Beruflich Dokumente
Kultur Dokumente
Mathematics
Edited by A. Dold and B. Eckmann
Subseries: Mathematisches Institut der Universit~it und
Max-Planck-lnstitut for Mathematik, Bonn - voL 5
Adviser: E Hirzebruch
1111
Springer-Verlag
Berlin Heidelberg New York Tokyo
Herausgeber
Friedrich H i r z e b r u c h
Joachim Schwermer
Silke Suter
M a x - P l a n c k - l n s t i t u t fLir M a t h e m a t i k
Gottfried-Claren-Str. 26
5 3 0 0 B o n n 3, Federal R e p u b l i c of G e r m a n y
A M S - S u b j e c t C l a s s i f i c a t i o n (1980): 1 0 D 1 5 , 1 0 D 2 1 , 1 0 F 9 9 , 1 2 D 3 0 , 1 4 H 1 0 ,
14H40, 14K22, 17B65, 20G35, 22E47, 22E65, 32G15, 53C20, 5 7 N13,
58F19
I S B N 3 - 5 4 0 4 5 1 9 5 - 8 S p r i n g e r - V e r l a g B e r l i n H e i d e l b e r g N e w York T o k y o
I S B N 0 - 3 8 7 - 1 5 1 9 5 - 8 S p r i n g e r - V e r l a g N e w York H e i d e l b e r g Berlin T o k y o
~BERBLICKS VORTR~GE
Ad-hoc VORTR~GE
ANHANG
A. Connes
Institut des Hautes
Etudes Scientifiques
35, Route des Chartres
F-91440 Bures-Sur-Yvette
France
c) the orbit space of the action of a discrete group (or Lie group) on
a manifold.
Secondly, Brown, Douglas and Fillmore have classified (cf. [8]) short
exact sequences of C * - a l g e b r a s of the form:
0÷ K+A÷ C(X) ÷ 0
b') For a Lie group G the d i s c u s s i o n is the same, with C ~c(G) instead
of ~F.
A A
Space V V/F F G W/F
Measure v.N.algebra
theory L~(V) of (V,F) }< (F) ~ (G) L(W) ~ F
]ch. I <0C(~)
H . ( W x F El', ~) ~
~c(Inda(D)) = <ch,Indt(D),~>
11
K, (V) ÷ H, (V,~)
Besides parts I and II, which will soon appear in the IHES Publications,
our set of p a p e r s will contain:
References
[4] P. Baum and A. Connes, Geometric K theory for Lie groups and
Foliations, Preprint IHES, 1982.
[29] J.L. Loday and D. Quillen, Cyclic h o m o l o g y and the Lie a~gebra of
matrices, C.R. Acad. Sci. Paris, S6rie I, 296 (1983) p. 295-297.
[38] W. Rudin, Real and complex analysis, Mc. Graw Hill, New York (1966).
[42] B.L. Tsigan, Homology of m a t r i x Lie algebras over rings and Hoch~
schild homology, Uspekhi Math. Nauk. Vol. 38 (1983) p. 217-218.
[43] A.M. Torpe, K theory for the leaf space of foliations by Reeb
components, Mat. Institut, Odense Univ., P r e p r i n t (1982).
ABELIAN INTEGRALS
G ,. Harder and N. S c h a p p a c h e r
Max-Planck-Institut
f~r M a t h e m a t i k
Gottfried-Claren-Str. 26
5300 Bonn 3
This formula, in fact, goes back to the year 1897, as does the instance
of D e l i g n e ' s conjecture with w h i c h we start in § I. Tying up these two
relations in the m o t i v i c formalism, we hope to make it apparent that
both r e s u l t s really should be v i e w e d "comme les deux volets d'un m~me
d i p t y q u e '~, as A. Weil has p o i n t e d but in [WIII], p. 463.
18
Contents
§ I A formula of Hurwitz
§ 2 Algebraic Hecke characters
§ 3 Motives
§ 4 Periods
§ 5 The rationality conjecture for Hecke L-functions
§ 6 A formula of Lerch
§ 1o A formula of Hurwitz
1 4v
(I) E' 4v ~ × (rational number),
a,b6~ (a+bi)
rl dx £(
(2) ~ = 21_ - 2.62205755,.. =
u
Notice the analogy of these identities with the well-known formula for
the Riemann zeta-function at p o s i t i v e even integers:
1 2v
(3) E' 2~ - (2~i) x (rational number).
a£ ~ a
A : y2 = 4x 3 _ 4x .
The left hand side of (I) carries data c o l l e c t e d at the finite places of
k , as does the right hand side for the infinite places.
I ~ > k*
2
F ~
I2~ > Gal (~/k)ab > (k®~g)*
where p ranges over all p r i m e ideals of ~ [i] . Then the left hand side
of H u r w i t z ' f o r m u l a (I) is s i m p l y 4 .L(~ 4~ ,0). We h a v e shown how this is
a special value, of the L - f u n c t i o n afforded b y the l - a d i c cohomologies
V~(A)®k®~ 4~
B H -I'0 @ H 0'-I
H I (A) ®~ • =
B
Call HB the f i x e d p a r t of HI(A) under F , and let be a b a s i s of
this onedimensional G-vector space.
DR +
H I (A) ~ F D {0}
i+ : +
HB(A) ®~ .... > (HIR(A)/F +) ®~
~* , Q is the d e t e r m i n a n t of the i n t e g r a t i o n - p a i r i n g
3
~' (z,F) = 4~(z,r) - g2lF) ~(z,F),
§ 2. A l g e b r a i c Hecke Characters
8 : k* > E*
is a h o m o m o r p h i s m induced by a r a t i o n a l character
• oS : k* > ~*
is g i v e n by
: k* > E*
~,f
(xf) = B (x)
* ToY ~.
k]A,f .......... >
klA
* ToY ~.
klA/k. >
L (~,s) = (L(To~,S))~6 T ,
W
where, for Re(s)> ~ + I ,
I
#
the product being over all prime ideals p of k for which the
value ~(~p) does not depend on the choice of uniformizing parameter
wp of kp
(6)
L(To~,0) = T(X) . ~ ( ~ , T )
25
§ 3. M o t i v e s
• there is an isomorphism
N
3.4. When ~ has arbitray weight (%0) the homogeneity condition (5)
above still forces the infinity-type 8 to be of the form 6 = lilBi ,
with lweight (Bi) I : I , n i ( d , 7 ) 6 {±1,0} . Since twisting with finite
order characters is e a s y to c o n t r o l motivically one w o u l d naively expect
to be a b l e to assemble a motive for any given algebraic Hecke character
essentially as t e n s o r product of c o n s t i t u e n t s of the form HI(A) or
HI(A) like in 3.3.
§ 4. Periods
Ho(M) ® ~ : HP,q
P,q
Hn(o,x) , w-n(c,T)
L0 : ~ £0 r
(@,T)6ZxT 0,7
The unit
I
HB(~(-I)) = ~ ~ and HDR(~(-I)) = ~ ,
-I
(9) ~(~{,T) = D(~') T - o6~(~ToB) p(o,'~)
{p 6 G a l ( ~ / ~ ) I P @ (T°O) : ~(~o~) }
c ('[o 8) TIT
(~ ONF/k) D (~ONF/k)
A(F/k,B) • -
(~n) D (~n)
(10)
D(~°NF/k) D(~) n n-1
= 77 ~(6 • D(~)
D(p) n D ( ~ n) Ic I T 6 T
If ~ is a c h a r a c t e r of finite order K* on
/k* with values in E*
~,f
one passes from ~(T) to Q(~T) by leaving D(T) unchanged, and
multiplying the p(0,T) by c e r t a i n algebraic numbers with eigen-
properties under ~ . The details c a n be f o u n d in [Sch]. All we n e e d
to k n o w is the following invariance lemma:
~(T)
4~
a,b6Z (a+bi)
This latter approach was used by Katz in [KI], [K2]. Just like Shimura,
Katz did not stop to look at m o r e precise rationality theorems about
the special values he had determined up to a n a l g e b r a i c number. In fact,
Katz' main concern was with integrality properties and p-adic inter-
polation.
