one.

(Se~. Fig~
I

, ,I

'

.. 69 -

Fig. 10.L
;i,. .

.;'.t;<P>
rep' f($)
>r<1-~-t3' >F1 (a<. , ..."' ,,..
.

,Q

t x y)
, , ,

Sf u,..

A-1 /3' 1
f-A-t.al -1
00.C
v
(1-u-v) '" ,...
(1-xu-:-yv) dudv
u,v,1-u-v~O
.
.
..

a::

.'

... ,,

""

:;;::~):

1 1

=-

1 t

lfl /3-1
12-1 A'-1 (3'-1
~-~-A'-1 q-xs+xst-yst)
.
-0(
1
s
(1-t) .. s 1..
t
(1-s) t' ,..
sdsdt
0 0
.
.
.

f
J,( f

0 0

.a+a-t
1
1A'-1

s"'

. . .: . '

'

'

..
(1-s)

. .

. '-

(1 ~xs+xst :...ys t)

~-"-ll'-1
1
'

~-1

. (1-t) .

r-

. :..oc .

. .

. .' :

(1-xs+xst-yst)

(1-xs) " (1-(y-x)st (1-sx)

. ~ ~ ' co( ,m) (

L-.

m=O (1,m)

-1

-~

. )ni . mtm(l . )-{m+ol)

y-x s
-xs

~d..

dsdt.

- 70 ..

f(p)r(p')r([-(3-p') F
r(~)

f s~H'+m-l(l-s) ~-~- ~-i

1

f:

(o<,m) (y-x)m x

m=O (l,m)
(1-xs) -(m+o<) ds
.

1:

t p'+m-l(l-tiP-1dt .

Observe that

s6+6'+m-l (1-s)t+m-(/l+ ~'+m)-1(1-sx) -(m+o<) ds

rc~+A'+m)r(l-~-@') F(o<.+m
rc~+m)

a+ a'+m, v+m, x)
) \ 'J
0

and

Hence

In particular, putting

EKercise 2.

= x, we obtain

Derive the following formula:

~ lJ~ , oY , x, Y) = ~O
~ (<l(,m)(~+@
,m)F(
P.+ tP 1 / ) m
F 1 ( o( , I"',
( ~ , m) ( 1, m) -m, ~ , ,.. '-' , 'I x x
(It sould be noted that

F(-m, ~ ,~+~ ,1-y/x)

are polynomials.)

;?{.. .: Barnes

iti~e,ved

integral

~.:::;:;. ~

that the function

In Section 4 (Chapter I) we

.;>tlLl..aLlOClS .

~,

p,

F( o< ,

x)

admits an integral repre -

;:)~~q,~ation of the form

?\Hr:; .
i .
,'.;if;<
' '."1,-_,,1 )
.

.;

F(ol, A, 'f, x),..

::~i:~L .
/.? ol, p>

.');:~!\

<~ig.

r(r)

p.

''.~i~ed by oc and

. :::l~i .

r(iY+s)

)s d 5

determined by o<+m

determined

by

o(

is dete r -

F~rthermor~, 'if m is a nonnegative intege r,

and

:''.-.:!:

j'~~th

r(o(+s)r(~+s)rC-s)( ~

Notice that the pa th

(See Theorem 4 .1.)

4 .1.

21ti r(~)f'((3))

't .; 0, -1, -2, , where B is the path given by

:::~i(

:?'the path

and

can be replaced by the

p.

In Remark 4 (p .44) of Section 7 we proved the following formula :

.

.
(IQ

(
" '~ Fo(,r,P
,y,x,y

~ (oc m) ( 8 .m) F ( +
,
) m
""'~0<1,m)h,m)
o( m,p ,t+m_,yx
.

:./:;'.
;.

<;} ~ we utilize the Barnes integral representation (11.1) of

:'.!,;: :::!:; .

:.t(o< > p, Y, x), we obtain

\~if ()[(~')
A A' v
)
~
)\[:.
r<r> FI
1 '''r'v 'o,x,y = ~
....

(13,m)

(l,m)

r{ot,m)r(f)Ft~+
A'
rct+m)
~ . m,,

~ C/3,m)xm
C(o<+mtt)r(!} '+t)
m=O (1,m)
2rti B
r(~+mtt)

. . 1. .1

_!_ L. <ot+t .m> <~ ,m>

2n:i ~
('6+t ,m) (l ,m)

r<-t) (-y) t

xm rcot+t>r<f+t>
r<+t)

v+
,0

) m
m, y x .

dt

r<-t> C-y) tdt

- 72 -

. ~, ~+t, K) r.<~+tr<i+t>
)r( p'+t)
Jsr F (ct+t,
r<-c> <-y >t dt .

. 21[1

Thus we obtain the following results:

THEOREM 11.1: Suppose that

Then

t~e

ot.,

o<',

p, fJ', '(,

1 :f. 0, -1, -2,

hypergeoruetric functions

integral representations
(11.2)

p, ,,x,y)

rc~>r<f'>

F 1 (o1.,p 1

r(t

_L ( F(ot+t a 0v+t ) r(o<+t)r(i3'+t) r<-t) (-y)t dt ,

2n:i 1
'1"
,x
r(1+t)

JB

( 11.3)

[()C(f') F ( ~ A'
.,,,
)
r<~I
2 a(,IJ'I" 1> ,X,y

I I

f, " .

i
21ti
(11. 4 )

rC')f(6'>

r<.O

(11.5)

BF\+t,p,~.x)

F 3 o< , o<.

re -t >< -y >t dt'

p, ~ , , lax , y )

i!i f/ ,p .i+t

r{oe>r<tu
rCl')

rcoc:+t>rc~'+t>
r(t'+t)

,x)

re '";.~:l~ ff +tl r< -t) (-y) t dt

F4(o<.,'3,t, ,x,y)

.. ..i..
F(~+t wA+t , " ,x > rco(+t)r<@+tl
21ti
rci'+t)
. B

respectively, where

..and (.11.3). by

o(

The function

integral

rC-t)(-y)t dt

[!'

is the path detPrm1.ned by

i.n (11.4), and by

F(O(+t,~,+t,x)

repr~sent~tion:

o(,

p'

in (11.2) .

o<., (3 .in (11.~).

i.n (11.2) admits the Barnes

- 73 .--

. r(oc+t)C(~) F(o<.+t "r+t

r<i+t

)f#Sl.I

,X

)-=...L(

where

r(o(+t+s)r(Q+s) r<-s)(-x)sds.,

27Ci~

.r(f+l:+S)

p.

is the path of integration determined by oc.+t,

Bt

Hence

from (11.2), we derive

-C(~)C(8)C(0') F (~ p n' v
y)
.
f(J)
l ""' 'I' ,,,x,

::'Jf..

2
./:. .
, (.--L.
)
JJ(C(o<+t+slr<a+s)r((l'+t)
21tl.
r<i+s+t)

r(-s)r(-t)(-x) (-y)

dtds

d3

/3

: ;:where the domain

/3

of integration is given hy
= { (s, t);

E Bt ~

t E. B }

1.-,

(, ,_ ~ .:

.'

,::,Thus we obtain the following results:

THEOREM 11.2:

Suppose that

o<., oc', ~' ~', , 'l/

Then the hypergeometric functions

Fj

(j

= 1,

0, -1, -2,

2, 3, ') admit

.integral representations

:fJ(ll.2 ,)

r<o(>r<a>r<~')
. r(l)
Fl (o{,~,~

~(~)2
r
21Ci
)
.
.a
c(ll.3')
:i~

r(c(.+t+s>rrn+s>r<a'+t')
ro+s+t}

: r

re- s )r(-t). (-x )s (-y )t dt ds ,

.

rcot>rco>r<~'> F (ol,~,~,~,i',x,y)

r<i>r<i'>

.. (....L)2J~f
>s<- )tdtd
2ni
ll. :f(ol+s+t)r(fi+s>rc~'+t>
.rcit..s)r(~'+;t) . .. re- s >r<-t><.
x
y
s ,

:;
.:~. ;

. , .

., , 11 4

'7.

, ,i ,x,y)
.

:.

',

: ,,

r~

: , ;

,> r<at>r<c<'>rrn)r<a'>
r (')
,

1 )
~ ( 2rl

,.

.J, a. it ""x y)
3 "''"" ' I"' I' , a , '

F .<J

f rCo<+s)r(o('+t)f(6+s)C(~'+t)

l}
f.i

..

r<1+s+t)

..

f(-s)r(-t)C:x)s(-y)tdtds,

- 74

(11. 5')

1 )
= ( 27ti

ss

r(ot+s+t) r<p+s+t)

s
t
r(~+s)r(t'+t) r<-s>r<-t)(-x) (-y) dtds )

t,

where the domain of integration in (11.k')

{j

and

a:.

f (s ,

s Bt

t) ;

t E B } ,

B is the path of integration in (11.k), while

o< + t,

of integration determined by

for

= 4,

~+t

and by oC+t,

for

(3

for

= 2,

Bt
3,

is the path
by

' , p

k = 5.

By utilizing (11.4') we can derive the following result:

THEOREM 11.3:

The functions

and

are related by the

formula:
(11. 6)

II::

~J (

f (oC > o('

o(

Cl( I

p, t1 )x
t

p,

O(

~I

Xt

Y)

y -o< 'F2 (o<. +(.I('+1- r'"' 'o(' ,Q(+ 1-p' o( '+1- P' '1/x,1 /y)

f(o<.,~ ,~,o<')x-y-tl'F 2 (o<+~'+l-~,o(,~ 1 ,oe+l-~,13 1 +1-oe' ,1/x,l/y)

f ( ~ , tX' ' " '

fl I ) x - ~ y - o( I F 2 (~ +oC I+1- ~ ' (3 , o(' ~ + 1- ~, o( t +1 - (3 ' J 1 / x , l / y)

+ f(~,~, o<,o<')x-<'y-ll F 2 (/j+[3'+1-x,~, (3' ,(3+1-.x, ~+1-0(' ,1/x,1/y) ,

1

where

Instead of giving a proof ~f this theorem, we shall prove a

similar result in the case
THEOREM 11.4:
Then

Suppose

of

one variable.
ct,

P, (,

ci-

(i ,

0( ,

p f: integer.

:'.:. ~

- 75 -

= r<a->r~
-
r('a'-oi.)r((3) (-x) F(o<,ot+l-,ot+l-,-9,1/x)

F(o<,(3,~,x)

(11. 7)

+
Proof:

.,
!.;.

(11.1) of

re~>

r<ot -M

r(oe)f(t-p)(-x)

-r;

F(~+l-c)',~,~+l-o<,l/x).

Let us consider the Barnes integral representation

F( "',

~,

'(, x).

in x . for
.
.
(11.8)

This integral is uniformly convergent

l arg(-x) I

We shall restrict

<

1t

to the domain (11.8).

The multiple-valued

functions on the right-hand member of (11.7) are also well defined

Take a sufficiently large positive number

in this domain.

let

BR

be the arc of

s. = Re te ,

semi-circle:

(s)

-iR and

B between

I0 -

7t \ " ~ 7t.

C(ct+s)r(M-s)
i::.:..
-: r<o()r(~)
ro+s)

in the closed curve consisting of

. ,

'f(s)

Res r(s) = ( - l)m/m!

s=-m
Hence

= - . -m and s ,.. -

at these poles.

along

'f <s)

and

DR:

BR
~

and

-m.

which are contained

DR.

Those pole ri

~n

Let us compute the residues

Observe that
and

BR

..1

of

"f

~ Jor 'f(s)ds
R
-

is the sum of residues at the poles of

be the

iR .

rc-s)(-x)

Then the difference between two integrals of

located at

and

Put

_ ro>

2-!:t f t<s)ds . JBR

R,

r(s)r(l-s)

= __7t:_ _
sin(7Cs)

.. 76 ..

In this manner, we obtain

.

Res
sa-.x-m

'

(-xy-o< .{oe,m) (oe.+lr,m) x-m

+cs> = r<ir)C(ffe-0()
f(~)r(~-o1)
(+1-13,m)(l,m)

..

.. ....

On the other hand, we have

:: 11m

..i

'f (s)ds

a - ~~:

.;.

' .

'

tends to

+oo

DR . .

if

!x) > 1

.. r .. .

larg(-x)I
< 7t' . To prove this, it is
sufficient
I ."._!
. .
....' '
. . -

a n d

'

:'

'' :

to estimate

'

:_rclJ(~(~r!)+s) r<-s><-~>s~.
on

OR.

We can not use ' stirli~g Is formula dir~,ctiy~ because

'larg{~)-~l < f. .

Stirling's formula is not ,valid in a sector

,

'

l_ .'

~void this difficulty, ' iet us .use the formula

..:1

.._..

...

. . -

..

. r<s)f(l-s) - _
: '. .:

To
1

';

.' /.: '.' <

'

'

~~- :

. ._.~

, ,',;

:'

.!

.. . .

"IC

_,__..,..

$in(TCS) '

to obtain

r: (oe+s)
C(@+s) re. - ) (- )s
rct+s)
. . s x ..
;:.

.1.

r(l-~-s) (-s)
7tSin(7t.((+s))
( .,;~)
rc1--s)r(1-~-s) sin(7r(o(+s))'s~n(7t(p:ts)) ..

Now by utilizing Stirling's formula, we can estimate the integral

- 77 -

of 'f'(s)

DR.

on

Exercise 1.

Complete the proof of Theorem 11.4.

.

At :.the:. ~nd

of Se~tion 5, we mentioned that t:Wo functions

(-x)F(' ,9'+1-()' ,o<+l-t' ,1/x)

and

are linearly independent solutions of the hypergeometric differential equation.

The formula (11.7) was found by Kummer.

Since

these solutions are not single-valued functions, this formula will

not hold, if branches of these functions are chosen in a suitable

manner.

Thi.s fact was not well understood at the time of Kummer.

In order to prove Theorem 11.3, it should be noticed that the

domain of integration in the formula (11.4') is

a= B1 x B2
where

B
1

f(s, t);

s B1 ,

is the path determined by

path determined by

p.

oc ,

oc ,

p,

B },
2
while

B
2

is the

Therefore, the idea in the proof of

Theorem 11.4 can be directly applied in the proof of Theorem 11.3.

To prove Theorem 11. 3, therefore, we must compute the . residues of
the integrand of the right-hand member of (11.4') at pairs of poles
{" = -a(-m,
t

For example,

-ot.'-n>

r= -p

-o(-

m,

-n,

r
t

-p- m,
-o('-n,

{ 8 -

t -

- {J- m,

-p' -n.

- 78 - ;

Res f(ot+s)C(oc:'+t)r(G+s)C([i'+t)
rci+s+t)

s =o( -m

ta-~ 1 -n

f(- )r(-t)(- )s(- )t

s

{(

..

.,
.- 79/ -

J~ ;~ ;.

Systems of partial differential equations.. The function

.
.. .. . . . . . - ... .... . .
'

:'.rt~: , ~,

r.

satisfies the differential ~quation

x)

+: [ t .- ~o( + ~ + 1):2C.l.Y ' ; .:. .ot..p y. ~

x (1 - x)y"

11;

'

~.I!:....

'

'

ft.)}~

pa,~tial .

,,,._., . . ..-

'

' ....

"

, ,

hypergeometric functions

j1y13_tems of

"

,. I

' ': : :

- . ,

'

. ,. '

0 .;

'

: _,

'

Fj (j ' = 1, 2, 3, 4) also satisfy

diffe.rentia.l eq.up:;:ipns .
.:

It '

.i

: THEOREM 12.1:

F , F , F
and F
2
3
4 satisfy respectiyely the
1
:,'following
systems of partial differential equatfons ': .
:: .:. :

~.

:.

.'. . j . : ~. I .' , ,.

.~

..,

: . .

{:~~~:~=~~:; ::;:~::~::::~~~~~~~:~i~~~\.'o .

,.:::\:,;,_.
'.~c:i2'-' .~ 2 >

(1-x)r~~ys+( 1,-(ct+~~l)x] ~. -~yq: ..~~z";.: '( { <.~."_

y(l-y)txys+[ t-(6'+p'+l)y']q-~xp-'~' z = o ~,
x

x(l-x)r+ys~i'-<~+p+.l)~Jp:_~~z ~
{

0 ,

y(l-y)~+xs+[~-<'+p'+l)yJq-oc.'~'z ;. o, .
.
2 :: : . . ..: . . .. ,i:;' .:
.
.
. '. :
x(l-x)r-y t-2xys+["i-(ot+('*l)xl p- (c(+~+l)yq-Pl/fa

= o. ,
y(l-y)t~x.2 ~-2xy~+[ i'~(ot+p+l)y]q~(D(+~+l)~p-~~z = 0 ,

' Proof:

We

shal.l ~ show, that' F{ . satisFies (12 ~ 1). : Other~ can

!~

j~~:~f.:shown in a similar. manner:.~: " . Deqote by ..e ' . and

x'(J/'dx :.
;t,~:~p,ectively.

8(6+tH-l-l)F
1

..

.and

T .the

opera tors

:; , y;J/Jy ,

Then we have

""

m(m+n+i-1>

1 . m,n=O.

(,m+n><a.m><f .n)
(~,mi-n)(l,m)( ,n)

xmyn

... .

- 80 -

<f

) . (oCm+n+l>(~.m+l)
(y,m+n) (l,m) (l,n
,.m,n=O
. . .

.. x

f.

m,n=O

,n)xmt-1 n

Y_

.. . ,

EX.+mrnl~+m) (()(,m+n) ((3,m).(@' ,n)

.
(,m+n)(l,m)(l,n)

xmyn

_. : !:

Hence

In the same way, we get

'I (EH- 'f+ r -1)F1

~ ,y ce-r1.+oe ><':'"'~' )F1 - o .

From these equations we can easily

Similarly we s,e e. tha.t . r 2 ,

.
. .

(12 .2 1 )

(12. 3 ')

r 3;
.

~erive

..

(12.. 1).

and. .'1!'4

-1>~ 2

..

{ e.ce .+-t~t)F 2 , ~ ~~~t'+~><~+p>r 2

.... , .<er+ t

... . .

o, .

..- 1ce+r+ o() <~t f>F 2 "'.' .o ,

8(B+t+1-l}F3 - x(0+oe)(8+p)F
3

=0

{
. ,. ,(8+t+r.- 1)F3 - :y(g>+oe')('+ 1J'F3 = O ..

ece+r-1>F4 -xCB+,+Q(}(8+,+ tJ>F4 ... o ,

'fCCf+

'-l)F4 -y(8+'+~)(8+t+P>F4

From these the systems (12.2), (12.3) and

derived.

t_~

sa~~sfy r~spectively: '

(12~4f

""o .
can be easily
~

..

I.

"

81 . -.

~;: :

Systems of total differential equations. Before '~tudylng

the

;~i::.

hsystems of . par~ial ~~ff~;enti~l. equafions which . Fj :: satisfy, : ~e

[:~h~ll

explain some aspe_c ts of total differential equations.

;~{

..