(Note that any algebraic Hecke character of any number field takes
values in a C M - f i e l d . )
Theorem 2:
8o = ~o NF/k ,
8 : k* > E'o*
L(~0"T n,0) 6 E
~ ( w . y n)
But we know the behaviour of the periods Q under twisting and base
extension: see end of § 4. Theorem 2 therefore implies that
Go/k : GLn/k -
.~
<>ik : (Op) p 6 G L n _ I , t 6 GL I }
q 6 GLn_ 1 , t 6 GL I }
and so on.
We i n t r o d u c e the two c h a r a c t e r s
7P : g =
<0 } >
tn
det(g)
and
tn
39
by
Recall that
H o m ( P , G m} = @ H o m ( P o ,~m } ,
gCZ
for e v e r y • 6T .
i(~) = (l(o,<))oE Z
by the rule
{ (-n(@,~)-1)yQ
n(o,~)yp
if
if
n(o,T) < 0
n(o,T) > 0
Kf
SK ~ i > SK < ~B S K
Just as in [Ha] ,II, the right hand side turns out to be a d i r e c t sum
of modules, induced from an a l g e b r a i c Hecke character
n : B(QIA,f) > ~*
T(~IA, f)
G (QIA, f)
V = Ind
B (QIA, f)
\
h is C , and )
_gf £ G(~IA, f)
(Here, "C " means right invariance under a suitably small open com-
pact subgroup in G(~,f) .)
1
where d o : ~ [k : ~] , and the system of maps {iT]<6 T is Q - r a t i o n a l
has to use the spectral sequence which computes the cohomology of the
boundary in t e r m s of the c o h o m o l o g y of the strata. Then the c o h o m o l o g y
has to be r e l a t e d to a u t o m o r p h i c forms, and one has to a p p e a l to re-
sults of J a c q u e t - S h a l i k a on m u l t i p l i c i t y one, and of J a c q u e t on the
discrete non cuspidal spectrum.
iH : F* > GLn(k)
H being the torus with H(Q) = ill(F*) Using this torus we can con-
struct homology classes (compact modular symbols)
* /F*
X : F]A > E*
G ( ~ m , f)
Int(Z(iH,X)) : JTo ~ > Ind
H(~m, f )
G ((~]A,f)
int l°c : J > Ind T oX ,
To~
H (~IA, f)
44
§ 6. A formula of L e r c h
with "ZD, ~ (x) ---x (I~P-I)/D (cod P )" the D th- power residue symbol
J(p) = ~ G(P) ,
: K* > K*
is g i v e n by
in+h)/2 -(n-h)/2
X ~ > Z X t
~(~D ) > K
x ~----> tr~ (~D) / K(xm)
{ D .+xD 0}cipn-1
Z X K ~ ( D D) = x1+.. n =
M ( J I N - (n+h)/2) : M(J) @ K K ( ( n + h ) / 2 )
7T F {<~>)-
a I
X (a)=-1
d 6 GaI(H/K) : CI(K)
a
6 C£ (K) ~ X (a) =-I r (<~>)
(15
d 6 C£ (K) a6 (~/m~) *
REFERENCES
[Da] R.M. D a m e r e l l , L - f u n c t i o n s of e l l i p t i c c u r v e s w i t h c o m p l e x
m u l t i p l i c a t i o n . I, A c t a A r i t h m . 17 (1970), 2 8 7 - 301; II,
A c t a A r i t h m e t i c a 19 (1971), 311 -----317.
[DMOS] P. D e l i g n e , J. M i l n e , A. O g u s , K. Shih, H o d g e C y c l e s , M o t i v e s
and Shimura V a r i e t i e s ; S p r i n g e r Lect. N o t e s M a t h . 900 (1982).
[Ei] G. E i s e n s t e i n , U b e r die I r r e d u c t i b i l i t ~ t u n d e i n i g e a n d e r e
E i g e n s c h a f t e n d e r G l e i c h u n g , y o n w e l c h e r die T h e i l u n g der
g a n z e n L e m n i s c a t e a b h i n g t , - a n d the s e q u e l s to this p a p e r -;
M a t h . W e r k e II, 536 - 619.
[GS] C. G o l d s t e i n , N. S c h a p p a c h e r , S @ r i e s d ' E i s e n s t e i n et f o n c t i o n s
L de c o u r b e s e l l i p t i q u e s ~ m u l t i p l i c a t i o n c o m p l e x e ; J.r. ang.
Math. 327 (1981), 184 - 218.
[GS'] C. G o l d s t e i n , N. S c h a p p a c h e r , C o n j e c t u r e de D e l i g n e e t F-
h y p o t h @ s e de L i c h t e n b a u m sur les c o r p s q u a d r a t i q u e s i m a g i -
n a i r e s . C R A S P a r i s , t. 296 (25 A v r i l 1983), S@r. I, 6 1 5 - 6 1 8 .
[KI] N. K a t z , p - a d i c i n t e r p o l a t i o n o f r e a l a n a l y t i c Eisenstein
series; Ann. M a t h . 104 (1976), 459 - 571.
[La] R.P. L a n g l a n d s , A u t o m o r p h i c R e p r e s e n t a t i o n s , S h i m u r a
V a r i e t i e s , a n d M o t i v e s . Ein M i r c h e n ; Proc. Symp. P u r e Math.
3 3 (1979), p a r t 2; 205 - 246.
[Le ] M. Lerch, S u r q u e l q u e s f o r m u l e s r e l a t i v e s au n o m b r e d e s
c l a s s e s ; B u l l . Sc. M a t h & m . (2) 21 (1897), prem. partie,
290 - 304.
[Schl] N. S c h a p p a c h e r , P r o p r i & t 6 s de r a t i o n a l i t 6 de v a l e u r s
s p 6 c i a l e s de f o n c t i o n s L a t t a c h 6 e s a u x c o r p s CM; in:
S 6 m i n a i r e de t h ~ o r i e de n o m b r e s , P a r i s 1 9 8 1 - 8 2 , B i r k h ~ u s e r
(PM 38), 1 9 8 3 ; 267 - 282.
[Shl ] G. S h i m u r a , On the z e t a - f u n c t i o n of a n a b e l i a n v a r i e t y w i t h
c o m p l e x m u l t i p l i c a t i o n ; Ann. M a t h . 94 (1971), 504 - 533.
[Sh2] G. S h i m u r a , A u t o m o r p h i c f o r m s a n d the p e r i o d s of a b e l i a n
v a r i e t i e s ; J. M a t h . Soc. J a p a n 31 (1979), 561 - 592.
Joe Harris
Mathematics Department, B r o w n University
Providence, RI 02912, U.S.A.
d e f i n e d by
wm_n
X
where w is the r e s t r i c t i o n to X of the g e n e r a t o r of H2(~N,~)
Of course, these are invariants of the u n d e r l y i n g differentiable mani-
fold of X , and do not reflect its complex structure. What does
determine the complex structure of X is the d e c o m p o s i t i o n of the com-
plexified tangent spaces to X into h o l o m o r p h i c and a n t i - h o l o m o r p h i c
parts; or, equivalently, the d e c o m p o s i t i o n of the space An(x) of
differential forms of degree n on X by type:
An(x) = AP,q
p+q=n
52
H~R(X)w = HC = ~ HP'q
p+q=n
HP,q = Hq,P
and
Q ( H P ' q , H p''q') : 0 unless p+p' : q+q' = n .
G(Z0, . ,Zn+l)d(~--@)A..^d(~)A.
. . . . . ^d(Zntl~\'-~G/.ZOn
~Z~i (Z 0 , .... Zn+ 1 )
Hn, 0 = Sd_n_ 1
Hn-k ,k
pr (X) = (S/J) ( k + l ) d - n - i
k
Fk = ~ Hn-~,~(~) __
~=0
(**)
Q(@k(V) (e) ,~) + Q ( e , @ n _ k _ l ( V ) (B)) = 0
~ e • Hn-k'k(x) , 8 • Hk+l'n-k-l(x) ,v • T
t u r n out to be n o t h i n g b u t p o l y n o m i a l multiplication.
It s h o u l d be n o t e d h e r e t h a t this in i t s e l f has some nice c o n s e -
quences: for e x a m p l e , w h i l e w e are as i n d i c a t e d e a r l i e r u n a b l e to
~n+l
determine Hk,k(X) n H 2k (X,~) for any g i v e n h y p e r s u r f a c e in ,
n = 2k , the fact t h a t for d a n+l the m a p
is s u r j e c t i v e i m m e d i a t e l y i m p l i e s that for g e n e r a l X ,
Hk'k(x) n H 2 k ( x , ~ ) = 0 , and so Hk'k(x) n H2~(X,~)- = ~ Thus on a
pr
g e n e r a l h y p e r s u r f a c e e v e r y a l g e b r a i c s u b v a r i e t y is h o m o l o g o u s to a
56
for g e n e r a l p c X , ~ q c X :
References
Yu. I. M a n i n
Introduction
Twenty-five years ago Andr6 Weil published a short paper entitled "De
la m@taphysique aux math@matiques" [37]. The mathematicians of the
XVIII century, he says, used to speak of the "methaphysics o f the c a l -
culus" o r the "metaphysics of the t h e o r y of e q u a t i o n s " . By this they
meant certain dim analogies which were difficult to g r a s p a n d to m a k e
precise but which nevertheless were essential for r e s e a r c h and dis-
covery.
%i"
Before the advent of ringed spaces in the fifties it would have been
difficult to say p r e c i s e l y what we mean when we speak about this geo-
metric object. Nowadays we simply define it as an "affine superscheme"
Spec Z [xi,~K], an object of the c a t e g o r y of t o p o l o g i c a l spaces local-
ly ringed by a sheaf of Z2-graded supercommutative rings (cf. n°4 be-
low) I have tried to d r a w the "three-space-2000", w h o s e p l a i n x-axis
i~ is s u p p l e m e n t e d by the set of p r i m e s
and by the "black arrow", c o r r e s p o n -
ding to the odd dimension.
. . . . . Spec
Three-space-2000
Starting with section 4 the o d d dimensions enter the game. The algebraic
geometers are well accustomed to e n v i s a g e the spectrum of the dual
numbers Spec ~[s] ,s 2 = 0 ~ as t h e infinitesimal arrow and w i l l hardly
object to a s i m i l a r visualization of Spec ~[~]. Still, there is a n
essential difference between these two cases. The even arrow Spec ~[c]
is n o t a m a n i f o l d but only an i n f i n i t e s i m a l part of a m a n i f o l d . This
c a n be s e e n e.g. in the fact that ~I[c]/~ is n o t ~[s]-free, since
2
from s = 0 it f o l l o w s that cds = 0. By c o n t r a s t , the o d d a r r o w
Spec ~[~] is an h o n e s t manifold from this point of v i e w , since the
Z2-graded Leibniz formula for, say, the e v e n differential d~ , is
valid automatically, <.d~ + d ~ - ~ = 0 and one easily sees that QI~[~/~
is R[~]-free.
~ h e even g e o m e t r [ is a c o l l e c t i v e
effect in the ~ - d i m e n s i o n a l odd geometry
Xi ÷ [ <n~n+i
n = _co
zs i n j e c t i v e .
(3) F ~[Xl,X2,...]
63
Then I-~ la[ v = I for all a 6 K. Moreover, R= {aE KI lalv~1 for all v6Sf}.
V
(2) L = (Lf; hv , v E S ).
H°(X,[) + ® H ° ( X v , i v) : H° (X,[)
v6S
D : (Df;rv I r v 6 ~ , v 6 S ).
D : Df * [ rvX(V)
v6S
Furthermore, we set
div I = vCS
[ rvX (v)
as is to be expected.
D ÷ class of 0(D)
c) ~ log Gv(Dv,X)'vol v : 0
Xv
Furthermore,
-I
<Y, ZrvX(V)> = X rv(I v°l v)
vES vES Xv
From the p r o d u c t formula one sees that the r i g h t hand side of (11) re-
mains unchanged if one takes a different non zero section of 0*(0(D))
instead of O*(ID) . This justifies the following general definition of
degree of an i n v e r t i b l e A-sheaf i on the A - c u r v e Y = Spec R:
d
I
(1) x(L) = < [ ~ c1(i)i tdd_i(X) >
i=0
where < > in the right hand side of (I) m e a n s the intersection index
calculated in the C h o w ring or in a c o h o m o l o g y ring with characteristic
zero coefficients. In p a r t i c u l a r
71
d i
(2) X(L) : [ n i
<c1(1) t d d _ i (X) >
i=0
(n~d) (Sn(T))
for i >0 , n >_ 0 , we m u s t choose a volume form w n6 Aft
and then define
w
XA(O(n)) = log I~1 n
#
n (nad,
® ~ J n ,n+d,
which maps vn onto vI (if ~-ft d ~ ~ ~ one can still
correctly define ~n@(d+1) , which suffices for our needs). Now choose
somehow wI and set
® n ,n+d)
wn (<0n~ idlR) \W 1 )
Then
w I ® ~ + 1 ( n d d)
n (n+d, (0(I))
×A(0(n)) = log iV~11 d~l d J XA
72
r r(d+IL
XA(0(]) ® 0( n ]? (~))) : XA(0(1)) + n vo i P
and finally
n (n+d. .n+d r
(3) XA(0(n) ~0(rm(~))) : d$~ d }XA(0(1)) + ~ d ) vol ~
Comparing (3) w i t h (I) and (2) we see that pd looks like a (d + 1)-
dimensional geometric manifold. We can also guess the Todd A-classes
tdAd+1_i(~d) .
2n
w : q~ wv ,
2 r i T r2 v6S
X A ( L ) = d e g L + X A ( 0 x)
where £K is the d i s c r i m i n a n t of K.
o
Furthermore, set H~(L) = { s £ L I Isl v ~ 1 for all v} Then HA(0):{0]U
{roots of u n i t y in R}, which is the a n a l o g of the constant field. From
the M i n k o w s k i theorem one e a s i l y deduces that XA(L) k 0 implies
o
HA(i) + 0
~v : V ° i v = I k =~1
2zi Vk,v A --
Vk'v
I
<m,v'> 2"~i S m A ~'.
Xv(~)
det H ° ( i v ( Y ) ) ® det-]H1{iv(y)) ® Kv
I
6. T h e o r e m . a) XA(0{D)) : ~ <D,D-K> + XA(~) , where 9 = 0(K) , < > is
the intersection index defined in n°1.6.
3. P r o s p e c t s and p r o b l e m s of A - g e o m e t r y
where (h ab} = (hij)-1 . Set ~v = - ~ g~8 dz~ A d~8 and (gy$) = (gas)-1
Then ([v,hv) is c a l l e d a Hermite-Einstein sheaf if
One should study from this view point the moduli spaces of stable sheaves
on a curve, with rank and degree relatively prime. The first u n s o l v e d
problem is to g e n e r a l i z e the R i e m a n n - R o c h - A r a k e l o v - F a l t i n g s theorem to
the A - s h e a v e s of a r b i t r a r y rank, where the second Chern A-class c2A(E)
should emerge,an intersecting new invariant.
2d+I
(z 0 n Zl) ~ = ~(%0L~%1)6 H D (X,IR(d + I) )
Let ~:X~÷ Spec R be the structure morphism. The final formula for
79
(z 0 N z l ) f C H K2 d + 2 ( X f ,U 0 U U 1 ; ~ ( d + I))
4. S u p e r s p a c e
cO dO dI c.