[.~i\,; ~i' Let

tota~ ,differential

us co_n sider. a .system of

eqli.ations : of

;t~~~ form:
~zj
.

= .~j '.~'.~ ,z 1 '. ~- ,zn>,~ + gj (x,y ,~ 1 , . ,~zn~~-!

. . ...

. .

. -

...

i,l,,n),

. .,.. (j

,.

,~~'.~~ere

f 1 ; .. , f 0

g1 , , g0

are functions of

x, y, z 1 ,

;~~0::~ ~hich ai;e . define,d .and co~t;inuously ._differentiable in a domain

n+2

::i~: '

~~;.~- contained in ,R

_If w~ ..C?OQ.sider ; x,:_: Y - ~s independent

~#"~

\t:.~riabl~-~: and ,~ 1 , :

zn

as :dependent variables (i. e ; unknown

~'.:tti~ceicms of. ;_ ~, y)., ' then the system. (13 .1) is equivalent to the . .
f,~~~: :t

:s ystem . -. :, .::

.:-.

~ .

- _i :. ~ ....

r=.

. . , lzj/ax == . fj (x, y, z 1 ,_ . ~, z 0 ) ,

~zj/'t>y ,;;. ; gj(x,y~ z 1 ; ~ zn)

.

...~.

. . :... .. _i, ,:

;, ;

zj ..

..:

' j (x,. y) .... (j .a:; l

~.

i .:. , '

n)

(.

~ :

0)
' ,. .

:~.~hich is defined and twice co~tintlousiy differentiable in a

;;,,; .

''

:, D c

2 .
R

1"

,.

l'

domain'

are independent

ilii~j;

:of the order of i .differ~nti,.ation with: respect to

"; O

. i).2 'f ~
.~

~: ' ~2 'f j
~y
. ay a x

(j - 1, n)

x . arid

y: i.e. ;

- 82 -

From (13.2) we derive

= 1, ,

(j

n).

Hence

and hence
n ._,

(13.3)

~fj/"dy + Ei_gk 'd fj/dzk "" ;>gj/ax + ~ fk dgj/dzk

(j

This condition holds for

(x,

y, <fi (x,y),

= 1,

, n).

, <j> (x,y)),
0

(x, y) D.

there exists a solution '

of (13.li satisfying the initial condition
(j .., 1, , n)

zj(x , y ) = zjO

0
and the smoothness condition, then (13 .3) holds

identically in J} .

Therefore, (13.3) is a necessary condition

for the existence of such solutions of (13.1).

In general, in

order that overdetermined systems such as (13.i) have solutions,

we need some conditions such as (13.3).

On the other hand, the

theory of total differential equations guarantees that (13.3) is

also a sufficient condition for the existence of solutions of (13.1)
satisfying

a~

arbitrary initial 'condition .

.'

THEOREM 13.1:

Suppose that

fj(x, y, zl' .., zn),

g.(x, y~ zl, .... ' z) are "all continuously differentiable in r5'

J
n
n+2
.
c R . and satisfy (13.3). Then for any point . (x 0 , y0 , z 10 , ,z00>".
there exists a unique solution of (13.1) which is defined and

- 83 -

.' .twice contiriuously differentiable. in a neighborhood of

~nd

(x , y )
0
0

satisfies the initial condition

.(13 .4)

(j 1 >

DEFINITION 13.1:

n)

A system of total differential equations of

the form (13.1) is called completely integrable if (13 . 3) is satis

... tied.

!..

:~

f.

We shall now consider the case when

holomorphic in

and

g.
J

are all

x, y, z 1 , , .zn

THEOREM 13.2:

Suppose that

,<x , y 0, z , , zn )
0
10
0

and

fj

are holomorphic at

gj

and satisfy (13 .3).

Then there exists a

unique solution of (13 .1) which is holomorphic at

(x , y 0)
0

and

. s atisfies the initial condition (13 .4).

Let us consider the case when

:'1h1' z , , z :
n
1
'

'

.Jl3.5)
. .~: .
'

and

are all linear

i.e.

..,

f.

dz. = .L, f .k(x, y)zk dx

J
k=l J
.

+ L,

k=l

gJ.k(x, y)zk dy
(j

= 1,

n).

If we denote by
;

..

wjk = fjk(x, y)dx+ gjk~x, y)dy ,

(j,k=l,,.n), .

the system (13.5) can be written in -the form

(j - 1,

n) .

.. 84 '

Moreover if we put

[~.11.

z=

ln

... (.()
11'.......
,

we can write the system (13.5) as

dZ = J2. Z

The condition of integrability becomes

.: zk[-af.k/jy+ ~

ks:1

f 'kghk]
h=l J

k=l

zklag .k/ax+ t_g.hfhkl.

l:

h=l J

1,

= 1,

n .

Since these are identities, we obtain

n

(13.6)

~fjk/;,y+ ) . f.h~k

h=i

()g ..ktax+ _Lg.hfhk

h=l J

j, k

As it is well known, solutions of linear ordinary differential

equations exist in an interval or a domain where coeff.icie nts of
the equations are continuous or holomorphic.

The same fact holds

for a system of linear total differential equations.

We shall

state such a result for the holomo;rphic case.

THEOREM 13.3:
in

Suppose that

fjk

and

and satisfy the condition (13.6) there.

zj

..

'f j (x,

y)

(j "" 1,

are all holomorphic

gjk

... ,

be a solution of (13.5) which is holomorphic at

Let
n)

(xo, Yo).

Then

- 85 -

>.1'f111 ..... ,'i'n are analyticatty continuable along any Eath in

starting at
It.is clear that the set of all solutions of a system of
linear total differential equations is a vector space.

Since

solutions are uniquely determined by their initial values, the

dimension of this vector space is

iHEOREM 13 . 4:

The solution space of (13.5) is an n-dimensiona l

vector space.
~olutions

Consider n

of (13.5):

~.'I .; '

-1 J. : ' : .. ' - ~ . '

(j , k "" 1, p n)

and the determinant obtained from these solutions:

,, 11

'f ln
..

In the sawe way

as

for linear ordinary differential equation~) we

derive
n

dA

ss

L_fjj dx
jml

n
~g . dy)d

j""l JJ

or
n

d(log A) ... ,L.fjj dx

jml

THEOREM 13. 5:

We have

n
L,gj. dy
j..:l J

- 86
.

. (13. 7)

A(x ,y) ... '1(x ,y0) exp[

(x,y)

fj. (s, t)ds+ 2:.g .. (s, t}d:t];

,y ) J=l J
J=l JJ

0 0

We have completely similar theorems for the general system of

total differential equations

m
~zj ~ : 2',fjk(x 1 , ~ ,x ,z 1 ,. ,z )dxk '
.
. . kl
..
m
n
.
Exercise 13 . 1:

l, ,n .

Show th~t, if we write' ' system (13 .5) in the

. form

dZ "" Jl Z ,

then the integrability condition (13.. 6) can be written as

- 87 -

Transformation of the differential equations for

systems of total differential equations.

F.

-into

We saw that the systems

!: .

:,~f

partial differential equations satisfied by

are

Fj

~ritten

in

;the form
Al r
{

.where

Aj

'
and

where

A2s

A3t

A4p

Asq

Blr + B2S + B3t + B4P + Bsq

and

F 1 , F2 and

Bj

F3 .

are holomorphic in . x

These equations are

A6z = 0'

-i- B6z

and

solv~ble

y.

A --= B
1
3

=0

with respect to

for

so that we obtain

a.

and

b.

3-

are

rationa~

functions in

x .

and

y.

Let us differentiate-- the first equation with respect to - y : ' ;J r

I";) y . =

a J s I ~ y + _s ~a 1 / ~ y
1

+ a 2 d p I 'd y +

p d a / 'd y

I I

Thus we derive

'

:(14.2)

'

..

,-s/';)x

;1-

a ds/'dy ... .(da /;;)y+.a )s + a t + . C>a t~y)p

(aa /dy+ a )q + (7'a-, /'dy)z -'

3
4

Similarly, by differentiating the second equation of (14.1) with

respect to
(14. 3)

x,

we obtain

b ~s/"ax+

~s/~y

(()b /dx+b )s+b r+(db /~x +b )p

+ (3b 3 /ax)q+

(;)b /ax)z.

. the left-ltand members ..o{ (14 .2) .and (14. 3) are linear forms in

.} _s/":i!' . and . "'ds/''dy ~

The determinant obtained from the coef-

.ficients
of
these linear forms is
.
.
.

We distinguish two cases:

Case

t:

Case I :

1 - a 1b1 0 ,

Case II:

l a 1b1

tri

th~s

case, we can eliminate

dS/';>x

''ds/"'Jy from

and

(14.2) and (14.3) at the same time to obtain a linear relation

r, s, t, p, q, z :

between

c r + _c 2s + c 3 t + c p + c q + c 6z
1
4
5
Suppose that from (14.1) and (14.4) we derive

(14.4)

r O(lp

whet~

olj' . . J

definition

and

O.

+ cC.2q + 3Z '

~1P ~ ~2q + .P3z ,

f 1p + 12q + Y3z ,

Yj

are rational functions in

x and y.

w~ have

~~/ax

P. ,

~ .z/3y ~ ,

and herice
Ctp/t)x . <illp+ 2q+ 0(3z '

~p/;,"1 = ~lp+ fl2q+ p3z. ,

~q/~h ~lp+ P2q+ ~3z

.f>q/~y .. Y1P+ )'2q+ 3z

By

- 89 -

Thus we obtain a system of total differential equations:

(14. s)

{:: :
dq

p dx

+ q

dy ,

(ot.lp+ ol.2q+oc.3z)dx
( ~ 1p +

(~lp+ ~2q+ e3z)dy,

$2 q + /3 3 z ) dx + ( ~ 1 P + ;r 2q + ;y- 3 z ) d y

In vectorial notations, we can write this system as

J.x

dy

.x dx+~ dy

~1dx:+P1 dy

~dx+P2dy

p3dx+t3dY

~1dx+a1dy

~ c1x+1 dy

z
d

In general, this system may not be completely integrable.

If this

system is completely integrable, then the solution space is a

three-dimensional vector space.
Case II:

In this case, we can express

-as/ ~x

and

()s/6y

as

r, s, t, p, q, z:

linear forms in

c r+c s+c t+c p+c q+c z,

6
1
2
3
4
5
d r +d s+d t
2
3
1

+ d 4p + d 5q + d 6 z

where coefficients are rational functions of

and

y.

: (14 . 1) into the right-hand members of these equations, we

f . ~StdX

= o<ls

6t;-L

+ cXzP +oi.3q +oC4Z

~s/'dy =~ls+ ~zP

Inserting

~3q

+ /34z

',

and

By using formulas such as

t

~q/dy,

'd z/"Jx =

we obtain a system of
p

di~ferential

~z/~y

equations

- 90 -

~q/~y

dq/C>x

=s ,

as/'Jx

= 0(.1 s +otzP + <X3q + 0(4 z

b s+b p+b 3 q+h4z

2
1

~s(Jy = ~ls+~2p+~3q+(34z

'

This system can be written in the form ; dz = pdx + qdy ,

dp = (a s + a p + a q + a z)dx + sdy

dq = sdx + (b s + b p + b q + b z)dy
1
2
4
3

(14.6)

ds = (1s+oc2p+o<3q+oc4z)dx+ ( f31s+e2p+~3q+~4 z }dy '

or in vectorial notations
dK

_a dx
4

a dx

b dy
4

4dx+~4dy

d
q

s
..

. .

dy
a dx .
3

a dx+dy
1

b2dy

b3dy

dx+bldy

o<2 dx+(3zdy

3 d:c+ ~3 d y

o(ldx+{\dy

:l

:j

This system of total differential equations .i s :not necessarily

completely integrab_le.

If this system is completely integrable,

then the solution space is a four-dimensional vector, space.

Let us now consider the systems of partial differential equations sa t .isf ied ,by
we shall find

F.

(j
J.
and b

1, 2, 3, 4) . . ror .each of these . systems,.

so that we may determine which of Cases ..

I and II occurs.
System. for

F :
1

We know that

satisfies (12.1).

Therefore,,

in this case we have

b

Hence we have

1 - a b
1 1

= O.

= - x/y

. ..

This means that (12.. 1) belongs to Case I.

- 91 -

System for
a

The function

F2:

b1 = x/(1-y)

= y/(1-x) ,

System for

...

al

Hence

. 1 - a 1b 1 -/J 0 . .

'

In this case, from (12 . 3) we derive

F3:

-y/x(l-x)

= -x/y(l-y)

' l - a 1b 1
From (12.4) we derive
b

'

F4 :

System for

satisfies (12.2).

F2

l.

b1 "" 2x/(l-x-y) '...,. 1 - ~ 1 b 1 1= 0

a 1 = 2y/(l-x-y) ,

This shows that (12.2), (12.3) and (12.4) belong to Case II.
Sy using the procedure of deriving systems of total differential

equations, we obtain such systems for

F , F 2 , F . and
1
3

be verified also that these systems for

integrable.

F.

. .J

It can

F .
4

are all comp.l etely

The proof of this fact will be left to the readers .

. :?hus we come. to. th.~ fo11owin~ conc1usio~.

THEOREM 14.1:

The dimension of the

.:

l_,:::
. .

system for

sol~tion

space of the

is three, while the dimensions of the

F1

spaces of the

::_ .

syste~s

..

,:

_.,.:,

for
.

.=- ..

.. .

s~lution

. '

F2 ,.F 3
.

and
.

are all

four~ .

'

Let us examine the system

for , . F. in more detai,1. . . . . . . .:. l. '.>
tiating the first equation of (12.1) with respect to

. ,or
~

Differenwe obtain

.x(l-K)cls/C>x ,+ y(l~~). c}s/Jy! LY+l~(o(+p~2)x}s .-~yt - (oC+.l)~q = 0 .

,'

'.

Similarly,

'.

differenti~ting

'

'

'

'

the second equation of (12.1) we obtain

.y(l-y)~s/~y+x(l.-y)as/ax+ [+1-(0(.+~'+2)y]s

To eliminate as/~x

and

ds/~y

- ~'xr - (+l)p'p

= o.

from these two equations, we

- 92

multiply .the fir~t equation by

1-x

and th~ second equation by

1-y

and we subtract one from the ocher.

Then we get

[ (+ 1) (1-y )- ( +e+2 )x (1-y) -<i+ 1) (1-x)+ (C4+13 '+2) ( 1-x)y ls

+ fx(l-x)r
By

- ~y(l-y)t + (ix+l)~' (1-x)p - (Ol+l)fl(l-y)q

solving this equation and (12.1) with respect to

=0

r, s

and

t,

we get
2

x (1-x)(x~y)r + ['t (x-y )-(ot+~+l)x + (o<+~- ~'+l)xy+ p y J p

- ~y(l-y)q -o<#(x-y)z

(t4.7)

(x-y)s -

p'p + (3q = 0

=0

y (1-y)(y-x) t + [ (y-x)-(a(+$ '+l)y

In

fact~

~'x(l-x)p

rewriting

+(oe.-~+p'+l)xy+~x)q

cep' (y-x)z = 0

x(l-x)r

and

y(l-y)t

in the above equati.on by

using (12.1), we get

( ('~-ix-p-l)x - (1--~'-l)y+ (j3-~')xy)s
1

- ~'y(l-x)s - /3'[)'-(o<+p+l)x]p+ pp yq+Q1.~0'z

.

.
1

+ ~x (1-y)s + ~ [ )'-(o+~ +l)y 1q

+ (ol+l)~' (1-x)p -

- ~P' xp - ot~~' z

(c<.+l)p(l-y)q = 0 ,

or
(Y-Q(-l)(x-y)s - [~'l-p' ~+l))p+ [fJ)'-/Hc<.+l)Jq

Thus we derive the second equation of (14.7).

=O

The first equation

of (14.7) is derived if we multiply the first equation of (12.l) by

(x-y)

and if we replace

(x-y)s

by

~'p-~q.

multiply the second equation of (12.l) bv

(y-x)

Similarly, if we
and if we replace

- 93 ..,; .

(yx)s

~q-~'p,

by

we obtain the third equation of (i4.7).

Prove that

Exercise 1.

(x-y)s ~'p +pq

=0

has a solution of the form

where

In particular, if we take

is an arbitrary function.
'1-( ) = (cx,m)
l m
,m)

<r

we find

again.

If we write the system (14.7) in the form

Exercise 2.

(14 ]I)
{

= (;l(.lp+c(2q+3z

'

= ~lp + e2q + ~Jz

'

.t ""

Yip+ r'2q + Y3Z '

we can reduce the system (14.7) to a system of total differential

equations for

z, p, q

= l,

(z j ' p j ' q j ) (j

Let

by the procedure given in this section.

2 ' 3)

total differential equations.

zl

z2

Z3

P1

P2

P3

be solutions of the systems of

Prove that

B' -r y p-r
r-~-p-1
. (1-x)
(1-y)
= const.x

r-o<- ~ -1 (x-y) -p-p'

ql q2 q3
If we write (14.7) in the form (14.7')., the coefficients
~

and

j.

l.
J

have poles on the lines in

~

0,

= 1,

y 0,

y 1

C x (:
and

0(. ,

- 94 -

F1

This implies that any solution of the system for

is regular

analytic in

c ~ c-{x.

= o~

u {x =

1}

u/y = o~

v{y

qu{x

Y)

However, solutions are in general multiple-valued functions .

y

x
/

In the same way, if we write the systems for

F , F
2
3

and

F
4

in

the form

r = a s + a p + a q + a z ,
1
2
3
4
t =b s + b p + b q + h z
1
2
3
4
~s/ ~x =

'1 s + o<.2p + o(3q + 4 z '

. 'as/~y eels+ ~2p + ~3q

then the coefficients have poles in

o,

.=

= 0,

x .. 0,

x .... 1,

y = 0,

= 0,

1,

'I ..

o,

x+y = l,

+ p4z ,

x (:

= 1,

y = 1,
(x-y)

at

x-*:y = 1

xy-x-y = 0

for
for

F2

,
F.3

2 -2(x+y)+l ~ U for

F4

- 95 ..

. y

I\_

Remark 1.

A natural compactification of the complex plane

.

is the Riemann sphere which is the unit sphere of real dimension

2
two, i.e. s This can be regarded as the complex projective line

P1 .

On the other hand,

The projective plane

IP

C x (
2

has two natural compactifications.

is one of them.

The other is

e1 x

1
1P

Both of them are compact c.o mplex manifolds, but they are not bi-

holomorphically equivalent.

This means .t hat there is no biholomor-

phic mapping from one into the other.

They have, however, a common

modification and hence they are equivalent in a certain sense.

F.

Since coefficients of the systems for

are rational in

and

these systems are well d~fined in the compactifications of

y,
(

l(.

In various cases, a choice between two compactifications of

is not a serious problem.

Remark 2.