I 0
xI 0 ~I dO
0 I
(I) zI =
] 0
~I 0 xI dI
0 I
< )
I
even places being in the u p p e r left and lower right blocks. These vari-
ables (xi,<i) are the coordinates of the local chart U I. D e n o t e by
BIj the submatrix of ZI formed by the columns with indices in J.
-I
Then the transition rules are Z j : B I j Z I.
One must not forget that the p a s s a g e from the left (row) coordinates
to the r i g h t (column) coordinates of an e v e n element implies sign
change in o d d c o o r d i n a t e s etc. The matrices are o f t e n written in t h e
standard format, like (I) w h e r e the e v e n - e v e n places are k e p t in the
upper left block. The group GL represents the functor of the in-
vertible matrices corresponding to e v e n m o r p h i s m s .
I
Ber B 3 B4] = det (B I - B 2 B 4 - 1 B 3 ) d e t B 4
With the help of these c o n s t r u c t i o n s we can define the most simple and
important class of superspaces. We shall call a superspace (M,0 M) a
supermanifold, a n a l y t i c or algebraic, iff a) Mrd is a pure even mani-
fold of the r e s p e c t i v e class; b) the sheaf 0M is locally isomorphic
to the sheaf Gr0M, which is in turn isomorphic to the G r a s s m a n n algebra
of the locally free (over Gr0M) sheaf JM/J~ of finite rank. (Note
that this G r a s s m a n n algebra should be called symmetric in the super-
algebra since JM/J~ is of pure odd rank).
11 x3 ~3 0!
0 D3 Y3 I ,
11 x4 o
0 T]4 I
~I
Y4
<x1iI <x4 -I
r~1 Yl q4 Y4
-I -2 -I _x[ly 4 I~4
: I x4 -I+ x4-I Y4 ~4q4_i ' 1
-x4 Y4 84 ' Y4 -x4-1y4-2~4q4 /
2 ~ £2(pI p1
JG x )
One can check directly, using the ~ech cocycle in the standard atlas,
that this obstruction is (I,1). Hence G is not split. (Notice that
the projective superspace is split: pmln = (pm,s(770~m(_1))) "
This example shows that the use of projective technique in the algebraic
supergeometry is restricted, and one is obliged to generalize those
methods of algebraic and analytic geometry which do not rely upon the
existence of ample invertible sheaves.
5. S c h u b e r t supercells.
m+n
Furthermore,
m
L ...........
]
where (plq) c = qlp • Finally, QI = {~I11 ..... I11)} (the one element
set). We s a y that the flag f: 0 = SoC S 1 c . . . CSm+ n = T is of type
I 6GI , if 6i(I) = rk Si/Si+ 1 . For groups G = OSp, -~'Sp, Q the
flag f is c a l l e d G-stable if the following conditions are fulfilled:
m
where (rls) = [ 6i(I) . •
i=I
w 6 GW , J = w(I) • .
m+n
T~ : (£,m+ n+ I - k) , ~ = [--~--] for- G = OSp, ~Sp.
SL OSp l lSp Q
K-I
Gilj(W) = [ G~{ (o i+I) , if S = w z);
i=0 ~Ii'Ii+l
GZIj(W) : 0 , if J % w(1) .
of d i m e n s i o n Gilj(W) .
if J = w(I).
6. G e o m e t r y of supergravity
Let us first describe from this viewpoint the simplest example, the
Pl~cker-Klein-Penrose model of M i n k o w s k i space.
On the other hand, the relative tangent sheaves T/,rM are integrable
distributions, i.e. locally free subsheaves of Lie superalgebras in
TM, of r a n k 012N. Every point of M is c o n t a i n e d in two closed sub-
supermanifolds of d i m e n s i o n 012N tangent to TiM and TrM
respectively. They are the Schubert cells we are interested in. The
supercommutator between T/M and TrM defines the F r o b e n i u s map
= ® (~210)
b) There is a w e l l defined isomorphism TOM ($210) * *
making the m a p s (I) and (2) to c o i n c i d e . •
6. P r e p o t e n t i a l . To c o n c l u d e , we g i v e some c o o r d i n a t e calculations
which make it p o s s i b l e to i d e n t i f y our g e o m e t r i c picture with that
of the a r t i c l e [28]. Set N = I and c h o o s e in Mi a local c o o r d i n a t e
system (xza,@~). Assume t h a t the f o l l o w i n g properties are true:
functions, (X~)rd are p-stable and f u n c t i o n s (xa. = ~ ( x za + x~),
I
04,8 ~ ) are local c o o r d i n a t e s on M, w h e r e Or~ =(~ ~ ) P, x a
r = (x~)P)
Such coordinates (xz,@ Z) on M£ , (Xr,0 r) on M r and (~,0£,@ r) on M
w i l l be c a l l e d distinguished ones.
N o w we set
a a)
(5) Ha = ~ i (x£ - xr .
(6) V£,r 3
: ¢Z,r D * ( d O ~ , r dO~ ,r )
coefficients
Step 2. The volume forms vz, r define the spinor metrics [£,r 6 BerSl, r:
REFERENCES
[24] Yu. I. Manin. Gauge fields and complex geometry. (in russian).
Moscow, Nauka, 1984.
Michael Atiyah
Mathematical Institute
O x f o r d OX1 3LB
is t h e r e f o r e a v e r y p e r s o n a l r e s p o n s e to M a n i n ' s article, c o n s i s t s of
S c h o t t k y p r o b l e m a b o u t the c h a r a c t e r i z a t i o n of J a c o b i a n v a r i e t i e s of
p r o b l e m s on 4-manifolds.
104
is b e s t i n d i c a t e d by the f o l l o w i n g s c h e m a t i c diagram:
Topology
Homology K-theory
w i t h the r e c e n t d e v e l o p m e n t of g a u g e t h e o r i e s of e l e m e n t a r y p a r t i c l e s .
this a s p e c t later.
§2. A r i t h m e t i c laanifolds
An a l g e b r a i c c u r v e d e f i n e d by e q u a t i o n s w i t h i n t e g e r c o e f f i c i e n t s
The A r a k e l o v - F a l t i n g s t h e o r y is then c o n c e r n e d w i t h e x t e n d i n g as m u c h
g ~ 2), one being the P o i n c a r 6 metric and the other being the metric
of t h e s e m e t r i c s arose in a p h y s i c a l context.
§3. Fermions
geometrically. On a p u r e l y a l g e b r a i c level of c o u r s e t h e r e is no
in p h y s i c s w h e r e g e o m e t r i c i n s i g h t and i n t e r p r e t a t i o n g r e a t l y a s s i s t
incorporated.
The d e v e l o p m e n t of s u p e r - m a n i f o l d s as o u t l i n e d by M a n i n a p p e a r s
f e r m i o n i c variables.
107
expressed in t h e form
E f (x)~G dx~ A dy 5
~BY
mind.
provides one bridge between geometry and quantum field theory. For
example ideas from supersymmetric field theories have cast new light
made.
in h i s Oxford D.Phil. thesis shows how the Nevanlinna theory fits into
Quantum
Langlands ÷ Connes + Field T h e o r y Quantum
References
4. D.B. Ray and I.M. Singer, Analytic torsion for complex manifolds,
INTRODUCTION
*) T h i s p a p e r s u r v e y s the r e c e n t i n t e r a c t i o n b e t w e e n t h e t h e o r y of
p h a s e t r a n s i t i o n s in s t a t i s t i c a l m e c h a n i c s a n d the t h e o r y of c o m -
plex dynamical systems.
their typical fractal properties play a very significant role in the in-
teraction b e t w e e n the Yang-Lee theory and the r e n o r m a l i z a t i o n group
approach. None of t h e s e and the findings in [15] w o u l d have been possi-
ble without t h e a i d of e x t e n s i v e computer graphical studies and experi-
ments.
k = I
<> k = 2 k = 3
(]. = I I*o-i q •
1
I
- U , if O. = 0
(2) E(i,j) = l j
0 , else.
Hence,
(3) E(o) : ~ E(i,j)
bonds
(4) x = exp(U/kB.T )
kBT
(5) fn - n in Z k , n = n(k)
Let
(6) N k = {x 6 • , Zk (x) = 0] ,
(7) N oa 0 ~ + = {x c } ,
<<!iii!iii!iiTiiiiiiTi?Cii
:i:!iiii!!i{iiii!!iiii[i!iiiiii!.!.i:!:!:i:{:"Lgneti c
. :i!iiii!!iiiiii!!iiiiii!iiiii!i!iiiii!iiiiii e . . . .
Xo iiii!ii!iiiiiii!ii!iiiiiii!ii!i!ii!iiiiiiiiiiiiii
~ ~ : t i ! ! i ii!iii![iiiiitiii!ii!iiii!ii!iii!i!li :ti:!iiii[!
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
(8) C ~ I T- Tc I- ~
R(x) = x = ( x2+q-lh
(10)
\2x+q-2/ 2
Z I (X) = q(x+q-1)
2.4 k-2
~0(x) = (2x+q-2)
Aq(1) = {x 6 ~ : Rn(x) ~ 1 as n ~ ~}
q
(12)
Aq(~) = {x 6 ~ : Rn(x) ~ ~ as n ~ ~} i
q
is the J u l i a set of Rq
J U L I A SETS A N D P H A S E T R A N S I T I O N S
CONJECTURE 1.1.
N = Jq , i .e.
(I 5) l f
g(x)
(x) =-~-f (Rq(X))
:
Ii n
(2x+q-2)
+ g(x) , with
117
PROBLEM 1.2.
(a) In w h a t sense is Nk ~ N as k ~ ~ ?
(b) F o r w h i c h q 6 ~ is N = J ?
q
(c) F o r w h i c h q E • does R admit further
q
attractors, other than I and ~ ?
(16) Rq(1-q)}i E Z
THEOREM 1 . 4. (Sullivan)
eie ( x-a ) 2
x i ~ )
x I-~x
Thus f can have only finitely many cyclic stable regions. But it
is s t i l l an o p e n p r o b l e m whether 2d-2 (d > 2 the d e g r e e of f) is
a sharp upper bound. Another open problem is w h e t h e r a Siegel disk al-
ways has a critical point on its b o u n d a r y . M. Herman [11] in a r e -
markable paper proved this conjecture recently for f(z) = z2 + I .
Note that theorem 1.5 a n d theorem 1.4 p r o v i d e an e x c e l l e n t basis for
computer experiments. For the detection and characterization of all
cyclic stable regions of a m a p f one simply has to f o l l o w the forward
orbits of a l l c r i t i c a l points. The following example illustrates the
strength of t h e s e results:
x-2 >2
EXAMPLE 1.6. f(x) = ~ , J =
Our map
Rq(X) = < x222+xq+-q1- 2 )
I , ~ , 1-q , ± ~ , (2-q)/2 .
A I :: {q6{ : Rn
q(1-q) ~ I , n ~}
M R := • ~ (A I u A )
q I ~q
S o S
q q
q
b~ I
I J
yl rl
>
I L
I
7"
J J
122
2.5
-2.5
-I
Figure 5. MR in b l a c k
] .70
I .54
0.2 0.36
Figure 6. Detail in M R
D e t a i l of MR ( s e e f i g u r e 6) i n b l a c k s u r r o u n d e d by A I i n y e l l o w and
A~ i n g r e e n .
The M a n d e l b r o t s e t M i n b l a c k t o g e t h e r w i t h i t s e l e c t r o s t a t i c poten-
t i a l g i v e n by t h e Douady-Hubbard c o n f o r m a l homeomorphism ~'.D -~ ~'-M.
123
I .70
I .54
0.2 0.36
Ato in b l a c k
I .70
1.54
0.2 0.36
where
x+q-1
~q(X) ........x_ I
(20) and
Sq(X) = x2+q-1
x2-1
This means that
(21)
with (x+ i)2
Dq(X) = x-1
Thus, Mq exchanges the hot phase (x=1) with the cold phase
(x=~) and the two crucial critical orbits of (l-q) and O .
Figures 3-6 are e x p l a i n e d and d e s c r i b e d in greater detail in [15].
In p a r t i c u l a r p r o b l e m 1.2 (c) is answered. R o u g h l y s p e a k i n g the main
body of MR and each of its buds as well as the main body of the
detail in figure 6 and each of its buds identify p a r a m e t e r s q for
which there is a p e r i o d i c attractor. Their basins of a t t r a c t i o n estab-
lish a third magnetic phase and the b o un d a r y of these basins, which
is the J u l i a set of R , b e i n g also the b o u n d a r y of A (I) and
q q
A (~) , is a c a n d i d a t e for a formal locus of p h a s e transitions. Note,
q
however, that even though N m a y be given by Jq , the J u l i a set
of Rq , its points may not be s i n g u l a r i t i e s of the free energy f in
the t h e r m o d y n a m i c limit. This seems to c o n t r a d i c t (5), but note that
in the t h e r m o d y n a m i c limit the free energy may simply allow an analytic
continuation.
In summary our e x p e r i m e n t s leed to the f o l l o w i n g interesting conjec-
tures:
CONJECTURE 1.7.
(I) MR is connected.
M = {c6~ : fn
c (0) / , ~ , as n ~ ~}
fc(X) = x2 + c .
2.0
-2.0
1.75
-2.25
2.0
1.4
-0.20 0.4
? i i
,D
t9
1 . 70
I .54
0.20 0.36
{~ E C : g~(O) 7~ r o o t of Pl f n
129
2 C AI , 3 6 A1 , 4 E A
! -I
I
e:
C~
~Q
0 C~
o
.,'a
I-'-
M
L~
0
II il
~Q
°
L¢I
I
L i
II II
0 0
0 ~D
I I
It II
0
133
REFERENCES
9. S. Grossmann: A n a l y t i c P r o p e r t i e s of T h e r m o d y n a m i c F u n c t i o n s and
Phase Transitions, in: F e s t k ~ r p e r p r o b l e m e IX, Ed. O. M a d e l u n g
V i e w e g 1969.
Wilfried 5chmid*
D e p a r t m e n t of M a t h e m a t i c s
Harvard University
Cambridge, MA 02138
a) as K-module, V is a d i r e c t sum of f i n i t e d i m e n s i o n a l
irreducibles, each o c c u r i n g o n l y f i n i t e l y often;
(I) b) the actions of g and K are compatible;
c) V is f i n i t e l y generated over the universal enveloping
algebra U(g).
HP(D,O(L)) = 0 if p ~ s ,
(5)
HS(D,O(L))(K) is a non-zero, irreducible Barish-Chandca module
Kashiwara [9]. 5ome preliminary remarks are necessary. The flag variety
X may be thought of as a quotient GC/B; here B is a particular Borel
subgroup of G C, i.e~ the normalizer of a Borel subalgebra b C g. The
differentials of the algebraic characters of B constitute a lattice A.