The Gauss differential equation

x(l-y')y" +
is

~educed

to a system

[-(+~+l)x}y'

'Jlpy

= 0

- 96

( 14. 8)

f yz''
.l

z/x ,
:c

xi3

x-1

Y+ (-1_-_r + r-1)(-8-1]
x

x-1

'

if we put
z

= xy'

Note that

z ' xy" + y '

and hence

x(l-x)z'

= x[x(l-x)y'I +

(1-x)xy'

a:o(~xy - LY-(o<+~+l)x)xy'

(1-x)xy'

-= fl(flxy + [ (1-)')+(oe.+~)x]xy' .
The system (14.8) can be written in the form
(14, 8 I)

where

Note that coefficients ' of (14.8') have

si~ple

now .that the system (14. 7) or (14. 7 ') for

system of an analogous form.

z ,

as unknown quantities.

poles.

We shall show

can be reduced

To do this, let us take

x;)z/ax,

y )z/ <> y

Then

:x[xtiz'tax] ,.. xr+p ,

~
ay
(x'iJz/ox)

=XS

Ln

- 97 -

+cyaz/dy] = yt + q
. *Y

Therefore, in a way similar to that in the case of the Gauss equations, we derive from (14.7) the following system:

d ~)z/)x

{A~x + Bgy_ +C

y .

dx +ok-+Ed(x-y))

x-1

x-y

y-1

y ~z/ ~y

x';>z/Clx
y

;Jz/ 'dy

where

1-1+13'

-(3'

D=

-~

. (!>'

-~

1-)'+ ~

-p

i-oc-jl' -

Note that the coefficients of this system have simple poles at

x

=0

=1

0 ,

y ... 1

and

y .. x

 98

CHAPTER III
Monodromy Group of the Gauss Differential Equation

15.

Euler transform.

In Section 3 (Chapter I) we derived the

Euler integral representation of the function

(3 .1)

F(o(,e, '(,x) -

a(+

1,

(1-xu)

-ar

f(u) =

Dropping the constant

let us write this integral .in a form

= ~ (1 -xu)~-l

y(x)

A.= -

where

y, x):

rcr>

r<~)/r(~)r(r-p),

(15.1)

p,

1
"-1 _ r-~-1 _ --cc
r(~)r(l-~) Ou
(1 u)
(1 xu) du.

The kernel of the integral is

factor

F(oc ,

t(u) du,

u~-l(l-u) r-~-l and

C = [O, l} .

This integral (15 .1) is a solution of the hypergeometric differential

equation
(15.2)

L(y)

x(l-x)y"

+ [ 't-(OC+~+ l)x]y' -

o<~

= 0.

We shall find other solutions of (15.2) also in a form (15.1) by

choosing

)\ , qi , C . in suitable ways ~

Suppose that

we

can differentiate the function

by (15 .1), with respe.ct to . x

Then we have

under the integral to obtain

y'(x)

f
=Jc

y" (x)

y(x),

~x (1-xu)

'i>

~-1

-2
'ax

(1-xu)

f(u)du,

X-1

T(u) du.

give~

- 99 -

L(y(x))

ic

xu)

L((l -

4'-1

) f(u) du,

where

L((l~~m)~-l) =
' .' ,

x(L-x)[(,\-l)(A-2)u 2 (1-xu).\-J]
.

,\ 2

rr-(O(+~+l)xJ[-()\-l)u(l-xu) - ]

..

al~(l-xu)

'

L((l-xu)A-l)

Let us write

L((l-xu)

.\-1

A.-1

in the form

) "' (1-xu)

A-3

H(x, u, ...\),

where
H(x,u, ~)

(~-1) (~-2)u

The function

of

in

2
x(l-x} - (A-l)u[ t-(oe.+p+l)x] (1-xu) - 0<(3(1-xu) .

H(x, u, ...\)
H

is quadratic in

x,

and the coefficient

is given by

[-{A-1) (A -2) - (~ -1) (ot +~+l) -

a<p Ju

Note that
(A -1) ( >.. - 2) +

( A -1) (o< + p+ 1) + o< (3 = (~ + o( -1) ( J\ + /J

. Therefore, if we choose
becomes linear in

)\

-o(+l

or

x.. We . shall fix

=:

-~+l

-1)

the function

. .iri this manner.

Since

the hypergeometric differential equation (15;2) is symmetric with

respect to

o<.

and

In case when

. A

, we shall consider the case when

-o<

+ 1,

we have.

H(x, u, .: \) =ct [{ (ot - +l)u 2 - .(-' +l)u } x+ ru - ~]

= oC. ( - ( c( +1 ) XU ( 1 -u) T
and hence

( 1 -xu) ('1 U - /J ) ) ,

A = - + 1.

- 100 -

L(y(x))

o(lc (l-xu)-2 [-(cc.+l)xu(l-u)+ (1-xu)(~u-~)} j(u)

o(

ic

:u

ol ( [

Jc

+oe.{

-Ol-1

4-l

(tu-f1)<j(u)]du

(1-xu)--l()'u-p)<f(':l))d~

u(l-u)t(u)JC

~~ /~(1-u),(u)l+

(1-xu)_'_ [

Therefore, if we choose
.

u(l-u)f(u) + (1-xu)

{(1-xu)-o<-l} {-u(l-u)j(u)} +

[-(1-xu)

(15.3)

--2

[-(oc+l)x(l-xu)

du

and

(Ju-p)1(u)) du

C by the conditions
.

Tu [u(l-u) f(u)] + (l u-/0 ~(u) = 0

and
(15.4)

[{1-xu)

then we have L(y(x))

(15.5)

-e<-1

= 0.

'f(u)

u(l-u),(u)]C

= 0,

The function
4-l

= u 1"

(1-u)

satisfies the condition (15.3).

~-4-1

'"

Let us fix

bYi (15.5).

Then

the condition (15.4) becomes

~

(15.4')

. [u (lu)

This means that, if

C is a closed path on the Riemann surface of

"

(15.6)

r-n"'(1-xu) --1 1c ...

u"'(l-u)

t'(l-xu) -ot.-1

r-~

or if (15.6) takes the s.ame value at the starting point and the
end point of
we can fix

C,.

the condition (15.4') is satisfied.

C in the following ways:

For example,

- 101 -

(i)

the path joining

to

(ii)

the path joining

-oo

(iii)

the path joining

to

to

if

Re~>

if

+oo if

0,

Re~>

Re( - {J) :> O;

0, Re(o<+l-r) > O;

Re( -(J)> 0, Re(o<+l-.r)

>0 .

On the. other hand, Jacobi showed that the following three

curves satisfy our requirements under certain conditions on the

parameters:
(iv)

the path joining

to 1/x :

(v) the path joining

to

l/x ;

the path joining oo

to

l/x .

(vi)

In order to find the conditions on the parameters that these three

curves satisfy our requirements, we must examine not only the

cor.dition (15.4'), but also the assumption that

y'(x)

=Jc

y"(x)

ax

(1-xu)

~-1

'(u) du '

. In deriving the conditions on

and

C,

we

actually assumed

.that the order of integration and differentiation can be interchanged.

This requirement m ust be satisfied by the three curves

(iv), (v) and (vi).

d ff (x)
d; ja . F(x,u)du
Therefore, if

Note that we have

Jf '(x) ...L

= .a

F(x,f(x))

i)x F(x , u)du+ f' (x)F(x,f(x))

0, we get

- 102 .

d Jf(x)

d; .
a

F(x,u)du

Jf

<j>(u) ,

other hand,

if

A-1

f(x) ~ l/x

then the formula for

= -~,

and

F(x,u)

is verified .. On the

y' (x)

F(x, f(x)) = O for

we have

3
. ~-1
F(x,u) = ~ (1-xu)
<y(u) ,
Since

~x F(x,u)du

a-

If we have F(x,f(x)) = 0 with

.
~-1
(1-xu)

(x)

f(x)

= 1/x

y" (x)

the formula for

and

.
is verified.

these two conditions are satisfied, if

{15. 7)

Re o<.

< -1.

In order that the integral (15.1) is well defined, it is sufficient

to assume that

{ Re ~
(15.8)

>0

Re ( 1 -

(3) ;>

Re ( o<. +1- )')

Re

o,

if

is the path (iv),

<1

if

is the path (v),

Red(< 1

if

c is the path. (vi).

<l

Re o<

> 0;

Furthermore, the integral is holomorphic with respeL:t to

and 'j

in the respective domain (15. 8) for each case.

that the condition (15 .4') is satisfied, it

is

oc.

(J

In order

sufficient to assume

that

{Re ~ 0, Re.< -1
;>

(15.9)

Re (a' - -~ ) ) 0 ,

>

Re (0( +1- J ) 0 ,

Reoc.<-1

ltt- r'-<-1.

if

is the path (iv),

..

if

is the path (v).

lf

is the pa th ' (vi) .

The ref ore,, the integral (15 .1) is ho lomor phic for (15. 8) and
satisfies the hypergeumetric differenti.;il .e quation for (15. 9)
Since the domain (15.8) contains 1he .domain (l~!~) for ea~h _case,
the integral (15.1) is a solution of (15.2) for (15.8).

Thus we

- 103 ...

proved the following result:

THEOREM 15 .1:

If we put

U(x,u) u 6 - 1 Cl-u) 1 -~-l(l-xu)-

(15.10)

then the following integrals are solutions of trw hypf'rgeometr i. c

differential equation (15.2):

i
.so
t

FOl(x) =

F~ (x) =

if

U(x,u)du

if Re,,>O,Re(o<+l-r)>O '

Rep >0, Re(!-/3)

00

(x)

100

tA

U(x,u)du

if

Re('(-{0>0,

Re(o<+l-~)>O,

(15.11)

fl/x

>0 ,

U(x,u)du

(x) =
1

.Ox

J.

U(x,u)du

> 0,

if

Re~

if

Re(c.X+l-,r)

<1

Re Q(

1/x

F i(x) =

tx-

F1

(x)

x oo

= J:

U(x,u)du
00

U(x,u)du

> O,

Re<X

<1

l/x

ln particular, all the conditions on the parameters are satisfied

if
(15.12)

The function

< Re ~ <

U(x,u)

. oo as a function of

u.

the upper half-plane and

Re

< Re ( o( + l) <

has singularities at
Suppose that

l{x

Im x

> O.

2
u

= 0,

Then

1, l/x

is in the lower half-plane.

precisely, we make the convention:

is in
More

and

0<argx<7r:,
-1t. <

.
l
-7<arg-<O,
x

and

-7t<arg(l-x)<O

arg{-x) < 0 .

Let us

the paths of integration in the following manner:

spe~ify

, the path is the line-segment joining

01
u 1. along the real-axis;

For

(ii)

for

FooO' . the path is the negative real-axis which joins

u -

- OJ

. u '=

+cio.

=1

to

along the positive real-axis;

l ,

the path is the line-segment joining

=0

l ,
1x

the path is the line-segment joining

=1

Fl '
-oo
x

the path is the half-line joining u

. ox

(v)
u =

(vi)
c

~loo.' . the path is the half-line joining

=. itx'

to

to

u = O;

.J:~r : F;

(iy)
to

to

for

(iii)

u =0

. (i)

co

for

1/x;
for

in the direction:

= l/x

to

arg(l/x) .

the u-plane
co

In order to define the integrals (15.11), we must also specify

the branches of the function

U(x,u)

given by (15.10).

this function is multiple-valued with respect to

u~

Note that

To fix a
'-.

- 105 -

-- branch of
~ each

it is sufficient to

U(x,u),

factor

and

u, 1-u

of integration.

1-xu

det~rmine

of the function

the argument of
U

on

th~

paths

We shall. fix those arguments in the following

manner:
For

F 01'

arg u

= o,

arg(l-u)

= O,

- 7t ~ a rg ( 1-xu)

(ii)

for

FooO'

arg u ,., 'lt',

arg(l-u)

= 0,

(iii)

for

F loo'

arg. u

= 0,

(iv)

for

arg u

{i)
"

ol'

-7t

0,

o-a arg(l-u) "'1t ,

(v)

for

(vi)

for

F 1,

-1t ~ arg u "' 0,

Fl ,
-oo

-1t ~ arg u ~ 0,

. 1x

arg(l-xu)

0;

~ 7t ;

-7t ~arg(l-xu) ~ 0;

arg (1-u) :.: -'It",

~

0 ~ arg(l-u)

'

O" arg(l-u)

= O;

arg(l-xu)

-lt ~ arg(l-xu) ~ O;

1l7,

~ 1'C,

arg(l-xu)

= - 7t

As we have shown before, the integral

integral representation of the function

01

F(,

(x)

p, r,

is the Euler
x)

multiplied

by a constant:

rep> r< r-A >

re r>
In other words, we have

F01 ()
x

rcl}>r<r-n>F<r <">
- , ,A. , 1 , x >

If we make the change of variable

u .. v(v-1) -l

the second integral

F (x)
040

F~o

(1 ~ v ~ O) ,

becomes

o [v(v-1)
. -1 ] ~-1[l-v(v-1)
- , -1 r-~-1
-1 :..'
J
[l-xv(v-1) ]

S1

-~v
(v~l)

- 106 -

arg v = 0

Let us suppose that

from

1.

to

Then, since
v(v -1)

arg(l-v) = 0

and

=7t,

arg u

as

goes

we must have

= e1ti. v(l-v) -1

-1 .

and hence
[v(y-l)-ll ~-1 ... e1d..((J-1) "~-1 (i-v)-S+l .

= O,

arg(l-u)

On the other hand,

arg(l-v) = 0

and

1 - v(v-1)

imply that

(1-v)-l

.Cl - v(v-1)-11 cr-s-1 ""

(1-v) r+p+1.

If we suppose that
0 ~ atg[l - (lx)v] ~ 7C,

(15.13)
then we get
(1 - xv(v-1)

-1

-0(.

[1 - (1-x}v]

-oe.

0(

(1-v)

since

l - xv(v-1)

-1

= (1

- (1-x)v] (1-v)

._l

and
0

arg(l-xu)

~ 1c

The assumption (15.13) is justified by the convention:

arg(l-x)
arg x

<

<
TI:.

O.

Note that we assumed at the beginning that

6120

<
0

<

Thus we have
1

-1t.

(x) =

1Ci <~ -1) "-1

~- r
"""dv,
vP (1-v)
[1 - (1-x)v]

where
arg v ~' 0,

This means that

arg(l-v) = 0,

0 ~ arg[l - (1-x)v) ~ 11:

-1

.. 107 -

rca
> r<Oi+1-v>

r(Q(+p+l-)
F(ot, (3,o<+~+l-Y, 1-K).

= e1d(~-1>

F000 (x)
Similarly,

f (v- )~- (1-v-l)-,4-l(l-xv-l)-(-v- )dv

0

F co(x) =
1

-=

f,

1
v

-~+1

[e

-7tl

(1-v)v

-1 i-~-1

[(-xv

-1

- -2
)(1-x v)] v dv
. -1

e -tl(.r-~-l)(-x)"'o(

1
.
.
1
v'-r(l-v)r- ti- c1-x"' 1v)-ll(dv

fo

-7tl(r-6-l) r<o<-1+1)r(t-fl)(-x}-~(oc QC-"'+l J-tt+l l/x)

r(cX-p+l)
.
~, . '"' r ,
,

= e .

-7ti~ r~ >rn-"'~
-1&
. rp-+l)
(-x) F(~-)'+1 , ~,/J-.+1,1/x)

=e

. .

-1
~-1
-1
Y-~-1

-ct.
1-x
F (x) = ) [x (1-(1-x)v)]
[1-x (1~(1-x)v)]
[l - (1 - (1-x)v)] (-7)dv
1
11
x
..

-=

[x

-1

/3-1

(1-(1-x)v)]

ri -1
'r-P- 1
-CK _:1
x (1-x)(l-v)]
[(1-x)v] x
(1-x)dv .

[e

eid. <r-P- 1 >f 1 xl-Y (1-x) r-o<-t1v -ct (1-v) r~.:. 1 [ 1-(1-x)v1~- 1 dv

Jo

., 1Ci(r-~-l) C(l-oe)r(J-(l) 1-t(l- )'1-rL-~F(l-~ 1 _ ~- _a+ 1 1 _ )

e
. r()'-Dt-~+l) X
X
I"'
o(, a rJ
, X

- 108 -

e7ti.(-p-l) f(l-~)r(r-O)(l-x)r-~-~F(Y-~ -~ 1-~-~1 1-x)

rcr--~+_1>

F1 (x) =
(v
-~
. 1
x
1

a:

f,0

f,

(v

1
x

'

-1 -1 (3-1
-1 -1 (-~-1
-1 -o(
-2 -1
.
x )
(1-v x )
(1-v ) (-v x )dv
.

-1 -1 (3-1
x

[e

Jti

(1-vx)v

-1 -1
x ]
.

r-p-i [e -ti (1-v)v

. -1 -at -2 -1
] v x dv

t-~ v ~~ ("v) - (1-xv) t-~-1 eu(t+oe-p-1) dv

1

~ eiti.(r+Dl.-~-l) C(Q{+~(I~r5 1 -0(> x 1 -~F(p-r+l,+1.;.,2-t,x).

We shall summarize these results in Table 15.1.
The integral of the form (15.1) is transformed by the change
of variable
u

= l/f

into
(15 .14)

where

'l'< 5) .= -(~, ~ )1-A t<.~ -1) S-2

and

is the image of . c . by the change of variable u

f:

The integral operator which assigns to

or

'f

the integral

(xs>HifC~)d~

= 1/

s.

109 -

is called the Euler transform.

The integral
b
.
.
.
8-1
~-(1-1
'-ot
a u
(1-u) . . (1-xu)
du ,

where

a, b

= 0,

1, 1/x,

&0

const.

where

c, d

= 0,

u = 1/5

is transformed by

into

~-r
r-~-1
-~
s
(1- s)
(x- s) ds
J

1, x, oa .

From this we derive the following

theorem.

THEOREM 15.2:

then
(15.15-1)

f:

ECx,

S)d5 = const.

c(x,

s)dS = const. F(o<.,~,(,x) ,

~(x,

5. )~ ~ = const. F(~, p,ae+p-r+l, 1;..x)

S)d S

= con$t. x

5)d5

= const ..Cl-x) r:.-.-p F(l-o<,l-p, t-o<-~+1,1-x ).,

(15.15-2)

itio1

(15 .15-3)

Jo

(15.15-4)

s:

ECx,

(15.15-5)

L:.(x,

eo

x ...

(15.15-6)

If we put

Ix' ECx, SJd ~

(-x)-"F(o<,o(-+1,o<-f!+l,l/x) ,

1-r

F(~-+1,

oc-)'+1, 2-'t, x) ,

const. (-x)~F(p, p-r+l, p ....+1, l/x)

.. 110 -

are all solutions of the hypergeometric differential equation

(15.2).
Supposing that

Im x

> O, taking the paths shown in the

figure given below and fixing branches of

the formulas (15 . 15).

E(x,

5),

we can verify

.. 111 -

16.

Connection formulas and monodromy group of the Gauss dif-

ferential

Suppose that

equati~

(16.1)

0 <Rep

<

Re "j (

<

Re(O(+l)

and

{16 . 2)

< arg

-7t.