Each x ¢ A -- or more precisely, the corresponding character ex of B
-- associates a Gc-homogeneous, holomorphic line bundle Lx --~ X to the
principal bundle B --~ G C --+ X. Its cohomology groups are finite dimen-
sional Gc-modules, which are described by the 8orel-Weil-Bott theorem
[8]. In p a r t i c u l a r ,
The center Z(g) of the universal enveloping algebra U(q) acts on the
cohomology by scalars (Schur's lemma! ), so
(13) F : M ( D x) --+ M U x)
A : M(U x ) ) M(D x ) ,
(14)
AV = D,@ U V ;
%
(19) F : M ( D ~ , K c) ~ M(Ux,K) •
and Springer [47] have worked out the closure relations between K c-
orbits; their results make it p o s s i b l e to interpret this geometric irre-
ducibility criterion quite explicitely.
orbit Y. Once the parameter x has been fixed, the duality D ~--+ Y
extends to the line bundles which enter the two constructions: a homo-
geneous line bundle L --+ D is dual to Ly, x --+ Y if the tensor product
L@Ly, Z restricts to a trivial K-homogeneous vector bundle over DDY .
start with the choice of a G-homogeneous line bundle LA, ~ --+ 5 I. Its
extension to C ~, w h i c h I denote by the same symbol, comes equipped with
an a c t i o n of g and an a l g e b r a i c structure. 5ince C~ is open in X ,
the pushforward construction attaches the space of a l g e b r a i c sections
H O ( c ~ , O ( L x , ~)) to the datum of the Kc-orbit C~ and line bundle LX, ~
Integration over 51 pairs this Harish-Chandra module nondegenerately
with C®(SI,L*x, ~@T~) (K) • the m o d u l e corresponding to the G - o r b i t S 1 and
line bundle L'X,~@T~. The hypothesis of the equivalence of c a t e g o r i e s
translates into the inequality Re k > -I . On the Beilinson-Bernstein
side, this implies the existence of a unique irreducible submodule: the
entire module generically, when Lk, ~ cannot be c o n t i n u e d across 0
and - , otherwise the finite dimensional submodule consisting of s e c -
tions regular at the two punctures. The realization of the dual module
C'(51, LI, ~ @ T ~ ) ( K ) exhibits both Harish-Chandra modules as m e m b e r s of the
principal series.
the level of Euler characteristics, and the known vanishing theorems for
the t w o constructions are sufficiently complementary to p e r m i t a compari-
son after all. In p a r t i c u l a r , it is p o s s i b l e to c a r r y techniques and
results back and forth between the two constEuctions [233.
References
G.B. Segal,
St. C a t h e r i n e ' s College,
Oxford.
§I General remarks
In this talk a loop group LG will mean the group of smooth maps
from the circle S I to a compact Lie group G. One reason for study-
ing such groups is that they are the simplest examples of infinite
dimensional Lie groups. Thus LG has a Lie algebra L~ - the loops
in the Lie algebra ~ of G - and the e x p o n e n t i a l map L ~ ÷ LG is a
local d i f f e o m o r p h i s m . Furthermore LG has a c o m p l e x i f i c a t i o n LG~,
the loops in the c o m p l e x i f i c a t i o n of G. Neither of these properties
is to be e x p e c t e d of infinite dimensional groups: neither holds, for
example, for the group of d i f f e o m o r p h i s m s of the circle [17].
From this point of view the group Map(X;G) of smooth maps
X ÷ G, where X is an a r b i t r a r y compact manifold, seems almost as
simple as LG. Such groups are of great importance in q u a n t u m theory,
where they occur as "gauge groups" and "current groups"; the m a n i f o l d
X is p h y s i c a l space. Thus loop groups arise in q u a n t u m field theory
in t w o - d i m e n s i o n a l space-time. In fact it is not much of an
exaggeration to say that the m a t h e m a t i c s of t w o - d i m e n s i o n a l quantum
field theory is almost the same thing as the r e p r e s e n t a t i o n theory
of loop groups.
If dim(X) > 1, however, surprisingly little is k n o w n about the
group Map(X;G). Essentially only one irreducible representation of
it is known - the r e p r e s e n t a t i o n of Vershik, Gelfand and Graev [9] -
and that r e p r e s e n t a t i o n does not seem relevant to q u a n t u m field
theory. For loop groups, in contrast, there is a rich and
extensively developed theory. They first became popular because of
their c o n n e c t i o n with the intriguing combinatorial identities of
Macdonald [16]. They are the groups whose Lie
156
+
X = LG/G ~ LG~/L G~ . (2.1)
y : {z 6 • : Izl < I} ÷ G~ .
y = yu.y+
+
with ¥u 6 LG a n d Y+ 6 L G ~ . This is analogous to the factorization
of an element of GL (~) as (unitary) x (upper triangular).
n
158
(ii) For each invariant inner product < , > on the Lie algebra
of G there is an invariant closed 2 - f o r m ~ on X w h i c h makes it a
symplectic manifold, and even fits t o g e t h e r with the complex structure
to make a Kahler manifold. The tangent space to X at its b a s e - p o i n t
is L ~ /~ , and ~ is given there by
2n
(2.2)
0
0
is the H a m i l t o n i a n function c o r r e s p o n d i n g in terms of the symplectic
structure to the c i r c l e - a c t i o n on X w h i c h rotates loops. The c r i t i c a l
points of ~ are the loops w h i c h are h o m o m o r p h i s m s qr ÷ G. Downwards
g r a d i e n t t r a j e c t o r i e s of ~ emanate from every point of X, and travel
to c r i t i c a l points of ~ . The g r a d i e n t flow of ~ and the H a m i l t o n i a n
circle action fit t o g e t h e r to define a h o l o m o r p h i c action on X of the
multiplicative semigroup ~I = {z 6 • : 0 < Iz[ ~ I}.
The c o n n e c t e d c o m p o n e n t s C[l] of the critical set of ~ are the
c o n j u g a c y classes of h o m o m o r p h i s m s I : T ÷ G. They c o r r e s p o n d to
the orbits of the Weyl group W on the lattice z1(T), where T is a
m a x i m a l torus of G. The g r a d i e n t flow of ~ stratifies the m a n i f o l d
X into locally closed complex submanifolds X[I], where X[k] consists
of the points which flow to C[I]. Each stratum X[I ] is of finite
codimension.
y = y_.l.T+ ,
+
with ¥± 6 L-G~, and I : S I ÷ G a homomorphism. This is the analogue
159
(lower t r i a n g u l a r ) x ( p e r m u t a t i o n m a t r i x ) × ( u p p e r triangular).
§3 The G r a s s m a n n i a n e m b e d d i n g of X
Ia b] (3.1)
c d
whose o f f - d i a g o n a l blocks b,c (with respect to the d e c o m p o s i t i o n
H+ @ H_) are Hilbert-Schmidt. The blocks a and d are then auto-
m a t i c a l l y Fredholm.
i : X ÷ Gr(H)
Proposition (4.2).
for each positive coroot h of G, where < , > is the basic inner
product on ~ .
(iii) Unitarity
(*) A n i n t e r e s t i n g f a m i l y of m e a s u r e s o n Y is c o n s t r u c t e d in [5],
but it d o e s n o t i n c l u d e the m e a s u r e n e e d e d to p r o v e u n i t a r i t y .
165
E? ° = IUp;
P
E pq
!
= 0 if q ~ 0
Ly 0 H~O } ( N y ; ~ ) ,
E pO m (~ S(T* @ N w) ~9 det(N w) @ L w ,
w
Z I fi(8)Bi (8)d8
i S1
lim e-IB
s÷0 ~,@,s
167
REFERENCES
[2] M.F. Atiyah and R. Bott, A. Lefschetz fixed point formula for
elliptic complexes: II. Applications. Ann. of Math., 8 8 (1968),
451-491
[5] I.G. Frenkel, Orbital theory for affine Lie algebras. To appear.
[18] A.N. Pressley and G.B. Segal, Loop groups. Oxford Univ. Press,
to appear.
[20] G.B. Segal and G. Wilson, Loop groups and equations of KdV type.
Publ. Math. IHES, to appear.
Yum-Tong Siu
Department of M a t h e m a t i c s
Harvard University
Cambridge, MA 02138, U.S.A.
from the curvature tensor. This survey talk discusses the situation
when the q u a d r a t i c forms used in p r o d u c i n g holomorphic objects are
only positive semidefinite instead of strictly positive-definite. In
certain c a s e s we m a y e v e n a l l o w certain benign negativity. One may
wonder w h y one should bother to s t u d y the s e m i d e f i n i t e case. There
are a number of reasons. Let me give two here. One is that some
situations are n a t u r a l l y semidefinite, like the s e m i n e g a t i v i t y of the
sectional curvature for a bounded symmetric domain. Another is that
when limits of h o l o m o r p h i c objects are used in p r o o f s (like in the
continuity method), the l i m i t of s t r i c t l y positive definite objects
can only be assumed first to be semidefinite though in the final
result it may turn out to be strictly positive definite. The
semidefinite case is by far m u c h m o r e c o m p l i c a t e d t h a n the d e f i n i t e
case.
n
I $(Pi ) - $(P2) I <_ C ( x Izi(Pi) - zi(P 2) I2 + ~ Izi(Pi) - zi(P212)
i:2
which is s t i l l open.
Theorem 2__
t F__o~ ~ y e r y positive integer n there exists a constant Cn
de__2mgnding o n l Z o n n w i t h the following ~r_~o~r_~ty.- L e t M be a compact
K~hler manifold of complex dimension n and L be a Hermitian line
bundle over M. L e t G b e a n o ~ _ n subset of M and a, b ~[ positive
numbers such that the curvature form of L admits a as a l o w e r bound
at ever Z point of G a n d admits - b as a l o w e r bound at ever Z point of
M-G. Assume that
Mok [29] also showed that for any harmonic map from a compact
K~hler manifold to an irreducible compact quotient of the n-disc
(n h 2) w i t h real rank 2n s o m e w h e r e , each of t h e n c o m p o n e n t s of t h e
map between the universal covers induced by it is either holomorphic
or antiholomorphic.
a n d C so t h a t
for all 9 . The subelliptic multipliers form an ideal lq. Let cij
( l ! i,j ~n-l) be the L e v i f o r m of t h e b o u n d a r y of ~ n e a r x in t e r m s of
an o r t h o n o r m a l frame field of (i,0) vectors tangential to the b o u n d a r y
of ~ . The starting point of K o h n ' s theory is the f o l l o w i n g results
concerning the ideal lq of s u b e l l i p t i c multipliers. For notational
simplicity we describe the case q = 1 and the general c a s e is s i m i l a r .
References
6. D. C a t l i n , in p r e p a r a t i o n .
21. J. Jest & S.-T. Yau, Harmonic m a p p i n gs and K~nler manifolds, Math.
Ann. 262 (1983), 145-166.
22. J. Jost & S.-T. Yau, Strong rigidity theorem for a certain class
of compact complex surfaces.
• I
27. R. A. L z v n e , On c e r t a i n covers of the u n i v e r s a l elliptic curve,
Ph.D. thesis, Harvard University 1981.
44. Y.-T. Siu, The Levi problem, Proc. Symp. P u r e Math. 30 (1977),
45-48.
52. Y.-T. Siu & S.-T. Yau, Compact K~hler manifolds of positive
bisect i o n a l curvature, Invent. Math. 59 (1980), 189-204.
56. S. W e b s t e r , On m a p p i n g an n - b a l l into an n + l ) - b a l 1 in c o m p l e x
space, Pacific J. Math. 81 (1979), 267-272.
Jacques Tits
CollAge de F r a n c e
11PI Marcelin Berthelot
75231 Paris Cedex 05
More correctly: two such data correspond to the same isomorphism class
of a l g e b r a s if a n d o n l y if t h e y differ only by a p e r m u t a t i o n of the
indices I,...,~ . Following C. C h e v a l l e y , Harish-Chandra a n d J.-P.
Serre, one can give a simple presentation of the algebra corresponding
194
[H,H] : {0} ;
[ei,f i ] = -h ;
1
[ei,f.] = 0 if i ~ j ;
3
-A. + I
(ad e i) 13 (ej) = 0 if i ~ j ;
-A. + I
(ad fj) 13 (fj) = 0 if i ~ j
(Z41 I ~ i ..... I
is standard, whereas
is twisted.
C..
W(A) = <ri [ 1<i<Z ; r2 = I ; (rir j) 13 : I if i ~ j ,
as A. A = 0, 1, 2 o r 3 >
13 ]l
(cf. [MT], [Mal], [Ti3] a n d also, for the a f f i n e case, [IM], [BT2] and
[Ga]) . In p a r t i c u l a r , G has a Bruhat decomposition G = U BwB ,
w 6 W
leading to a " c e l l decomposition" of G/B : the q u o t i e n t s BwB/B have
natural structures of finite-dimensional affine spaces. Now, in the
minimal situation, the same N is the g r o u p N of a n o t h e r BN-pair
(B-,N) , not conjugate to the p r e v i o u s one except in the finite-dimen-
sional case (i.e. when ~ is p o s i t i v e - d e f i n i t e ) . Furthermore, one
a l s o has a partition G = U B wB , called the B i r k h o f f decomposition
w 6W
of G (because of the special case considered in [Bi]; for the g e n e r a l
result, cf. [Ti4]) . W h i l e the c e l l s BwB/B are finite-dimensional,
the "cells" B-wB are finite-codimensional in G , in a s u i t a b l e
sense, and, unlike the B r u h a t decomposition, the B i r k h o f f decomposition
always has a big cell, namely B-B if one chooses B- in its c o n j u g a c y
class by N so t h a t the intersection B N B is m i n i m u m with respect
to the inclusion (we t h e n say that B and B are o p p o s i t e ) . In the
formal situation, a Birkhoff decomposition (and h e n c e a b i g cell) still
201
l ~--> i - <l,h.>-@. of A* ,
i 1
(6 • 2 ) m2 = (-1 <l,hi>
i = ~i 6 T = H ° m ( i * ' C × ) ' with 0i(
for I E A*
and
(6 .3) if A, .A.. : 0 (resp. I;2;3), then m.m. = m m.
13 3x 13 3 1
coincides with that scheme only over the p r i n c i p a l ideal domain. This
suggests that, in g e n e r a l , ES m a y be the g o o d functor when restricted
to t h o s e rings. This is u n d o u b t e d l y so in the affine case. Indeed, in
that case, one can characterize the functor ES restricted to p r i n c i p a l
ideal domains - call it E (pid) - by a s y s t e m of v e r y natural axioms
which, it seems, s h o u l d be satisfied b y the "good" functor GS (cf.
[Ti4], 7.6 b)). Another application of t h o s e axioms is t h a t they enable
one t o d e t e r m i n e explicitly the functor E~pid)- (whereas the m o r e
abstract definition by generators and r e l a t i o n s is m u c h less m a n a g e a b l e ) ,
and that the r e s u l t one obtains suggests (always in the a f f i n e case)
what must be the functor GS for a r b i t r a r y rings. We shall come back
to t h a t q u e s t i o n in t h e next section (and in A p p e n d i x 2), b u t l e t us
first conclude the p r e s e n t one by t w o r e m a r k s .
A
GS(~) = ~(K((z))) ,
Gs(K) = F4(K((z)) )
onto RL,/LHo(a)
itself.
arises when one studies the case of the G C M (_~-~) ; the 8-dimensional
A --
Having thus described both G/B = G/B and B , we h a v e g a i n e d
^
some understanding of the algebro-geometric nature of G itself. But
a more direct and promising picture is g i v e n b y V. Kac and D. Peterson
[KPI] who attach to G (the m i n i m a l group) a "coordinate ring", or
rather two rings, the ring ~[G] of "strongly regular" functions, and
the ring ~[G] r of "regular" functions. The first one is g e n e r a t e d by
the coefficients of all highest weight representations (in [ K P 1 ] , t h i s
210
9. Applications.
A
Now, let us e m b e d G in a c o m p l e t e topological group G and
suppose that, if U and U denote the c l o s u r e s of U and U+
A -- + A --
in G , the p r o d u c t mapping in G defines a h o m e o m o r p h i s m of
^
_ x H x ~+ onto a dense open subset ~ of G . L e t us a l s o e n d o w
H' with a topology making it into a c o m p l e t e topological group, such
that Ker z is c l o s e d in H' and t h a t the c a n o n i c a l algebraic iso-
morphism H'/Ker ~--> H is an i s o m o r p h i s m of t o p o l o g i c a l groups as
well (observe that, by hypthesis, H is locally closed, hence closed
^
in G , and is t h e r e f o r e a complete topological group).