.(7t.,

:1Then the six integrals

01

, F

<

103
x

arg(l~x)<

FeaO' F

0,

,
11
x

-re<

F lCCJ' F
0

arg(-x)

< O.

are well

defined and we have three fundamental systems

{F
' Fl
01
-x GO

t ' . {FooO ,
.

F i
1x

l'

{F

loo

, F 1}
o-x .

of the hypergeometric differential equation

x{lx)y"+[~-(o(+~+l)x]y'

(16 .. 3)

- o<~y

0.

:. In this section, we shall find relations between these three

fundamental systems.
Consider first three integrals

FMO'

01

: and

F M.
1

Consider

also the following curves in the u-plane:

a line

-R

u ' -r,

a semi-circle

c2

a line

t-2

a semi-circle
a line

where

= re i

(n:-a)

>r >0,
where

r ~ u ' 1-r ,

1+r

a semi-circle

= 1 + re 1 (1t-&),

where

0 ~ 6 ~ 7t,

R ,

u =Re

i9

' 7t .

0 ' 6
These six curves form a closed path in the u-plane.

Let us denote ,

- 112 -

''
'

C1

-r

R
We

sha~l

fix a branch

1-r

of the function

ul (x.,.., u)

this branch is single-valued .on

arg(l-xu)

u (x, u)

on

the curve

arg(l-u) = 0 ,

arg u =it' ,

on the segment
on

This
arg(l-u)

~r

Hence

is uniquely determined by the conditions:

(16.4)

(i)

change cbntinuously

arg u,

so that

u)

CRr and its interior.

condition is equivalent to the condition that

and

U(x,

~l'

c1

arg(l-xu)

~7t

Under the assumptions (16.4), we observe that

arg u

decreases from

7C

starts from

arg (1-u)

to

0,

0 and comes back to

after taking negative values,

arg(l-xu}

changes continuously from a positive

value to a negative value;

(ii) on c2'

arg u = 0,
between

at
(iii)

on

12,

arg(l-u)

-1t

= 0,

and

arg(l-xu)

and takes a negative value

and

= 1-r

u 1-r ;

arg u

=0

arg(l-u)

-1t: :.;;

at

changes from

arg (1-xu)

' 0;

varies

and
0

l+r,
to

-:rt

. 113 -

on

(iv)

(v)

on

C3,

arg u

r-3,

arg u

arg(l-u) = -'7t,

0,

:::

changes from

the initial value

to

(i.e.

changes from

arg(l-u)

arg(l-xu)

- 7t 1i: arg (1-xu)

and comes back to

7t

7t.

-11:

' 0 ;

),

to

0,

increases and arrives at the initial

positive value.
By virtue of Cauchy's theorem, we have

ul (x, u) du = 0 .

CRr
Let

tend to zero and let

lim

J .u1

tend to infinity, then

(x,u)du = 0

(j

= 1, 2 ,

3)

1j

and

lim

u1 (x, u) du = F000 (x) ,

ul (x, u) du = F 01 (x) ,

cl

lim

. c2

and
. lim (

u (x , u) du = F ~ (x)
1
J.O> .

le3.

Thus .we obtain the following relation:

F 000 Cx)

01

(x)

+. F 100 (x) = o

Consider next three integrals

000

, F

case, we shall use the following curves :

ox1

In this

- 114 -

u ~ -r,

a line

-r-1 :

a circular arc

c2 :

a line

~2

C3

lJ :

-R

'

cl :

u :;;;

= re i9 ,

R >r > 0 ,

where

where

f exp(i arg(l/x))

1t~

arg(l/x)

~ f ~ lxt- 1 -r

< 0 ,

.J

-1
i6
-1
-1 .
a semi-circle u = x +re , -n:+ arg(x ) ~ e ~ arg(x ) >:
.
-1
a line u = exp(i .arg(l/x)) , tx l + r ~
R ,
-i0
-1
a circular arc u = Re
,
-arg (x ) ~ 6 ~ 7t:.

Determine

u (x, u)
1

f"

uniquely by taking
-1C

< arg
0

1
u,.. arg(x- )

< arg(l-u) <

<

:).

O ,

n:..

arg(l-xu) . = 0
on the line segment

c2 .

Letting

r -> 0

R --> +oo ,

and

obtain
e

-2rcift F

ooO

.l.

ox

+ e -2.1tiotF 1

xi)()

-R

>,, 1/x .

"" ...
Rlxl/x

1"3
"-

Similarly we derive'---.._. ____ _____,.....

F 01

and

F 1 - F 1
lx01r

we

integral
1

s
0
0

0()

00

J
1

1/x

F 01 (x)'"'

F.,0 (x)

---

Fleo(x)""

.. F l (x)==

O;c

l/x

[ .;c, OJ

..

OJ

-7ti~

'

identification

' , ,

-------- -

r(o{rf:!-i{1({f>} (-x)-oCF(ot,ot-1+1,_~+l,1/x)

0('

-~ i- -p+11 )
' oc
' -'X

re~ -'l~(li!"r\1-) xl-)'F (ol-1+1, ~+1-)' '2-)',x)

r(l-ot>rcx-n>
<1- >"-PFcircr-0(-p+l)
x

r<8>
ro-ocL <-x) -PF <p ' 6-~+1 ' ~-t+l ' 1/
>
r(~-0(+1)
. x
1ti(t-~-1>

eTCi ((+ Oi-(3-1)

e Ki(f-(3~l)

e'1ti(i'l) f<@)r(ee.+l-~) F( p c<+~+l-11- )

f(c<+~+l-)
' '
. ' x

r<~>

r(l3)r(-ft) F(ae ~ y x)

Table 15.1

ul/(vx)

ue[l-(1-x)v] /x

u=v/x

uol/v

umv/(vl)

arg u arg(l-u) arg(l-xu) transformation

0

[ 0' 1t]

[ 0 ;n:]

[ 1t,

O]

[O,n:) .

.. '1t

[~tt,

[ 0, 7t]

- 7t

7t

[-7c:,OJ

( -7t, OJ

F.l(x)sJ [-7r,O]
1 x
1.
eo

1/x

Fl (x)a )

x 00

- 115

-2'1ti(r-p)
F1 0 .
.. Floo + e -2u(r-~) 1 + e
1x
x"

THEOREM 16.1:
and

F 1
0x

The six integrals

FOl' Fl-oo, F~O' F1l' F loo

x
satisfy the following four relations:

l .

-2'71:Ll<

-2n:i.,9

.0

-1

e 21ti(~-'t) -1

FOl
F1

x-eo

Faoo

(16. 5)

1.

e 27ti(~-r>

- 0

F 1

1i"

F loo

F 1

Oi"

It is easy to verify that e.very four-by-four submatrix of

the matrix of (16.5) has rank four.

Hence, if we select.any two

.integrals from the six, then the other four integrals can be expressed as a.linear combination of the two . Such relations are
called connection formulas.

For examp_le, we have

l-e -27ti (~-oc)

e -2?t:i _
1

,.. e -270.(Y-"."~)-l FooO + 1-e -27ti(+~-r) F1 ~

On the other hand, Table 15.1 shows that

F 01 (x)

F600(x)

= enivg-1)

r(f>){l)- (4) F (ol. ' ~, i'' x)

r(t.1) r(O'+l-y) F(
rc+~+l-r)

A
Qt, ,.. ,0(

+a v+1 1
, .. -a

-x

- 116 .._

and

F l (x)
1-:-

ti(tr-~-1)

rCl-o<)r<r-e>
t-~-p
.
r( t-~-p+l)
(1-x)
F (1-, - ~, --~+ 1, 1-x) .

Therefore

(16.6)

F(oc.,~,~,x)

rcnr
c~-a1.-M

f(;J-o<) f(-r-~)F ( o<, p,o<+ p-~+ 1, 1-ic:)

r<1>r
Co<+/3-tO (1-x) 1-ot-L
.
1 -F(~-ot, r-p, -oc.-~+1,1-x)
r()r(~)

Note that
sin(n:( a'-o<)) .

= sin(7t("(( .. ot.- fo))

'

27tioc

e
-1
e77:i(1-~-l)
l-e -2n-.i (o<.+~-1)

sin(7toc)
sin (7t(+~- ~))

and
sin(7ts)

The formula (16.6) can be obtained also in the following manner:

Set
F ( oc , ~ ,

'(, x) = AF ( oe ,

p, o<. + (3 - +1 , 1-x)

+B(l-x)'t-ot-~F(1-o<,1-~, 1--~+l, 1-x).

Let

tend to

Chapter I)

1.

Then by virtue of Theorem 3 . 3 (Section 3,

we get
r<ll>rca--ot-~)

re 1-> re 1- ~>

=A

under a suitable restriction on the parameters

Next let

tend to 0.

0(

'

/3

and'

r.

Then again by using Theorem 3. 3, we get

- 117 '.-

1 ... A

r<cit+-t+l)r(l-~)

'

r < - + 1 > r <~ - +1 > +

8 rcr-oc-f3+l)C(l-)
re i- >r <i - p>

'

From this we obtain

B =

In Section 11 (Chapter Il) we derived a connection formula

(11. 7) which represents

F (ell(,

p,

', x)

as a linear combination of

(-x)-ocF(O(, +1-, o(.+1-p, 1/x)

and
(-x) -pF ( 0, ~+1- r, ~+1-cx, 1/x} .
(Cf. Theorem 11.4, p. 74.) . As this connection formula holds under

the condition that

(16.7)
.' all connection formulas of this kind also hold under the condition
(16.7).

If we know one of

t~ose

connection formulas, then other

formulas are derived by using changes of variables and by using

the identification of twenty four solutions of Kummer.

- 118 -

17.

Monodromy representations of the Gauss differential eguatio.Q_.

This section is a continuation of Section 6 (Chapter I).

we

tion 6,

the

defined

In Sec-

monodromy representation of the Gauss

differential equation (6.1) with respect to a fundamental system

y}.

{ <j>,

We also stated Theorem 6.1 (p.39) without proof.

In this

section, by using the six integrals (15 . 11), we shall compute

monodromy representations of the Gauss differential equation explicitly.

As in Section 6, let D
1Cl (D, .x )
0

denote by

point

x .

1 (D, x 0 )

By the definition of

7t (D,

x )
0

x = 0

not encircle the point

not encircle the point

Then

E D.

We

D with the base

a homotopy class of loops

Let

1. 0 be a loop which

once in the positive sense, but does

= 1.

De.n ote also by

L 1 a loop which

x = 1 once in the positive sense, but does

encircles the point

.l.O and

i~

which was given in

"7l:'

D which start and terminate at

encircles the point

and let

the fundamental group of

Section 6, an element of

in

= C - {O, . l}

= 0.

The homotopy classes containing

.l 1 are denoted respectively by ( .t 0 ] and [ .l 1].

7rl (D, x )
0

{cf, "t}

Let

is a free group generated by

be

a fundamental system

l ol

and

equation . Then the monodromy representation with respect to

i 'f, 'f}

is a homomorphism:

f:

7C (D, x )
1
0

~GL(2,

,t 1 ].

of the Gauss .differential

I
I

C).

... 119

:\ . This homomorphism

.. . (17 .1)

is uniquely determined by two matrices

= p<C -lo1),

The matrices (17.1) . are called the circuit matrices around

and

=1

with respect to

{ ~-,

y},

respectively.

present two methods for computing A and A1

0

=0

We shall

in this section.

The first method is based on the connection .formulas

Method I:

among the six integrals (15.11).

0

< Re ~ <

Re_1

Suppose that

<

Re ( +1)

We shall use the fundamental system

oi (x)

Im x

> 0 and that

<2

{ F 01 , F 1 }
x(lo

= const. F ( oc ,

Note that

p, 1, x)

and
F

(x) .,. const. x

1-1

F(O(-C(+l,

~-+1,

2- '(, x)

jf 00

The solution F (x)' is single-valued in the neighborhood of

01
-21ciY F
(x)
if it
x = O, while the . solution F
(x)
goes to e
1
1
.
-oo
-t>o
x
x
is continued analytically along

Therefore, in this case, we

have

in order to compute

A ,
1

we use

th~

following connection formulas:

- 120 . -

where
...
-27Ci
"' e

<r- "'.:t) - e -27tUic

.

By using the fact that

F.., (x) .= const. F(ot,
0

F 1 (x)

,,.+p-r~l,

1-x),

.
r-c1.-"l"F(t-ac, ~-(J,r-ot-(3+1, 1-x) .,
= const. (1-x)

ii'

we derive

C is the matrix of the linear relation (17.2) .

where

...--- ~

..... .

('Method II !i
-

... -----::.~-:.::::: _/ .

Assume aga.in that

{Fdt'

fundamental system

(~1.3)

.j

=
'01 (x)

ix-

ix-

1 ~... 1
.

fo

u .

l/x

F l (x) =

F . 1 },

f1

(1-u)

Im x

>

0.

We shall use the

where
. 0 -1
1-oc:
~
(1-xu)
du ,

u~-1 (1-u) "I- -1 {1-xu) - du

The paths Of integration are fixed in the same way as in Section 15.
Let us consider first the analytic continuation of . F

01

along

l2l -

~i

: 0

We must investigate how the quantity

We can assume that
a .l in.e -segment joining

. (t)

lxl

r0,

where

r0

arg(l-xu)

changes

.(.0 consists of two parts:

to

ir

and

the circle

(ii)

is a small positive number.

(See Fig. 17.1.)

Fig. 17.1
; Now it is easy to see that the change of
(-

. loop

is zero.
0
continuation along

Hence

01

(x)

arg(l-xu)

along the

does not change by the analytic

10

Let us next consider the analytic continuation of

.f1 .

As

moves along

encircling x = 1

.l 1 ,

the point

in the positive sense.

l/x

01

along

moves along a loop

(See Fig. 17.2.)

In

order to keep the same branch of the integrant for the integral
:. F 01'

1/x

we must deform the path of integration so that the point

is never on the path of integration.

along
path of

Let

1. 1 at x = x 0 . As x moves a long
integration~

When

x start moving

.l 1, we deform the

comes back to

the path of

integration becomes a curve shown by Fig. 17. 3. This curve can

be further deformed into a curve
not difficult to show that
F

sho~n

F 01 (x)

01 (x) + (1 - e

by Fig. 17.4.

changes into
-2~

)F !.(x)
1x

It is then

~-

- 122

1 1

by the analytic continuation along

- 1/x

0
Fig. 17.2

Fig. 17.4

Fig. 17.3
'

Let us consider the analytic continuation of F l

As

moves along

encircling the points

Fig. 17 .5.)

the point

and

tx-

along . J.
0

l/x moves along a loop

in the negative sense .

We must deform the path of integration for

that the points

and

are never on the path.

(See

F 1 . so

ix-

When

after moving along ~ .f. , the path of inte0

0
gration becomes a curve shown by Fig. 17.6. This curve can be

comes back to

-further deformed into a curve given by Fig. 17.7.

'

difficult

~o

prove that

F 1 (x) changes into

.
.1x

{ 1+ e - 27ti~} F

01

(x)

by the analytic continuation along

+ e.- 2 ~F ~ (x)

.e o

It is then not

- 123 -

1/x0
Fig. 17 .5.

1/x0

Fig. 17.7

Fig. 17.6
It is easy to see that

F i(x)
1x

changes into e 2'1ti(t--~)F 1 (x)

1i

. by the analytic continuation along

ll.

Thus we obtain

A0 ""

(-(1-e ~ 2~)

_:rlr

and
l

=
0

1-e

-21tiOC

e -2?ti (O(+p-r>

(Cf. Theorem 6 .1, Chapter I, . on p .39.)

- 124

CHAPTER
---
-- . .

t J

-~--

Solutions of Systems 0f Partial Differential Equations

for Hypergeometric Functions of Two Variables

18.

F .
1

Solutions of the system for

In Section 12 (Chapter II)

we derived the systems (12 .1) ~ (12 .2), (12. 3) and (12 .4) of partial
differential equations satisfied by

F , F , F
2
3
1

and F

respec- ..

Let us consider the system (12.1)

tively.

x(l-x)r+y(l-x)s+ [y-(ol+~+l)x]p -~yq -o(~z = 0

(12.1)
.
. . { y(l-y)t+x(l-y)s+ (1-(oe+p'+l)y]q - ~'xp - O(~'z = 0

In Section 5 (Chapter I) we proved that the Gauss differential

equation has twenty four expreiraions of solutions of the form
x

f1

( 1-x)

1.

and

x.

1.

x,
(Cf~

.
F (Ol , ~ , 1 . , x . )
.

1.

r.1.

and

&'. ,

tr'i

l"i.

1.

are linear in

()(

and

is one of the six transformations

1-x,

l/x,

1/(1-x),

Theorem 5.2, p.31.)

(x-1)/x,

x/(x-1)

These solutions are divided into six

groups and any two solutions of each' group are linearly dependent.
It is known that the system (12.1) has sixty expressions of solutions of the form
x

where

(1-x)

'"t y fi. (1-y)'l (x-y) 'ti F

f 1,

tr1 ,

f1.

f.j_, <r.j_,

't"1 ,

1 (~ 1 , ~i'

'i'

~i'

Pi

p;_,

Y ,

and

xi, y ),
1
'(i

are linear

- 125 -

and

in
expressions of

and

'I '

and

y.

yi

and

are linear fractional

Furthermore, these sixty solutions are

divided into ten groups and any two solutions of each group are
linearly dependent.

In this section, we shall find these sixty

expressions of solutions of (12.1).

To begin with, let us consider the simple integral representation of Euler for

(10.4)

Fl (o<,

...

F1 :

~, ~, '(,,

x, y)

1
r
<
)
J
ot-1
v-cc-1
-~
- ~'
r(oe) rcr-)
u
(1-u)
(1-xu) (1-yu)

du.

(Cf. Theorem 10.2, p.64 . )

In Section 10, we showed that the solution

defined by the integral (10.4) has six expressions "which are derived
from (10.4) by changing variable
(lS. ) { u
1
u

= v, u = 1-v,
= (l-v)/(1-vx),

respectively by

u = v/[ (1-x)+vxl,
u

u = v/"[ (1-y)+vy],

= (l-v)/(1-vy)

(Cf. Theorem 10.3, p.65 . )

The method of the Euler transform
).-1
SC (1-xu) f (u) du

which was explained in Chapter III suggested that, if we consider

the integral of the form

(18.2)

z(x,

we may determine
system (12.1).

y)

A. ,

f,c

fA. ,

(1-xu)

).-1p.-1 .
:(1-yu)
(u)du ,

and

so that (18. 2) satisfies the

Moreover the fopnula (10.4) suggestes that we may

- 126 -

take

A.= -p+1,

1' (u)

p.= -p'+1,

=u

oc-1

. t-oc-1

(1-u)

and that we may take curves joining any two points of

l/y

and

(18.3)

oo

as

0.

z(x, y)

This means that an integral

b at -1
r -ee -1 (1-xu) - ~ (1-yu) a u
(1-u)

u:iay satisfy the system (12.1), where

of

0,. 1, l/x , l/y

0, 1, l/x,

and

o.o.

a;; and b

~' du

are any two values

We shall verify that (18.3) is a

solution of (12 .1 ).

Set

'(u)

= u~~ 1 (1-u)l--l

in (18.2) . . Suppose that the derivatives of

z(x , y)

by differentiation under the symbol of integration :

q =

r -

Then

i.e.

ic

~(~+l)u (1-xu)

s -

can be obtaine'

r,

(3
P.

-~-2 (1-yu) - ~ r<u) du,

-ll-1

~ 'u (1-xu) ,.,

.
-I!' -1
(1-yu)

1- ' ( ~ '+l)u (1-xu)

-~

(1-yu)

r<u) du

-{3'-2

'.f(u) du .

- 127 -

L(z)

= x(l-x)r+y(l-x)s+

(t-(at+~+l)x]p -pyq

r [ ~(~+l)x(l-x)u2 c1-xu)-~- 2 (l-yu)-~

Jc

-ot.,z
1

+
+

~~'y(l-x)u

(1-xu)

-~-1

(1-yu)

-a'-1
P

~ ( r - (at+ ~+l)x)u (1-xu) - ~-l (1-yu) -I!'

- p~'yu(l-xu) -~ (1-yu) -ti' -l

- pot (1-xu} - ~ (1-yu) -/3'] f (u) du

Now

consider the function

G(x, y, u)

(18.4)

= (1-xu)-~-l(l-yu)-13' u' (1-u)r-ac

(1-xu)

-4-1
1
"

_a

(1-yu) '"' u(l-u)

f (u)

Observe that
1
. . A+1
a'
o<
'1-ot
-~G/'au = x~+y~+ - - G
1-xu
1-yu
u
1-u
and hence

= (~+l)xu(l-u)(l-xu)-.i'-Z(l-yu)-~ 'f(u)
1

'aG/du

+ ~'yu(l-u)(l-xu)-1!-l(l-yu)-~ -l

f(u)

+ [ ot (1-u) - ( r - oc.)u] (1-xu) -p-l (1-yu) - ~

'f (u)

If we use two identities :

u(l-u)

2
-(l-x)u +(1-xu)u

a.nd

(1-U) - ('1-0()U

we obtain

0( -

ru

=XU -1u+oe(l-XU)

- 128 -

~G/"au = [-(p+l)x(l-x)u2 (1~xu)-~-z(l-yu)-p'

+(p+l)xu(l-xu) -~.-l (1-}".i) -/3'

-~'y(l-~)u2 (1-xu)~-l(l-yu)-~'-l
+

yu(l-xu) -~ (1-yu) -

~ -l

+(otxu-Ju) (1-xu) -~-l (1-yu) -

+ (l-xu)-~(1-yu)-~'] T(u)

-[(~+l)x(l-x)u2 (i-xu)-~-z(l-yu)-~

. + 13 1 y(l-x)u2 (1-xu) -f,-l (1-yu) -13' -l

+ ( t-(0( +p+l)x)u(l-xu) -~-l (1-yu) -~'

-

~'yu(l-xu)

.. t.t-1
P(l-yu) Y

.. a

- ct(l-xu) -~ (1-yu) -~']

<f (u)

Thus we obtain
L(z) = -

Interchanging

x, y

p JC

d G/~u du =

and

p , p ',

M(z)

-p [G(x, y, u)lc

we also get

- p'[H(x,y,u)Jc,

where
M(z)

= y(l-y)t+x(l-y)s+ [}'-(+~'+l)y]q

- J3'xp -

and

(18.5)

H(x, y, u)

Therefore, if we take

C so that

t-~

-P' le = 0

0(

r-1( (1-xu) -arc1-yu) -4'-1

11
1c = 0

'

[u . (1-u)

(18.6)

(l-xu)-~(1-yu)-~' -lu'(l-u))':_cit

[u (1-u)

(1-xu)

-~-1

(1-yu)

o<p'z

- 129 -

l:len
L(z) = 0

his
s

~eans

and

M(z) = 0 .

that (18.2) is a solution of (12.1).

C a path joining any two points of

0, 1, l/x, 1/y
~

oc ,

f the real parts of the parameters

Thus we can take

~'

and

"1

and

co ,

satisfy

uitable conditions.
THEOREM 18.1:

18.7)

Let us set

U(x, y, u)

= u -1 (1-u) t-~-1 (1-xu) -4fl(l-yu) -Afl

hen the system (12.1) has the following ten integrals as solutions:

1
U(x, y, u)du

if

< Reo(

18.8-2)

J~ U(x, y, u)du

if

< Re c<

J8.S-3)

y, u)du .
.

if

Re

U(x, y, u)du

if

0 < Re oc: ,

Re {3

U(x, y, u)du

if

0 < Re ex ,

Re ~ '

18.8-1)

<. Re 1

:is. 8-4)

eo U(x,

l/x

J
f

Re (

' < Re '1 ,

r - '3 - ~') < l

Re ( ~ - ~ - ~ ' ) < 1 ,

<l

l/y

:18.8-5)

<1

rl/x

:i8.8-6)

)l

U(x, y, u)du

if

Reel< Re)' ,

~18.8-7)

Jr1'y. U(x, y, u)du

if

R~ o(

if

Re~<l,

<

Re

't ,

Re~<

Re ~'

1 ,

<1

1 .

(18.8-8)

(OO
)l/x

U(x,y,u)du

Re(1-f3-(3')<1,

- 130 -

(oo U(x,

(18.8-9)

y, u)du

l/y

.
U(x, y, u)du

(18.8-10)

Re ~ ' < 1,

if

J1/y

lf Re

l/x

Re ( ~ - ~ - (3 ' ) < 1 ,

p< 1,

Re

p' < 1

In particular, if
0 < Re ot. < Re "t

(18.9)

< Re ( ~ + (3'+1) < 2 ,

0 < Re f3

0 < Re

p' ,

then these ten integrals are solutions of (12.1) at the same time.
In order to fix paths of integrations and branches of
as functions of

u,

U(x,y,u)

let us suppose that

Im x > 0,

Im y > 0 ,

or more precisely
{18.10)
Then

l/x

(18.11)

and
-1r

<

l/y

< 1C.

< arg

< 1t'

are in the lower half-plane.

< arg(l/x)< 0,

On the other hand,

plane.

arg x

(1-1/x}

-1t< arg(l/y) <

and

(1-1/y)

We assume that

are in the upper hal f-

We assume that
O<arg(l-1/x) <"TC,

(18.12)

O<arg(l-1/y) < 1t:

Further assume that

-'Jt< arg(l/x) <

(18.13)

arg(l/y)

<

and

(18.14)
Then

and

< arg(l-1/x) <

l/y

a..:g(l-1/y)

< n:

lies in the domain bounded by three lines:

u

(b)

u (1-s) + s/x

az

(1 ~

(a)

~ +oo) ,

(O ~ s ~ 1) ,

- 131 -

and
u = s exp[i arg(l/x)J

(c)

(1 '

s ~ +oo)

Fig. 18.1
This relation of the positions of
y

1/x

and

1/y

is in the sector:
0

and that
0, 1

and

<

arg y

<

arg x

is within the circle passing through three points

x.

(See Fig. 18.2.)

We shall fix the paths of inte-

grations as indicated by Fig. 18.3.

>o

implies that

Fig. 18. 2

- 132 -

+oo

- 00

Fig. 18. 3

We shall determine branches of

18.1.

U(x, y,u)

as indicated by Table

Note that it is sufficient to determine

arg u,

arg(l-u),

arg(l-xu),

arg(l-yu)

on each path of integration. As it is shown

by (10.4), we have

Jo

u(x,y,u)du const. F (ot,p,p',y,x,y)

1

Other integrals of (18.8) are also expressed by means of F1 .

prove this, it - is sufficient to make the change of variable
as indicated also by Table 18.1.
( Q

J~

=e

U(x,

y,

u)du = e

To
u

For example,

-'lti(0(-1) ( 1 ot-1
Jo v (1-v)''A+(J' -)' (1~(1-x)v) -p (1-(1-y)v)-~'.

-1liCct-1) rcct>rrn+1f-r+1)
rroc+~+fl'-r+1)

Fl(.,~' f3 ,ot+~+ ~ -r+1, 1-x, 1-y).

As the first integral (18.8-1) has six expressions, (Theorem 10.3,

p.65),

each integral has also six expressions.

Therefore, alto-

gether, we have sixty expressions of solutions of (12.1).

Remark 1:

The integral (18.3) is transformed by the change

of variable
u

= 1/v

?able 18.1
integral

arg u

arg(l-u) arg(l-xu) arg(l-yu)

transformation

~
0

[ -7r",0]

[-7t,O] .

1t

[0,7r]

[ 0 .7t]

-~

[-7t, OJ

[-7(,0)

u l/v

[ -'Jt, O]

[ 0 ,7r]

[0,1t]

uv/x

[-7,0)

[ 0 ,1t] .

[-7t,O]

uv/y

[-7t,O]

[ 0, 7t]

[-1t,0]

[0,1t]

[-Jt,0]

[ 0 ;rr.]

[ -1t,-O]

[ -lr, OJ

[-1t,O]

[0,7tl

-1t

[O,ir]

u l/(xv)

[-'JC, O]

[ 0 ;re]

[ -1[,0]

-1t

u l/(yv)

uv/(v-1)

IO

00

)
1
1/x

t
l/y

~0
l/x

u = l/[x+(l-x)v]

l/y

5.
1

= l/[y+(l-y)v]

00

}
l/x

co

~
1/y
l/y

it

.
[ -7[, 7r]

[ 0 ,iC]

[ -l:', 0]

[0,1r]

= 1/[y+(x-y)v]

Table 18.2
arg(v-1) arg(v-x) arg(v-y)

arg v

integral

restrictions

0
0

,.

[-'It:~ O]

-7C

-1t

00

[ 0, n:J

[0,7&]

Re(1-)
Re I( > 0

>0

iI

Re(/3+~'

-)') > -1

Re /3

[ -7t,OJ
~

[ 0, 1C}

(-7,0]

[O ,211:]

[ 0 ;n:]

[ -lt, 1C]

[-7t,0]

[-'it,O]

[ 0,27ti

[ -1t,lt]

(-7C,O]

<1

Re(~+~ -r)

Re /!,'

<1

~1

[ 0, 7t}

[ 0, 11:]

Re (If-)> 0
I Re fJ < 1
I

[0,7t']

[0,Jr]

Re(Y-ot) >0
Re~'< 1

'

60

[0,'!r]

'

[ 0, 1t]

[0,1t]

[ 0 '211:}

(0,7[']

[ -'7t, 1t]

[ -7r ,O]

[ 0, 'Tt]

[ 0 ,27r]

[0,2TC]

. co

[ 0,1t]

> -1

Re> 0

Re(~+~' -1)

[ -7C,O]

i
x

.I

[-1t,O}

.0

60

Re (p+f3' -r) > -1

Re(t - cx) > 0

[-7t,O] !I . [ -1r, O]

-TC

[ -'1t, ?t'}

x
-

- - -

'-- - --

. -

--- -

Re~<

Re

Pi.

>0

Re /3' < 1
Re ot. > 0
Re ~ <. 1
Re~'< 1

> -1

- 133 -

into

V(x, y, v}dv ,

where
~ B' -i
V(x, y, v) vt''
(v-1) Y-cc-1 (v-x) -/3
. (v-y) - Pa

and

c l/b ,
Thus we derive ten integrals of

10).

d l/a
V(x, y,v)

from (18.8-1}""(18.8-

These ten integrals are solutions of (F ).

1
THEOREM 18.2: The integrals

fd

c v ,....'2'-J
P
(v-1) J~Ol~l
, (v-x) -Al"'(v-y) - - ' dv

(18.15)

are solutions of the

system (12.2) under the respective restric-

tions given in Table 18.2, where

O, 1, x, y

and

and d

are any two points of

oo The paths of integration are taken as

indicated by Fig. 18.4 and branches of integrand are determined

as indicated by Table 18.2.

Fig.
In Fig. 18.4 and Table 18.2,
{

< arg

< arg(y-1) <

18.~

we assumed that

< arg x < 1C

arg(x-1)

< 7t:

- 134 -

Remark 2.

In order to guarantee the convergence of the

integrals (18.3) and (18.15) we must assume certain conditions

on the parameters

p,

o<. ,

(3 '

weakening these conditions .

of integration.

and

There are two ways of

One of them is to take closed paths

The other is to introduce new idea, called a

The first method is. as fol ~

finite part of divergent integrals.

lows:

"( .

For example, we replace the integrals

Io U(x, y, u)du

and

l
V (x, y, v) dv

respectively by the integrals

fc

U(x; y, u)du ,

along a closed curve

V(x,

y,

v)dv

C which is given by Fig. 18.5.

l/x
Fig. 18.5
Then

fc:
=1.

if none of

U(x,y ,u)du

.
2ri(t-o<)
. 27ti.oc ( 1
(1-e
.
) (le
)Jo U(x,y,u)du

(l-e21ti(r-cl-e2'7ti.ct.>rcoe>r<r-oc.>
A'
>
..
r<t> F1.(a(~
,,..,,.. ,y,x,y
ae. ,

X-o<.

and

is an integer .

the second method, i.e. the method of the finite part of

- 135 -

divergent integrals is essentially based on. the concept of

For example

analytic continuation.

0 U(x, y, u)du

is originally defined for

(x, y),

t.o

0 <Re"

< Re 1 However, for fixed

this integral can be analytically continuable with respect

the parameters

ct ,

s.. ..

(3 '. and

integral in a much larger domain .

Thus we can define this

If we apply this idea to the

integral
_ 1_
r<~>

f~cx- ~jA-lcp<{>d;

then this integral becomes meaningful for every value of .A

'(x)

is a C~-class function.

as a linear functionaL
sense of L. Schwartz .

J0

Teo_ =
J

("
Ox

tion .

We can regard

This functional is a distribution in the

In particular:
...

r'l .

(A

= n),

'(X)

(~

= 0),

(A.

a: -

(o<s>d~ds
J
1

er::

'(n)(x)
where

if

ds n _1

n),

is a positive integer and . Sx is the Dirac-distribu-

- 136 - .

19.

Connection formula and monodromy representations for the

system (12.1) satisfied by

F . In this section, we shall briefly

1
explain connection formulas amcng solutions of the system (12.1)

and monodromy representations for the system (12.1).

To begin with, let us consider the ten integrals (18.15).

We

suppose that these ten integrals are well defined at the same time.
This is possible under a suitable assumption on the parameters

p,

'

and

't The paths of integration for these integrals

are shown by Fig. 18.4. (Cf. Theorem 18.2, p.133.)

We shall use

the same idea as in Section 16 (Chapter III) to find connection

formulas among solutions of (12.1).
(i}

The basic idea is

to take a closed curve consisting of parts of these paths

of integration and small circular arcs and large circular arcs sc

that Cauchy's integral theorem can be applied, and then
(ii)

to let the radius of small circular arcs tend to zero and to

let the radius of large circular arcs tend to infinity.

By using various closed curves of this kind, we can find more

than thirty relations among the ten integrals (18.15).

For example:

if we use the closed curve given by Fig. 19 .1, we get

JO V dv
to

1~
0

V dv

f,

00

V dv

=0

It is known that the system (12.1) has only three linearly independent solutions. (Cf. Theorem 14.1> p.91.)

This means that

there are only seven independent relations.

As these relations

- 137 -

show, any three of the ten integrals (18.15) are not necessarily
linearly independent.

It can be shown that three integrals

f:

(vdv,
are linearly independent.

Vdv

and

Fig. 19 .1

We shall now proceed to the monodromy for the system (12. :_).
In Section 14, it was explained that the system (12.1) has the
singular set which is the union of the five lines-:
(19 .1)

and that any solution of the system (12.1) is regular analytic

in the domain
(19 .2)

(Cf. Section 14, p.94.)

In the same. way as we defined the funda-

1tl (D, x 0 ) of D C - { 0> 1 J with the base point

in Section 6 (Chapter I), we can define the fundamental group

mental group
x

0
TC l (J) , {x , y )) of ~- with a base point (x , y ). Then
0
0
0
0
taking a fundamental system of solutions of (12.1), we can define

the monodromy representation with respect to this fundamental

system in the same way a s we defined monodromy representations

- 138 -

of the Gauss differential equation (6.1).

Chapter I.)

The monodromy representation thus defined is a homo-

morphism of

7t

(JJ , (x 0 , y 0 ))

1
It is known that

two elements.

7r

(oi@, (x , y ))
0
0

~ive

into

is a free group generated by

7t (D, x )

GL (3, C).

(See Section 17, Chapter III.)

1t" (~, (x , y ))

(Cf. Section 6,

is more complicated.

However,

It is true that

is generated by five elements.

C x IC

elements, consider a complex line in

through the base point

singular lines (19.1).

Furthermore assume

that this line does not go through four points

and

(1, 1).

which passes

(x , y ) and intersects with the five

0
0
This means that this line is not parallel

to any of these five singular lines (19.1).

(1, O)

To find these

(0, 0), (0, 1),

These four points are. intersection-points

between the singular lines (19.1).

identified with the complex plane

This complex line can be

C.

Then the intersection -

points with five singular lines (19.1) are represented by five

points
point

..

A , A ; A , . A and AS on th.is complex plane. The base

2
4
1
3
(x , y ) is also represented by a point B on this .com0
0

plex plane .

The intersection of ~

identified with
1t. (C

C. - {A , ~ ", A }.
1
5

and this complex line is

The fundamental group

-,{A1 , ,As}, B) of _c - {Al' ,

generated by five elements.

Asl

is a free group

Those generators are represented by

five loops surrounding . A , A

respectively. Let us denote
1
' 5
these loops by .l , , ./. . (See Fig. 19.2~) . It can be proved
1
5

- 139 -

'7r (~, (x , y ))

that

is generated. by five elements correspond-

0
ing to these five loops.

Fig. 19.2

' n:1 (C - {A1 , , A }, B) is a free group,

5
1ti ( .!; , Cxo, y 0)) . is not a free group. - In other words, there
Although

are relation$ among the five generators.

consider

i. 2 and

l. 3 .

Note that the complex line intersects

with the singular lines

spectively.

=l

and

The point

point of the two singular lines

O<jyj<.S,

number.

Yo)

and

is in a

is the intersectiony "" 0.

More precise-

is a sufficiently small positive

The domain (19 . 3) is homeomorphic

/ X

--+-4~1'----

.l 2

can be .deformed into a loop

lying in the line

y y0 ,

and this .deformed

zy

y "" 1

--....~~--y=O

x=-0 x=l

The loop

re-

A3

0 ~ y 0 ) is in the domain

(19.3)

(xo,

= 1 and

(x

where

A2

(1, 0)

ly speaking, suppose that

0<\1-x\<S,

at

y - 0

Suppose that the base point

neighborhood of (1, O).

To see this, let us

- 140

loop can be considered as a circle

Sz

.I,

The loop
x

= x 0,

83

= 1+

(x -1) e

2
i6

defined by
>

y "'" y O .

can be deformed into a loop lying in the line

and this deformed loop can be considered as a circle

defined by

83

y Yoe18

x - XO ,

The product of s2 and s3 is a torus contained in ~ '

82

and

S3

are two circles on this torus.

52

and

S3

are commutative on the torus.

is homo topic to
topic to
to

l3

l.2

S3S2

S3S2

on

in ~

in . ~

th~ to~s.

Thus we conclude that

are commutative in the group

7r

this group is not a free group.

It is well known that

In other words,

Therefore,

This means that

(,J,

and

S2S3

s2s3

is homo-

l.2 L3 is homo topic

c i 21

(x , y )).

0
0
Similarly, . [

and

[ .l 31

This shows that

1.1 ] and

.i 4 ]

7t (~, (x , y 0)) . . There are other relations.

1
0

are commutative in

For example, there are. _relations among

These relations arise from the fact that three singular lines
x ... 0,

=0

and

=y

intersect at one point

(0, 0).

In order to compute the monodromy representation with respect

to the fundamental system

Jx V(x, y, v)dv ,
0 .

y V (x , y, v) dv ,
0 . .. .

f O1

V(x, y, v)dv

of the system (12 .1), let us first consider three loops

and

5.

The line

y = Yo

l'

i2

intersects with three singular lines

- 141 -

x ... 0,

= 1

and

=x

at . (0, y

r,

(l, Yo)

and

<Yo' Yo>

but this line does not intersect with the other

respectively,

The base point

two singular lines.

l'5

loops
identify the line

= Yo

is on this line .

(xo, Yo)

.t l' 1 2

we shall deform three loops

Now

.l.5

and

lying in the line

into three

Yo

If we .

with the complex plane, the loop

. ..

. .

can be considered as a loop starting and terminating at

surrounding the origin

12

=0

in the positive sense.

l. 5

x 1

in the positive sense.

is a loop starting and terminating at

the point

= y 0 . in

the positive sense.

V(x, y, v)dv

= Jx &+~-r
v

analytically along
.

''
"l'

(See Fig. 19.3.)

..

(v-_l) r--1 (v-x) -~ (v-y) -~ dv

.
then this integral becomes

.
x
e2'1ti(1'' -1) ( V(x, y, v)dv

Jo

and surrounding

If we continue.

The loop

Fig. 19.3

. x

and

The loop

can be considered as a loop starting and terminating at

and surrounding the point

.f'l

- 142 -

This is due to the fact that

arg v+ 27ri

and

arg v

and

arg(v-x)

change to

arg(v-x) + 27t'i, but other arguments do not

change after the analytic continuation along

.li.

Let us next

continue the integral

f:

(19.4)

along the loop

that

li

V(x, y, v)dV

We must deform the path of integration so

does not go across the path.

Hence at the end of analy-

tic continuation the path of integration becomes a path as shown

by Fig. 19.4.

Fig. 19.4
Note that the path shown by 19.4 can be further deformed to a
path as shown by Fig.

19.5~

Fig. 19 . 5
Taking the change of

arg(v-x)

that the integral (19.4) becomes

into consideration, we conclude

- 143 -

{e- 21ti~ -1)

5x

Vdv

y Vdv

Similarly the integral

5
1

V(x, y, v)dv

becomes
2

(e ~

-1)

x
Vdv

li

Vdv

after the analytic continuation along

matrix along

./.

l.

Therefore the circuit

is
e

27t1cs-n

e -2~
. -1

e-~-1

In the same way we get other four matrices corresponding to the

7rl ~ , (x , y )). By these ma trices
0
0
the monodromy representation with respect to the fund~mental

other four genera tors of

system

~ y

x Vdv,

. 0

. 0
...

is completely determined.

Vdv,

14!~

(12.2)~

20 . General solutions of the systems

(12.3) and (12.4).

In this section, we shall derive general solutions of the systems

of partial differential equations which are satisfied by
and

F , F3
2
We shall first consider the system (12.2)

respectively.

which is satisfied by F 2 :
x(l-x)r - xys + [t c (ot+~+l)x] p - ~yq -Ill.~ z - 0 ~

(12.2)

{ y(l-y)t -xys+ ri'-(oc.+p'+l)y]q-{3'xp

-p'z

0 .

Let us make the change of variable

(20 .1)

Then the system (12.2) is transformed into

x(l-x)r' -xys '+[ 2A+1-(2A+fl+O{+p+l)x] p' -().+Myq'

-[(.A+p+oc.)(A.+~)-A(A+Y-l)x- 1 ]z' = 0,
{20.2)

y (1-y) t I -xys '+ [ 2fl+ )'' - (7p.+A+~+ ~'+ 1) y] q I

-[ (.\.+f+o<) Vt+f3'

~+~' )xp I

)-pJ,W-t' -l)y -1] z I

::

If we remove the two terms

A(X+ "j-l)x

-1

and

}A (p.+

'1' -l)y -1

from (20.2), then the transformed system (20.2) has the same form
as {12.2).

For this purpose we shall take

(20.3)

X("+ )'-1)

=0

A and

)"-Cf-+ Y'-1)

so that

= o.

The equations (20.3) yield the following four solutions:

!

(20.4)

f~=

0,

'fa.=O,

~>..=1-Y,

lp.= o,

{)..=0,
,..

{~=1-'t,

= 1-1', f = 1-Y'

The first case of (20.4) corresponds to the identity transformati on

- 145 -

z z'

The second case of (20.4) yields the transformation

(20.5)

z x

t-r z'

and the transformed system (20.2) is obtained from (12.2) by

replacing

Y'

~ '

Q(. '

-~'.

respectively.

'( , .

l ' by

oC+l-t "

f3 +1- r ,

~., 2- '( ,

Therefore, this system admits a solution

z' F 2 (0(+1-l'_,. ~+1-~, Ji', 2-r, y', x, y).

From this we obta~n a_ solution of (12 .2):
1-J
.
.
(20. 7)
z =x
F ( oc. +1- , ~ +1- t , f , 2 - l , l ' , x, y)
2

(20.6)

Similarly, the third case of (20.4) yields a solution of (12.2):

(20.8)

and the four-th case of {20.4) yields a solution of (12.2):

c20.9)

,
= x 1-1 y 1-r' F2 (+2-r-1,

~+1-r,

1J'+1-y', 2-r,

2-r,

x, y) . .

Thus we obtain the following theorem.

THEOREM 20.1:

A general solution of -the system (12.2) is

given by

(20.10)

= AF 2 (Q(,

p. f, r,

J',x,y)

+ Bx 1 -)'F 2 (o+1-l,~+l-t,~' ,2-1, ' ,x,y)

+

cy -r'F <o(+1-r'

r f+i-r

,r,2-i ,x,y)

+ Dxl-t yl- l'' F (+2-1-1' ,p+l-"t, ~'+1-t' ,2-1 ,2-1' ,x ,y),

where

A, B, C and

2
D are arbitrary constants.

Let us consider the next the system {12.3) which is satisfied

by

F .
3

We shall show that (12.3) is transformed into the system

- 146 -

(12.2)" by the change of variables:

(20.11)

x -

1/ ~ '

y =

1/71'

'ld. ' Z

s~

In fact, by this transformation, the syscem

. { x_<1-x)r + ys + [\'-("+~+l)x]p

-<tpz = 0

(12.3)

becomes
(20 .12)

y(l-y)t+ xs + [y - (cc'+f+l)y]q -

ae.'~

'z 0

1-Vr - h 8-t-[Ol'.-jl+l- (2o<H' +2- rH Ip - ""'l.i - "(<+' - +1) z ' = 0'

'lCl-~)t

s'lst[C('-13'+1 - <2ot'+ix.+2 - ~)~lCT-

oc's? - oe' (ot+Ol'-t+t)z'

- 0 ,

where

r -=

a2z'/~s2'

p -

~z1~s

s =

~ z'/~Ed7t

q ., ~z'/ji

The system (20.12) has the same form as (12.2) with parameters
ac+' ' -y+l ,

IX,

oc',

Q(. - ~+l,

ot' - p'+l .

Thus we can derive t he following theorem from Theor em 20 . 1 .

THEOREM 20 . 2 :

A general solution of the system (12.3) is

given by
'

(20.13)

z =Ax

..

-at

. +Bx

.
-oe

,.

F (c(+'+1-i,oc,o1.' ,c(+l-p,ot'+l-f ,1/x,1/y}

2

~y-ot'F2(p+oc.'+l-,p,oc' ,,+l~,oc.'+1-f ,l/x,1/y)

+ Cx -fll y - ~ '~ (oe.+pr+1-r ,O(,' p' ,ot.+1-p,

2
+ . Dx -p y -

In particular, for
coefficients

f + l .;.oc' , 1/x, 1/y)

~IF 2 ( ~+ ~I+1- r, ~, ~I .'~ + 1- , ~I+1- ot I , l /x , 1 I y)

z = F (at,
3

A, B, C and

ol 1

p, p', y, x, y),

we determined the

D by using the Barnes integral

- 147 -

representation in Section 11 (Chapter II).

(See Theorem 11.3,

p.74.)
Let us now consider the system (12.4) which is satisfied by
F4 :
2
x(l-x)r - y t - 2xys +
(12 .4){
2

y(l-y)t -x

.~

[r-(+~+l)x]p - (D(+(i+l)yq -lz = 0 ,

- 2xys + [ r' - ~+~+l)y]q -

(a(+~+l)xp

- o<pz "" 0 '

This system is transformed into

.
2
X (1 -x)r I - Y t I - 2XyS l -+: [ 2,\+ r-(2.\+2f'.+o<+p+ l)X] p I
(20.14)

-(2.\+2p.+1X+p+l)yq ' - ( ().+.+) (A+ft+~)-~(~+;r-1)~ -llz '

= 0

y(l-y)t I -x r I -2xys'+[2f.+ ~ -(2X+2.,f.l+c<+p+l)y ]q I

-(2A+2p.+O(+p+l)Xp I

-[(~+fl+) (A~+~)-p(f+ 1 1 -l)y- 1 JZ I =

by the transformation

z x"A.yP.z'

(20.15)

As we did before, we shall choose

X(A+~ -l)x -l
may be removed from (20. 14) .
(20~16)

A. (X

and

so that the terms

)4- ()A+ ~' -l)y'

and

Thus we determine

+ r -1) = O,

)A(p.+ t'-1)

>.

and

}J--

by

=0

The equations (20.16) yield

. { A.= 0 ,

f- = 0,

{A=l-t,

\ ~- o ,

Hence in the same way as

{.A=O,
fo

we - derive~

= 1- r',

~-1-1,

{ f'-=

i-

r'.

Theorem 20.1, we obtain the

following theorem:
THEOREM 20.3:
given by

A general solution of the system (12.4)

is

- 148 -

(20.17)

z AF (ot,

+Bx 1-1.F

p,
4

Y,

't'~x,y)

R.
I
((l(-i+1,,..-r+l,2-~1

,x,y)

+ cy1 - r' F4 (.- J'+l, 13- r'+1,1,2-r' ,x,y)

+ Dxl-yl-t'F (o<-~-r'+2,~--1'+2,2-)',2-~' ,x,y)

In the .definition of the hypergeometric functions of two
variables, we supposed that
negative integer.

~either

'I

nor

'1'

Therefore, in deriving Theorems 20.1, 20 . 2 and

20.3, we must suppose corresponding conditions.

must assume that

't ,
in Theorem 20.1.

is zero or a

r' + integer

For example, we

- 149 -

21.

Euler transform in double integral.

In the previous sec-

tion, we found four linearly independent solutions for each of

the systems (12.2), (12.3), and (12.4).

However, we would be

unable to calculate the monodromy representations of those systems

with respect to these fundamental systems of solutions.

In

Section 19, we explained how to use the simple integral representation of Euler for computing circuit matrices of the system
(12 .1) which is satisfied by

A similar method based on the

F
1

double integral representations of Euler may yield monodromy

representations for the systems (12.2) and (12.3) which are
satisfied respectively by F2

and

F3 , although, to the author's

knowledge, nobody has ever tried to compute the monodromy representations for the system (12.2) and (12.3).
First of all, we must generalize Cauchy's integral theorem.
Such a generalization was given by H.
f

(x~

Let

y)

be a holomorphic function of

S (CD)

and

as follows:
y

Let

in a domain D.

be .a closed s mooth surface of real dimension two.

If there exists a set V

(i)

V C D ,

(ii)

the boundary of

then

Poincar~

of real dimension three such that

V is

ffs

S ,

f (x, y )dx dy 0

In Section 10 (Chapter II),

we derived the double integral

- 150 -

representations of Euler for

on p.61.)
(21.1)

F , F
1
2

and

F .
3

(S" e Th ''' rcm 10.1

We showed that the three integrals

JJ

'3-lv 13' -l (1-u-v) r-,-~, -l (1-xu-yv) at du dv ,

u,v,1-u-v~O

and
(2.13)

rr
} }

~1 ~-1
v

(1-u-v)

~-~-$'-1

(1-xu}

(1-yv)

-ot'

du dv

u , v , 1-u -v~ 0

are solutions of (12.1), (12.2) and (12.3) respectively, if these

'integrals are convergent.

The method of the Euler transform which

was explained in Chapter III suggests, as in Section 18, that we

may replace the domains of integration
u, v, 1-u-v ? 0

and

.~

1,

.~

by other suitable domains of integrations so that the integrals

thus obtained also satisfy the systems (12.1) , (12.2) or (12.3).
Ac tually we cari find .more than four solutions of (12 . 2) and (12 .. 3)
in this manner.

However, an y new solution of (12 .1). can not . be

obtained by this method.

- 151 -

22.

Characterization of the systems of partial differential

equations satisfied by

and

F , F , r
2
1
3

Consider first a

F .
4

diferential equation of the second order

(22 . 1)

y"+p(x)y'+q(x)y

Suppose that
means that
both.

ls a singular point of (22 .1).

x "" a (:f co )
x

=a

The point

is a singular point of
x

(22.1) if x =a

=0 .

=a

p(x)

q(x)

or

is called a regular singular point of

is at most a simple pole of

a double pole of q(x).

gular point of (22.1).

or

This

x ""a

Suppose that

p(x)

and at most

is a regular sin-

Then the equation (22.1) can be written

in the form

2
(x-a) y" + (x-a)P(x)y' + Q (x)y = 0 ,

(22.1 '}
where

P(x)

Q(x)

and

x = a.

are halomorphic at

As the general

theory of linear differential equations guarantees, the differential equation (22.1') admits a solution of the form
y

where

x = a

and

(x-a/

is a complex const:ant and

~(a)

:# 0.

<f (x)

'(x~

is holomorphic at

As it is easily checked, the quantity

is a root of the quadratic equation:

fCf-1) +

P(a)f

Q(a)

=0 .

Changing the letter, we call the equation

A(~-1)

+ P(a)). + Q(a) ""0

the indicial equation of (22.1') at

x =a.

Two roots of the

indicial equation are called the exponents of (22 . 1') at

a.

- 152 .

f1
f1- f2

Let

f2

and

I integer,
(x-a)

:/: O'.

f 1 - .r2

If

where

'

x.

= a.

If

admits two solution.:>

'fi(x), - - ex-a)

f2

cr 2 Cx)

and

1'i (a)

I 0,

integer, we may suppose that

f1 - f2
-

at

and <}>z _ are holomorphic at -x = a

<f i

'f 2 (a)

then (22. l ')

- f1

(22 .2)

where

be the exponents of (22. l ')

n '

is a non-negative integer.

Then (22.1') admits, in this

case, two solutions

(22. 3)

where

(x-a)

'ri

and

' 2 (a) ~ 0 and

particular, if

fl - f 2 > 0,
point

x a

f1

f2 f (x)+
2

S(x-a)

are holomorphic at

x =a

11 (x),
'f 2

(x-a)

f'1

<f 1 (x)log(x-a), _

and

-,

(a) I 0,

S is a constant which is either 0 or 1.

f1 - f 1 = 0,
the constant

then

s ,. l.

may be zero.

is called logarithmic if

= 1.

In

However, if
A regular singular
A system of

solutions (22.2) or (22.3) forms a fundamental system of solutions

of (22.1').

This fundamental system is called a canonical system

of solutions of

(22~1')

- Suppose next that

at

x a.

p(x)

and

q(x)

are holomorphic in a

domain

If we make the change of the independent variable:

x ...

its

then the differential equation (22.1) becomes

- 153 -

d y/d5 + (~--2 p(l/~))dy/d~ +-:;z;-q(l/~)y

(22 . l")

We shall call the point at

(22.1) if

5-

x ~ ~

equation of (22.1)

at

x oo

regular singular point of

is a regular singular point of (22 . 1").

indicial equation of (22. l") at

The

= 0 is called the indicial

The exponents of (22.1") at

x oo

are called the exponents of (22 .1) at

regular singular point of (22.1), then x

of multiplicity at least one and . a zero of
at least two.

= 0.

=~

x oo

If

,. O

is a

is a zero of p(x)

q(x)

of multiplicity

Therefore, the differential equation (22.1) can be

written as
(22. l" f)
where

P(x)

and Q(x)

equation of (22.1"')

y"+xP(x)y' +Q(x)y - 0 ,
are holomorphic at

at

x oo

~(A+l)

= oo

The indicial

is given by

- )..P( oo) + Q{oo) - 0

The differential equation (22.1) is called an equation of

Fuchsian type, if all singularities of (22.1) are regular singular
points including the point at infinity.

The Gauss differential

equation:
(22.4)

. "+{ ! + +@-1+1}
Y

x-1

+ ot./J

x(l-x) y

- 0

is an equation of Fuchsian type which has three regular singular

points at

x - O, 1

and oo The indicial equations and the

exponents at these three singular points are as follows:

- 154 -

singular
point

indicial equation

exponents

A_(;t-1) + 't A= 0

0, 1-Y

~(~-1) +(cl.+~

).(~ +l) - (oc.+~+l)A + o(.~

00

If

-l+l);\

...

o, t--P

o( ,

1 f: integer, then
F(oc.,~,y,x),

1-l'
F(cC-f+l,

~-+1,

2-r, x)

form a canonical system of solutions of (22.4) at

x ""

0.

If

"t-0{-ft:f integer, then

F(,p, oe.+p+1-r, ~-x),

(1-x)

'--,

F('(-oc,

1-13,

form a canonical system of solutions of (22.4) at

c< -

p:/-

d'--~+1,

1.

1-x)

If

integer, then

x-~F(~,

x-oeF(, D(-1+1, Ol.-p+l, l/x),

form a canonical system of (22.4) at

p-t+l, p-oc+l, l/x)

x = oo .
..

A second order equation of Fuchsian type with three regular

.,.1.+

(22.5)
where

'

= a,

singular points at

and

oo

can be written as
2

Ax+B
,+
Cx +Dx+E y
(x-a) (x-b)y
{x-a )2 ( x-b)2

A, B, C, D . and

of (22.5) at

= a,

=O
'

E are constants. The indicial equations

and

oo

are respectively given by

 lSS

at

f2

fl'

Let

x '"' b

and

1 ~2 the exponents
the exponents at x = oo Then the

be the exponents at
't" ,

x a,

~ ,

relations between roots and coefficients of a quadratic equation

yield

r1 + f2
erl +

1 - (Aa+B) (a-b)

~2 = 1 - (Ab+B)

(b-a)

-1

2
2
f1 r2 - (Ca +Da+E) (a-b)- ,
2
2
dj_ '2 - (Ch +Db+E) (b-a) - ,

-1

't'l + '(2 .. - l+A ,

- c .

"tl t'2

From these relations, we obtain

ri + r2 + ~i + 0 2 + 't'1 + 't"2

{22.6)

... 1

This relation is called the Riemann or Fuchs relation.

Suppost! that two points

fl'

and six complex numbers

fying (22.6) are given.

and

~1 ,

f2'

(a
~2 ,

+ b,

a :;. oo , b :! oo )

'C 1

and

at

oo
x

satis-

Then a second order differential equation

of Fuchsian type with three regular singular points at

and

:i:z
x

= a,

b,

is uniquely determined in such a way that its exponents

=a,

b,

and

-c 2 respectively.

00

a re

f l,

and

fl

If" l,

If'l

and

L' l

In fact1 such a differential equation is given

by
(22. 7).

11+ ( 1- f 1 - f 2+ 1- 0.1 -

x-a

+ ( "C'l 't2 +

x-b

~ )

f 1 r2 (a-b)
x - a

is-1 ~2 {a-b))
y
...
x - b
(x-a) (x-b)

= O

The differential equation (22 . 7) is called Riemann's equation, and

the set of all solutions of (22.7} are denoted by

- 156 -

(22. 8)

The notation (22.8) is due to Riemann and it is called Riemann's

P-function. The set of all solutions of the Gauss differential
equati~n (22 .4) is thus given by

y ... p

(22.9)

ca

ct

1-)'

r-- ~

{o

Now we shall explain the characterization of Riemann's

ferential equation due to Riemann.

To do t his, for simplicity, sup -

pose that none of the singular points

Let

a, b

and

oo is logarithmic .

be the set of all solutions of Riemann's differential equa -

tion (22. 7).

(22.8).

dif ~

Let

This means that

D be the domain

Riemann sphere .

Then

denotes Riemann's P-function

D

=S

- ta, b, co},

where

is the

has the following properties :

.;

(i)

Every function in

is analytic in

D,

but it may be

multiple-valued.
(ii) For every point

in D,

there are two branches

f1

and

of two functions .or one in g. which are defined in a neighbor2

hood of
and which satisfy the condition
0

x
I

; 0 .

.. 157 -

A function
function in

defined i _n a neighborhood of

J. if and only if

is a branch of a

is a linear combination of

2 .

and

(iii) In a neighborhood of
in

a,

there are two branches of functions

which are of the form

(x .. a)f1_t1 (x) '

where

f 2 (a)

11

+ 0.

and

<f 2 are holomorphic at x a

In a neighborhood of

of functions in

3.

'f' 2 (b)

'f'1
I 0.

and

if 2

'1 ; .

f I (x)

..

'2

(x - b) .

are holomorphic at

In a neighborhood of

of functions in

there are two branches

which are of the form

(x .. a)
where

x b,

and

t 2 (x)

x b

x ... oo ,

and

li (b) + O,

there are two branches

which are of the form

-'t

.. x.

where

i,1

X2<00>

+o.

and

X2

X1 (x),

are ho lomorphic at

= 60

and

X1 ( oo )

f. 0,

Riemann proved that the converse is true, i.e. Riemann's differential equation can be characterized by the properties

(i), (ii)

and (iii).
THEOREM 22.1:

Suppose that a set of functions,

fies the conditions (i), (ii) and (iii),

+Tl+ 1:2 l '

ft r/: f 2'

crl

+0-2,

where

Tl ;. '?" 2"

jr ,

f 1+ f 2+
Then

satis-

r + 6""2
1

- 158 -

lw = p
An outline of the proof is as follows:

arbitrary branch of a function in

Let

y(x)

be an

which is defined in a

neighborhood of x . Then the condition (ii) implies that

0
and f 2 are linearly dependent and hence
y(x)

f (x)

f (x)
2

y' (x)

fi(x)

fz(x)

. y" (x)

f'{-(x)

f'2 (x)

in a neighborhood of

XO.

y,

=0

This relation can be written

as

y"(x)+p(x)y'(x)+q(x)y = 0
which is a linear differential equation.
prove that

p(x)

and

q(x)

are holomorphic at

p(x)

and

q(x)

and that they

and . This means

2
1
are uniquely determined by $ . On the

do not depend on the choice of branches

that

By using (ii), we can

other hand, the condition (iii) assures that

= a,

and

are regular singular points . of this differential equation.

00

This

proves Theorem 22.1.

Remark.

If some of a,

and

oo ,

say a,

is logarithmic,

then the first part of condition (iii) should be replaced by the

following condition:
.,

In a neighborhood of

x =a

there are two branches of func-

- 159 -

tions in

Jr

(x-a)

where

Tl

which are of the form:

P1

'( (x) ,
1

(x-a)

rz <j>

2 (x) + J (x-a)

f1

<f 1 (x) log (x-a)

<f 2 are holomorphic .at x a and

and

<j>

(a) I 0,

' 2 (a) :F 0.

It was Picard who first generalized this result of Riemann to

the. system (12.1) which is satisfied by
of functions of

and

F1 He considered a family

which satisfies three conditions corre-

sponding to the three conditions given above, and he .derived the

system (12 .1) from the family.

set of functions of two variables

be a

~l

More precisely speaking, let

x

and

y which are defined

in the domain:

.D1 = .S IC S - {x = 0 J U {x os lj U { x ooJ
~s

U { y .0} U_/y lj U { y

we explained before, the five lines

=1

and

x ~ y

= O,

Od\ U { x '"'Y} .

x 1,

Y.= 0,

are singular lines of the system _(12.1).

1' 1 as

order to describe the behaviors of functions in

In

(x, y)

tends to infinity, we must determine the set of points at infinity .

Ile

have remarked already that

tions, i.e.

2
JP

and

S x S.

Note that

Section 14, Chapter II, p.95.)

Eication of
(i)

C x C.

has two natural compactifica-

C x. C

1
S .. P . (See Remark 1,

Picard adopted

Sx S

as a compacti-

Suppose that

every function in

is analytic in . ~l'

.; 1

but it may be

nultiple-valued;
(1 i.) f or every p oi n t

(x , y ) in
0

~l'

there exist three branches

- 160

(x,. y),

f (x, y)
3

and

(x, y)

fl

f2

f (x, y)

(iii-1)

1 ,

for any point

31

and

if and only if

(0, y ), where
0
jl

'f1 (x , y) , <f 2 (x , y) ,
(iii-2)

1,

<J>

<r 3

and

for any point

'fl' t 2

(1 ~ y

t3

and

(iii-3) for any point

),

(iii-4)

t 1 (x' y) '

(iii-5)

fl'

.._

t 2, f J

f 2 (x '

(x ~1),

(x, y) ,

IO,

1,

ti0 ,

there are

(1, y 0 );

{: O, 1, oo, there are

0
which are of the form:

(co ,

x -~ X (x , Y) ,
3

y0);

x 1 O, 1, oo,
0

there are

which are of the form:

Y) '

are holomorphic at

for any point

there are

where

Y. 1

60 ,

(0, y 0);

""' ..

.._

x -~ X (x, y) ,

(x , O),

is a linear

which are of the form:

are holomorphic at

fo'F any point

Yo

where

Jr 1

three branches of functions in

where

~-1+1

where

(CC>, y ),
0

X1 , 1. 2 , X3

r/: 0, 1,

are holomorphic at

x -~ X (x, . y) ,

(x , y )
0
0

which are of the form:

x

3'1

.three branches of functions in

where

are holomorphi c at

three branches of functions in

where

at

three branches of functions in

where

such that

defined in a neighborhood of

is a branch of a function in
combination of

f3

r/: 0

and that a function

of functions in

wh~re

~ -l'+l
3 (x ' Y) '

'f

(x0 , O);
0

:I

0, 1, co,

there are

- 161 -

three branches of functions in

,...

t2

where
(iii-6)

'f 1 (x, y)
and

for any point

'f'J

).1 which are of the form:

~ 't'
- (x, y) ,
(y-1) t-ct..

'

(x

are holomorphic at

0 , oo),

three branches of functions in

where
'}

:f 0, 1, oo,

which are of the form:

..... (x, y), . .y -13'

Y.-~ X1
. X2 Cx, y),

x-1 , x. 2

where
{iii-7)

and

-x.

for any point

where
~

are three branches of functions in

~l (x, y)
where

; 1,

; 2

and

~ 2 (x,

y) ,

1' 1

some additional hypotheses, then

(x , oo) ;
0

.; 0, 1, oo,

there

which are of the form:

(x-y) '/-,-~,

t 3 . are holomorphic at

Picard proved that, . if

of the system (12 .1).

- x.
- cx, y),
3

y .

are holomorphic at

(x , y ) ,
0 0

there are

f 3 (x, y) ,

. (x , y )
0 0

satisfies (i), (ii) and (iii) and

;.

is the . set of all solutions

We ..do not know clearly what additional

hypotheses must be required.

Picard as well as Appell and Kampe de

Feriet gave the hypotheses in a _very vague fo~ in their works.

The principle of the proof is the same as in the case of Riemann's
equations.

3r 1

Let

z(x, y)

be an arbitrary branch of a function in

which is defined in a neighborhood of a point

(xo~ _Yo> oB i

Then the condition (ii) yields

zl

z2

ZJ

zl

P1

P2

P3

P1

ql

q2

<lj

ql

.rl

r2

r3

81

- 0 ,

z2
p

2
q2
82

Z3
P3
q3

- 0,

- 162 -

zl

z2

P1

P2

P3

ql

q2

q3

tl

t'l

=0 ,

where
p "'Jz/'"Jx,

q ")z/'dy,

..,.

r =

z/ ax

2
2
d Z/dy

and
p.

tlz./-ax, .
J

These three relations can be written as

y)z,

r _...

at (x, y)p+ oc (x, y)q + at (x,

~1 (x, y)p + ~2(x, y)q +

1 (x, y)p+ 12 (x, y)q+ 3 (x, y)z.

p3 (x, y)z'

After very long calculations, Picard showed that this system coincides with (12.1).
PROBLEM 1.

Complete the work of Picard.

Riemann's point of view ca~ be applied to the systems (12.2),

(12.3) and (12.4).

Goursat applied the same principle

systems -(12.2) and (12.3) .

,~o

the

However, the author does not know

whether or not the same principle has ever been applied to th

system (12 . 4).
PROBLEM 2.

Complete. the work of Goursat .

PROBLEM 3.

Deri~e th~

l;

system for

by using Riemann's point of view.

F4-

(i.e. the system (12 . 4))

- 163 -

CHAPTER V
Automorphic Functions, Reducibility

and
Generalizations

23.

Automorphic functions.

It is well known that the elliptic

function was discovered by Gauss, although he did not publish his

discovery.

Then Abel and Jacobi rediscovered the elliptic function

independently .

The discovery of the elliptic function opened an

epoch in the history of mathematics.

Indeed, this discovery had

influences not only on the analysis, but also upon all fields of
mathematics .

Gauss also found a new function which is related to

the elliptic function, quadratic forms, and arithmetico-geometric

mean.

This new function is called the elliptic modular function.

It was Dedekind who rediscovered the elliptic modular function .

This is the earliest example of automorphic functions.
In 1972, Schwarz derived, from the Gauss differential function,
automorphic functions other than the elliptic modular function.
Ihen Fuchs tried to generalize Schwarz's results to more general
linear differential equations of the second order.

Stimulated by

the work of Fuchs, Poincare founded the general theory of automor~hic

functions and called a kind of automorphic functions the

~uchsian

function.

We shall now explain how we can deri ve automorphic funct i ons

- 164

from a linear ordinary equation of the second order

(23 .1)

y"+p(x)y'+q(x)y = 0

Suppose that (23 . 1) is a Fuchsian equation with regular singular

point~ at

and

'

a 1 , a 2 , , an.

(x)

D - S ~ {a1 , , aJ.

Set

Let

'fi (x)

be linearly independent solutions of (23.1) and let

/

.,

7tl (D, x ) -+ GL(2, C)

0

be the monodromy representation of (23.1) with respect to the

fundamental system of solutions
(23.2)

We denote by T

:~

the set of linear fractional transformations

az+ b
t ( z ) .. cz+ d

r : GL(2, C)

and define a homomorphism

Then the composite map

into T.
map

"t

'1:

of

Let us denote by

+ O)

r:

G is a subgroup of T.

r (rr1 (D,
l

Let

ffLl

by

is a homomorphism of

G the image of

G "" 1:

'

(ad - be

1Ll (D, x 0 )
7r {D, x ) under the
1
0

x ))

be a loop in

[:

:].

then fundamental system (23.2) becomes

'"lt

x rf
0

- 165 -

by analytic continuation along

.1..

Consider the ratio

= f 1 (x)/ ~ 2 (x)

f(x)

Then

f (x)

becomes
8 f1(x)+b~2(x)

_ af(x)+b
C'f'l {x) + d<f (x) - cf(x) + d
2

by analytic continuation along

of D by

f.

. Suppose that

1 .

L1

func.t ion defined in

Whe.n

denote the image

is a univalent domain in

= g(z) of

the inverse function

We let

Then

S.

is a single-valued

f (x)

moves along the loop

comes back to the original point, the value

of

./,

f (x)

and

at the

starting point becomes

az+b
cz+d
at the terminating point.

This means that

az+b)

g ( cz+d

z:

g(z) _.

We shall now give the definition of automorphic functions.

Let

G be a subgroup of

and

.'.l

be a domain in

that each linear fractional transformation of

A meromorphic function
relative to

g(z)

G if for any

G maps

S.
~

Suppose
onto A.

is called an automorphic function

t E G and any

g(t(z)) g{z) .

z e A. we have-

166

Example l ~

n E. i },

where

Let
'J:

.:1 = t

and

={ tn;

m. ~

E.

::L
r

Let

where

".!)

' 1

.A = fC

U.J

G.
i

G "" ~ t

and

and

~ m~n

~lliptic

and

w 2.

= z + m w 1 +. n w 2 "

(z)

are complex numbers such that:

Then automorph:i.c funct:Lons relc.t:tve tc

u.L.J.

Then

is an automorphic function relative to

Example 2:

is the set of all integers.

g(z)

= z + 2n7'Ci~

t (z)

are e lliptic functioru3

A typical exampls- i2 t he Weierstr ass

function

2:

1n~nE~

I
;..

(lll~n)~(O,O)

~~

1,...
... =mr
, =rt''
)
"'"'l
~2

Schwarz cons:i.dered the Gaus s differential equation under the

assumption that all of the paramet ers
numbers .

{ oll

l \

and

He determined a.11 ca sES that the image

'1

.4 of

a.re real
D

under the map

is a univalent domain contained in

and hence the invers e

func ~

ti on

x = g(z)

of

is single-valued in
THEOREM 23.1:

Then

Assume that

We shall summa.rize 'his results<

are all r eal.

16 7 -

2>

(i)

g(z)

,t;;

single~valued l.J.

isl

rI

.A "" ~ 1 -

r=

lr~q{-i8l '" l/m,

v "" l ~

~ ~

>..

/ iA

where if one of

1/1.

3IZ

l ""

if

and only

l/n

1s 2~

m "'

19 2'

00

'

n """ 1:1 2,

and

...

:s

00

is one ll the others a re equal;

(ii) if

then

g(z)

(iii)

if

is a rational function;

J\

+)4+ i.> = l/ l + l/m + l/n = 1

then there exists a linear fractional transformation

such that

t(z)

in

is either a simply periodic function or a

g(t(z))

doubly periodic function;

(iv)

if
A+~+ Y"' 1/.1.

+ 1/m + l/n

<

then there exists a linear fractional transformation

such that

g(t(z))

In case (ii)

{2 s 2, k) '

this case

permutation of
(3 ~

3, 3)

and

translat i ons.

t(z)

is an aut:omorphic function defined in

(.I..

m, n)

ls a permutation of

(2 s 3 ~ 3) ,

1, 2'

is a finite group.
(1, oo, eo)

(2, 2s

(2 ~ 3, 4)

or

ln case (iii),

oo),

(2~

4,

~)

l)

(2 ~ 3, 5) .
( l ~ m, n)

(2~ 3~

gro~p

In case (iv),

is an infinite group.

6)

jz!<l.

(1, k) k)

G is an infinite
G

in

In
is a

or

isomorphic to a group of
In part .icu-

differentia l equation has the fundamental s ys t em of solutions

- 168 -

'f
and

r
1 Jo. Ju(u-l)(u-x)'
-oo

~ ~ f,'OO . .

du

.. 2.

is the upper half-plane

~u(u-l)(u-x)

>0 ~

Im z

du .

and

G is the group

generated by
t

(z)

= z+2

2z+ 1

We remark that

t ..
2.

It is known that

1
-1

= <ft (x)/'f 2 (x)

f(x)

equation of the third order.

Ii

z.

Zdv
Cl-v ) (1-xv )

satisfy a differential

We shall find it.

Let

y 1 and Yz

be' any fundamental system of solution of (23.1) and set

z
Since

and

Yz

1 (x)/y2 {x).

are linear combinations of

'f 1

and

Cf 2 ,

we

have

It follows that

oef+@

z ..

1f+S

depends essentially on three parameters.

note that

where

C is a constant.

z'
z
we have

from which

Observing that
I

Y1
.. - -

Y1

:z_

Y2

Y1Y2

Y1Y2

m:

Y1Y2

First

- 169 -

y"'

" ..
L

2
-p.
Y2

- 2-

Thus we have
(23. 3)
Differentiating both sides, we get
Yz ..

-\yi{p(x)+z"/z'}-;y2 {p'(x)+ [zn/z'] '}

.\ y { p (x)

+ z" /z '} 2 - ; y 2 { P' (x). + z'" /z' - [z" /z' J2} .a.

and hence

yl

(23.4)

Since

Ji;y {p(x)
2

2
-2p'(x) -2z"'/z'+3[z"/z'] +2p(x)z"/z' }.

is a solution of (23.1), substituting (23.3) and (23.4)

2
into (23.1), we get
\y2 {p(x) 2 -2p'(x) -2zm/z'+_3[z"/z'J 2 +2p(x)z"/z'}
2
.
.
- \y 2 {p(x) +p(x)z"/z'}+q(x)y2 -0.
y

Dividing both sides by

(23.5)

z"'/z'

-~

\ y2 ,

[z"/z'J

we obtain

-z{!p(x) 2 +ip'(x) -q(x)}.

The left-hand side of (23.5), i.e.

(23.6)

zttt/z'

-~

[z"/z 1 J

is called the Schwarzian derivative of

{z, x }

and often denoted by

The right-hand side of (23.4) or the quantity

I(x)

= tp(x) 2 +\p'(x)

-q(x)

is an invariant of (23.1) under a transformation of the form

(23. 7)

y a(x)u

In other words, if (23.1) is reduced to

u"+p (x)u' +q (x)u ., 0
1
1

 170 -

by (23.7), we have
\ p (x)

+ 12 p' (x)

In particular, if we take

- q (x)

~pl (x)

= exp { - ~

a (x)

+ ~ p 1 (x) - ql (x) .

Jx p (t)dt},

then (23 .1)

becomes
u" - I (x)u

0 .

For Riemann's equation with scheme

oo

we get
1z, x}
l

= ~[(l-'A.2)

a-b+ (l- . 2) b-a +l->'2]

x-a

/A

x-b

(x-a) (x-b)

where

In particular, for the Gauss equation, we have

.{z, x l.,. 1 { l-.X2 + 1-..2 - l-X2-.2+v2 }
I 2 --;r (x-l)Z
x(x-1)

where

~2

f- 2 = ( -

- ( 1 - '1) 2 '

~)2

'

}12 - (0( -

p) 2

About at the same time when the general theory of automorphic

functions was founded by

Poincar~,

Picard discovered an example of

automorphic function of two vili:iables by utilizing the system (12.1}

I

w~ich

is satisfied by

F1 .

Ao we explained before, the system

(12 .1) has three linearly ir1de eT1d~ni: solutions.

them by

Let us denote

- 171

z (x,y) ,
1
and let

f1

z (x,y),
2

z (x,y)
3

be the monodromy representation with respect to the

fundamental system

z (x,y)
1

(23 .8)

z2(x, y)

z (x, y)
3

ft

The representation

GL(3, G:}.

For a loop

.2

is a homomorphism of
at

x ,

'7C l

(~ , x )

the homotopy class

into

J. ]

is an

'

then the fundamental system (23.8) becomes

al

bl

cl

z (x, y)

a2

b2

c2

a3

bJ

CJ

z (x, y)
2
z (x, y)
3

by the analytic continuation along

l.

Set

z (x, y)

z (x, y)

s f (x, y) "" z (x, y) ,

3
rhen the functions

f(x, y) and

alzl + blz2 + clz3

8JZ1 + b)z2 + C3Z3

t = g(x, y)

g(x, y)

become

a + b g+ c

1
1
1
= a)f+b
g+c
3

tnd
a2zl + b2z2 + c2z3
aJzl + b)z2 + C3Z3

= z 2 (x, y)

a f +- b g+ c
2
2
2
a 3 + b 3g + C3

- 172 -

respectively by the analytic' continuation along

= j' (s,

f<s,t)

y =

t),

1 . Let

be the inverse of the transformation

=f

Suppose that:

'f

This means that

and

1'

(x, y) ,

and

g(x, y) .

are single-valued functions.

'f .are

Then an

automorphic functions relative to

a subgroup of linear fractional transformations which are derived

from the monodromy representation

Picard first considered a special case

C(

= p = p'

't = 1 .

= 1/3 ,

He showed that, in this case, we can choose

that
in

(s, t)

t<s, t)

and

ls\ 2 + \ t \ 2 < 1.

.
J

and

so

are automorphic functions defined

In this special case, the system (12.1) has

solutions

du

~u(u-1) (u-x) (u-y) '

f ~V-;:u=(=u=-=l

)=(=d:=_=x)=(=u=-=y=-) '

We have already mentioned that the Gauss differential equation has

similar integrals as solutions when

o(

= ~ =

12, 'I= 1. These

integrals are integrals of simple algebraic functions of

- 173 -

containing a parameter

or two parameters

x, y.

An

int~gral

of algebraic function is called, an Abelian integral which is: very

important in the theory of algebraic functions and in the theory
of algebraic geometry.

The theory of automorphic functions has

become a branch of mathematics related to number theory and alge.braic geometry.

Finally we shall state an extension of
derivative to the case of two variables.

the Schwarzian

_Consider a completely

integrable system of partial differential equations

r

(23. 9)
{

Let

z , z

and

~p+

c(2q + C(3z

~lp+ ~2q+ ~3z

r 1p+

Y2q+. Y3z

be linearly independent solutions of (23.9}

and set

Then

o(

2 '

- 174 -

where

The four differential expressions on the left-hand sides are a

generalization of the Schwarzian derivative to a map of. c2 to
2
C

: (x, y)

(u> v).

- .115 -

24 . Reducibility.

We shall .consider again a linear differential

equation
y" + p(x)y' + q(x)y

(24.1) .

-=

We shall write the equation (24.1) in a form

.( d 2
.).
d
. dxz+p(x) dx +q(x) Y =

o.

The . expression
d2

-z' +

(24. 2)

'

di"+

p(x)

q (x)

dx

is a differential operator.

It happens that the operator (24.2) is

decomposed into a product of two operators:

' ~~ + p(x) !

(24.3)
The

right~hand

(!

+s(x)) (
.. ..

+ q(x) (

+ s(x)) (

;-+ r(x))

member of (24 . 3) means that

+ r(x))

2
d 2 + [r(x) + s

dx

(X) I! +[r' (X) + s (x);,(x) 1

<

A polynomial is said to be reducible if it can be decomposed

into a product of two polynomials of lower degrees.

In general,

the _r .e ducibility of polynomia_ls depends on the field to which the

coefficients of polynomials belong.

'

'

'

'

~..

Similarly, for differential

op.e rators, the reducibility depends on the coefficient-field.

other words, it is necessary to prescribe a field to which .the

In

- 176 -

coefficients

p, q, r

and s . belong, if we discuss the decom-

position (24.3) of the operator (24.2).

We shall restrict our-

selves to the set of all rational functions

C(x).

Namely, we

shall suppose that coefficients of differential operators belong

to C(x).
Remark that

C(x)

is not only closed under the usual addition,

subtraction, multiplication and division, but also closed under the

differentiation with respect to
ferential field.

x.

Such a set is called a dif-

A precise definition of a differential field is

as follows:
A set

K is called a differential field if

(i)

K is a field in a usual sense;

{ii)

there exists a map

a, b E K,

D from K into

K such that, for any

we have

(a)

D(a+b) ,.. D(a) +D{b) ,

(b)

D(ab) .. D(a)b + aD_(b)

The map is called a differentiation.

called a constant if

O(c)

An element

K is

of

= 0.

Let us return to the operator (24.2).

the operator (24.2) is said

The equation (24.1) or

to

be reducible if the operator (24.2)

admits a decomposition of the form (24~3) in C(x). Suppose that
.

(24.2) admits a decomposition (24.3).

{24. 4)

is a solution of (24.1).

y'

+r (x)y

Then any solution of

... 0

This implies that there exists a non-

.. 177 ..

trivial solution of (24.1) which satisfies a first order linear

differential equation (24.4). We shall show that the converse is
also true.

Let .

satisfies (24.4).

~(x)

be a non-trivial solution of (24.1) which

This means that

{24. 5)

' "

q'

- 0

rcr - 0

p' '

and

,. +

(24 .6)

Differentiating (24.6), we obtain

'f" + r 'f' + r 'er -

(24. 7)

0 .

Subtracting (24.7) from (24.5), we have

(p~r)

(24.8)

niminating

f'

(q~r')1'

.. 0

and . . ~.
J . from (24.6) and (24.8),

.
I

.r

. p-r

q-r'

.. q "."' r' - (p-r)r

=0

we

get

iet

s .. p - r

:o derive
r+ s = p ,

r' +rs =- q
his means that (24.2) admits the decomposition (l4.J) .
THEOREM 24.1:
~ere

The operator (24.2) is reducible if and only if

exists a non-trivial solution of (24 . 1) which satisfies a

lrst order linear differential equation (24.4).

Consider the Gauss differential equation

- 178

x(l-x)y" + [

(24.9)

l'-(cx+~+l)x]y'

ll(p-y:.= 0_. _

Suppose that (24.9) is reducible. -Then there

a non-trivial

exis~s

solution of.(24.9) which satisfies a linear differential equation

Such a solution is given by

(24.4).

(24.10)
Since

= const.

f(x)

is a rational function in

r(x)

~x

rl (x)

T(x)

)x r(t) dtl

we have

x,

= rl (x) +~/Li log(x-ai)

r(t)dt

are non-zero constants.

where

exp_[-

= const. e

Hence

P.1

TT (x-a1 )

From the fact that (24.9). is an equation of Fuchsian type, it

follows that

T(x)

must be a constant.

rl(x)

in the form

(24.10.')
On

Therefore, we can write

y ..

.f (x)

= const.

-rr
11

P.1

(x-a1 )

the other hand, the equation (24.9) has its singular points at

x = 0,
1- '{ '

x = 1

= oo

The exponents at

while the exponents at

Therefore, if

'f (x)

and

ai

ii

and

=0

x - 1 are 0 and

t: l,

then

P.1

- 1.

is expressed as one of the following forms:

P(x)

(24.11)

'f (x)

x1

-r P(x) ,

(x-1) 't - - ~ P(x)

x 1 -<x-l)r--~P(x}

are

0 and

r -fJI.- {J
This implies that

- . 179 - .

while

P(x)

is a polynomial in

x.

Let

be of degree

Then

r<x)

is rewritten in one of the following forms:

0.

;ex) ,

xn

xl-l+n ~(x) ,
~(x)

Y-oe-~+n

J..
r<x>
,

xl-oc-~+n ;ex) ,

where

is holomorphic at

exponents of (24 . 9) at
(24.12)

. d.

or .

are

x = oo

p =-n

rf Coo)- + O.
and p , we

or

Since the

and

60

1-1-n

get

oc.+~-)'-n

or

or oc.+p-1-n.

From (24.12) we conclude that one of four quantities

o( '

must be an integer.

~,

1- at.

and

' -

We can also prove that the converse is true,

by using one of the transformations:

y - x

l~Y

z '

(x-1) r-ot-p z

and
y .. x

THEOREM 24.2:

1-'t

(x-1)

1-ot(3

The Gauss differential equation (24.9) is

reducible if and only if one of the four quantities

and

'I- f3

ct ,

p , r-0{ .

is an .integer .

Suppo~e

again that (24.9) is reducible.

Let . y

f(x)

be a

solution of (24.9) which is linearly independent of the solution

(24.10).

Then

 180

t(x)J
['fCx)

(24.13)

is a fundamental system of (24.9).

Let

be the monodromy

representation of (24.9) with respect to the -fundamental system

J. 0

If

(24.13).

and x

=1

con-st-.

T(x>

and

l. 1'

are loops at "xo

respectively, then the function

surrounding x - 0
'(x)

by the analytic continuation :a.iong .t

becomes
0

11.

and

Hence

f < lo1> = [

** *oJ ,~

Consequently, the monodromy group consists of lower t!iangular

matrices. - Conversely, it is . easily verified that, if- the monodromy
group consists of lower triangular matrices, then
of the forms (24.11).

(x)

takes one

This means that (24.9) is reducible.

Thus

we proved the following theorem.

THEOREM 24.3:
~educible

The Gauss differential equation (24.9) is

if and only if there exists a fundamental system of (24.9)

with respect to which the monodromy group consists of lower tri

angular matrices.
' .We shall now explain briefly how to generalize the concept of
-

reducibility
, to linear differential equations of higher orde~s.
.

Consider an n-th order equation

(24.14)

- 181

. or the corresponding differential operator

dn
dxn

(24. 15)

dn-1
p (x)

dxn-1

+ _ + Pn (x) .

The equation (24.14) or the operator (24.15) is said to be reducible

if the operator (24.15) admits a

dn
-dn-1
+
p (x)
+ +p - (x)dxn
1
c:b:n-1
n
-_

- (24.16)

- ( d:1 + .. +Pa
dx l _
where

decompos~tion

< n1 ,

<

n.

(x)Y( a;,

+ : +

-_l - ') - dx 2

~n~(x))_
__

T:

..

The following theorem is a generalization

of Theorem 24.1.
THEOREM 24.4:

The operator (24.15) is reducible ' if- and -only

if ther-e exists a non-trivial solution of - (24.14) which satisfies a

linear differential equation of a!l order lower than

n.

A ._monodromy representation of (24.14) is a - homomorphism

_7rl-_(D, x )
0
whic~

the

into

GL(n,_C), _where

c~efficients

of

D - i$ the. greatest domain in

of _(24.14) are holomorphic.

A monodromy

represe_n tation

* . *

}02

n2

iihere

and

n _
2

~re

determined by

and they are independent

- 182

f { 1t1 (D,

of each matrix of

x )).
0

'.9

The following theorem is a

generalization of Theorem 24.3.

THEOREM 24.5:
..

Supp0se that (24.14) is an equation of Fuchsian

-

type. Then the equation (24.14) is reducible if and only if there

exists a reducible monodromy representation of (24.14).

We shall now proceed to the systems (12.1), (12.2), (12.3) and

(12.4).

The system (12.1) has a form

Q{1 P + oc.zq + 0(3 z ,
{

(24.17)

s -

P1 P +

t -

"1 p + y 2q + '13z ,

~zq + ~3 z,

while the systems (12.2), (12.3) and (12.4) .have .the form
.{. r : :1 s+ a 2p+ a 3q + a 4 z ,

(24.18)

b s+b p+b q+b z.

2
3
4
1

In their book, Appell and Kampe de Feriet gave the following definition of reducibility.

The system (24.17) (or (24.18)) is reducible

if there exists a non-trivial solution of (24.17) (or (24.18)) which

satisfies a system of partial differential equations whose solutions
form a vector space of dimension <3 (or <.4).

For example, (12.1)

is reducible if and only if there exists a non-trivial solution of

(12.1) satisfying a system of . the form

aiq +a2 z ;
c

b q +b z ,
2
1

p -= az

or { q

etc.
bz

We do not know whether a systematic study had been already done for

- 183 -

reducibility. We do not

know

any necessary and suffi..cient condi-

tions for reducibility which are given explicitly in terms of

parameters.

Appell and Kampe de Feriet gave only a few examples.

rhe reducibility of moriodromy representations of the systems (12.1)

,,..,.(12.4) can be introduced in a natural way.

However, it seems to

is that there
is no clear
. statement concerning the relation . between
.
.
.
~educibility

of systems and that of monodromy representations.

PROBLEM 1:

Find a necessary and sufficient condition that the

:ystems (12.1) "'- (12.4) be reducible.

e ducibility of the systems (12.1)
onodromy representations.

Find also relations between

(12.4) and that: of their

-- 184 -

25.

Generalizations. We defined the hype:rgeometric func.tions of

two variables. by. utilizing the Gauss function.

we can int~oduce hypergeometric functions of

Lauricella. introduced the following

F .(oc..
A '

t' l

.:>:".:->.. (

I.

. L

:,

t' n,

' l, . ,

f~ur

In a natural way,
n variables.

functions:

. x.)
n, x 1,

, n

y .
0

m_) (~ - m -)(A

n' n
(~ ,~+ +m0 )

1'

m) ~(A m)
t"l' 1
t'n' n
(l,~) (1,mn)

(at ,ml+ +mn) (~,ml+ +mn)

(11,ml) ... ((Y'n,mn)

(l,~) .. (1,mn)

xl

m
. x n

and

(a( m.. + +m ) ( II. m ) ( n. m ) m

m
' l
n t'l' 1
'"'n' n
1
n
'' - ('j ,~+ +m )(1,m ) (l,mn) x 1 xn
0
1

In case when

n 2,

we have

.FA = F2 , FB = FJ , FC = F4 and FD = Fl .
There are many functions of intermediate types.
I

On the other hand, many functions of one variable are defined

'

as generalization of the Gauss function.

The Gauss function has

the three expressions:

(i)

..

2:,

(Cl( , m) ( ~ , m) xm

m=O

(1,m)(l,m)

'

- 185

(li) .

f,'

const ..

d..:....,

(u-1)

r'-{J-1

- 0(

(u-x) . du ,

1
const.

(iii)

+ieo

r(at+s) r(6+s) r(-s) (-x)s ds.

-ita .
t+s)

re

The second expression can be. written also in the .forlJl .

l

(ii')

const.

-1

u~,

.,_ -1

(1-u) ~

(lx\l)
.

-oc

du

The series

F(ac.l, ...
~11

O(p+l
~p

., -

( o(

m) (..1

1'

f:o (~ 1 ,m) ~

m)

"'pl'
m
x
(~p,m) (l,m)

is derived from the expression (i) in a natural manner.

~enerally,

More

we get the following series:

(at ,m)

(o<q,m) (1,m)

:t

is known that

rom the expression (ii), we can derive the following generalizaion:

oo

S
1

b -1
b -1
.
b -1
.\-l
1
2
(u-a 1 ) .
(u-a 2 ) . , : ~ ~u'."ap) : P (u-x)
. du

:om the . expression (iii)~

we

can derive . the following function:

- 186 -

.f ~ioo
-1.00

TI rcoc .+s>

i=l

1.

Tr rep

il

r(-s)(-x)

ds.

.+s)
1.

These ideas of generalization of the Gauss function are also

applicable to generalization of Appell's and Lauricella's functions.
Finally, we shall talk about the confluence-principle.
start with, let us consider the Gauss function

Introducing a new parameter

F(ac,

~,

To

'(, x).

to define a function by

. F(CI(, 1/E., J, x)

e.

and then taking a limit as

tends to zero, we obtain a new

.. function
lim F(t<, l/C., y, fx)

i~o

, This function

F(oe, l

>

= ~(ot,m)xm

"'-o< y ,m) (1,m) = F(,

is also denoted by G(,

x)

called the confluent hypergeometric function.

r, x)

y,x).

and is

Furthermore, the

confluent hypergeometric function satisfies the differential

equation
(25 .1)

xy"

+ ( "( - x)y' -

On the other hand, it can be shown that

o< y

F(ce-,

=- 0

y, x) admits an

integral representation:

(25.2)

rc1>
r(o<) r ( -ot)
.

1
0

ot-1

(1-u)

t-ac-1 ux
e

du

Note that, in this expression, Laplace transform takes the place

of Euler transform.

The differential equation - (25.1) admits a

- 187 -

regular singular point

at

x .. oo

at

and an irregular singular point

Such equations form an important class of equations

which contains Bessel equation:

(25.3)
The equation (25.3) has the Bessel function of order

as a

solution.
Humbert obtained from Appell's functions seven functions by

the confluence-principle.

i!W F 1 (l/t'

For example,

p,

~ ( ~ ,m) (

13', n)xtl)rn

~, .x, t.Y) .. }.;J "t ,m+n) (l,m) (l,n)

and the function defined by this series admits an integral representat ion

)reJ

C<?t>

,_l

4 r
r " a' n.....Mnr
v"' - (1-u-v) -l"'- 1~ e- r du dv.

u,v~O

1 -u-v~O

Furthermore, this. new function satisfies the system of partial

differential equations
(25.4)

... . ,... {' xr + ys + ( r -x) p - ~ z

yt + XS +

( -y) q -

=0
Z

From (25 . 4) we derive

(x-y)s+ ~q The confluence-principle

is

of more than two variables.

p'p = 0.

applied also to hypergeometric functions