Note that the left (or right) translates of all open subsets of
obviously form a basis of the topology of G' (hence the u n i q u e n e s s
assertion).
A
Appendix 2. The group functor GS in the affine case.
A
(P0) There is a Lie algebra functor Lie GS defined as f o l l o w s
(compare [DG], pp. 209-210). For any r i n g R , set
R(e,g') = R [ t , t ' ] / ( t 2 , t '2) , where s,s' are the canonical images
of t,t' in the q u o t i e n t ; in o t h e r words, R(e,e') is the tensor
product of two a l g e b r a s R(e), R(e') of dual numbers. For r 6 R ,
let ~ : R(S) --> R , 1 : R(e) --> R(E') , o : R(~) --> R(c,E') and
~r : R(e) --> R(s) be the R-homomorphisms sending £ onto 0,s',ee'
A
and re respectively. Then, the additive group (Lie GS) (R) is the
kernel of the h o m o m o r p h i s m
A A A
GS(~n) : Gs(R(e)) ----> Gs(R) ,
the s c a l a r m u l t i p l i c a t i o n by r is i n d u c e d by the a u t o m o r p h i s m
A A
GS(~ r) of Gs(R(s)) and the c o m m u t a t o r of two e l e m e n t s
A A
x , y £ (Lie GS} (R) E G s ( R ( ~ ) ) is the o n l y e l e m e n t Ix,y] such that
A A
GS(O) ([x,y]) = (X,Gs(1) (y))
A
where ( , ) stands for the usual commutator in the group Gs(R(£,e')).
A
(PI) (Lie GS) (~) is the Kac-Moody algebra associated to the system
( ~ ® A, (~i)0 ~ i ~ i' (hi)0 ~ i ~ i ) completed with respect to the natural
gradation (deg e i = I, d e g fi = -I, deg h i = 0)
A
(P2) The group Gs(C) coincides with the formal group over C
attached to S by any one of the construction processes described in
§§ 5 a n d 6; in p a r t i c u l a r , it c o n t a i n s (a c a n o n i c a l image of)
w x
Hom(A ,~ ) and its c e n t e r consists of all ~ 6 H o m ( A * , ~ ×) such that
~(~i ) = 0 for all i
A
(P3) Modulo its center, Gs(C) is the subgroup of Aut((Lie GS) (C))
generated by all converging e x p ad g , with g E (Lie GS) (~) (this
turns out to be identical with the adjoint group considered by
R.V. Moody [Mol] and J.I. Morita [Mo3]; about this group, cf. also the
215
A
(P4) The functor GS restricted to p r i n c i p a l ideal domains, together
with suitably defined functorial homomorphisms
A A
~i : SL 2 - - > GS , B : Hom(A*,? x) --> GS (which we leave as an exercise
to d e t e r m i n e explicitly), satisfies the axioms (i') to (iv') of
^
[Ti4], 7.5, and is c h a r a c t e r i z e d by them, once Gs(C),= ~i(~) , ~ (C)=
are given.
A
Those properties clearly indicate that the functor GS which we
are going to d e f i n e is the "right one", at least when restricted to
principal ideal domains but maybe also for general rings, considering
its fairly simple and natural definition (though it is c o n c e i v a b l e
that some algebro-geometric invariants of the ring, such as Pic R ,
should be brought into play).
type of S diagra m
(n+1 vertices)
(n vertices)
3D4 I
(i) an i s o m o r p h i s m K --> Ua ;
aE~_
77
Ua x T x a6=~ + Ha
A
Gs(R) = @(R((Z))) ,
-I
xa in case (i) ,
2~
Cn_ I if A has type Dn ,
F4 if A has type 2E 6 ,
iT
h(z;z') = Z(Z'i T Z i + Z i Z_i) ,
219
A
The description of the functor GS associated to an arbitrary
system S of affine type, i.e. a system satisfying (At) but not
necessarily (A2) and (A3) now amounts to a combination of extension
220
REFERENCES
[sw] G. SEGAL and G. WILSON, Loop groups and equations of KdV type, Publ.
Math. I.H.E.S. 61 (1985).
[Ve2] J.-L. VERDIER, Les repr~sent~io~ des alg~br~ de Lie affines: Applica-
tions ~ quelques problgmes de physique (d'aprgs E . DATE, M. J I M P B ,
M. K A S H I W A R A , T. M I W A ) , S@ra. N. B o u r b a k i , e x p o s @ n ° 596
(1981-1982) , A s t @ r i s q u e 9 2 - 9 3 (1983) , 3 6 5 - 3 7 7 .
D. Zagier
U n i v e r s i t y of M a r y l a n d Max-Planck-Institut fur M a t h e m a t i k ,
C o l l e g e Park, MD 20742 D-5300 Bonn, FRG
2
Yl = 4x3 4x + i (Yl = 2y - i); (2)
~i =
!
E ~
dx
Yl
- 2
S Y /4x3_ 4x+ 1
dx
- 2.993458644...; (3)
a +b
{an+ I, bn+ I} = {--~-~, n/~n}).
~2 = 2
I Y
6 /4x3-4x+l
dx
- 2.451389381...i (4)
¢/Ze I + Z~ 2 * E({)
1 ' 1 i
p(z) = -~+ Z ( 2-
z m,n (z-m~l-nw 2) (m~l+n~02) 2)
!
(Z means [ ), w h i c h satisfies
m,n (m,n)~(0,0)
g2
=
60 [ , ........ 1
4 -
4~ 4
~
(i + 240
oo n3 )
.2zin~i/~ 2 : 4,
m,n (m~l+n~ 2) 3~ 2 n=l e -i
, ~ 5
g3 = 140 [ 1 _ 8w6 ( n )
m,n (m~l+nw2)[ 27~J2~ 1 - 504 n=l~ e 2 Z i n ~ i / e 2 - 1 =-I.
15
2P 0 = (l,0), 3P 0 = (-i,i), 4P 0 = (2,3), 5P0=(~,~), 6P 0 = (6,-14)
n2h (P0) .
LE(S) = II 1 1 (6)
pfi37 l + N P ) - P + ~ s i+ 1
P P 37 s
n Ii~ 2 3 4 5 6 7 8 9 i0 ii 12 13 14 15
-Z (8)
a(n) Ii ~ 2 .........-3
....... 2 -2 6 -i 0 6 4 -5 -6 -2 2
o r d s = 1 LE(S) = rk E(~) = 1
and that
d (i0)
d--s LE(S) [s=l = 2h(P0)'~I'S
iLL is not known (for E or any other elliptic curve), this last
statement cannot be checked. However, L~(1) can be computed numer-
ically (cf. §3), and its value 0.3059997738... strongly suggests
(cf. (3), (5)) the equation
i.e. (i0) with S = I; the truth of this equation follows from equation
(18) below.
with ~2 as in (4) and some integer A(p). The value LE,p(1) can
be calculated numerically by the rapidly convergent series LE,p(1) =
2 [ (~) a(n) e -2~n/p 3/~ (cf. §4) so we can compute A(p) for small
n=l P n
p. The first few values turn out to be
for some A(d) 6 Z; the first few values not in (13) are
230
The most striking thing about the v a l u e s in (13) and (15) is t h a t they
are all squares. This is e a s i l y understood from the B i r c h - S w i n n e r t o n -
Dyer conjecture: the D i r i c h l e t series LE, d is j u s t the L-series of
the "twisted" elliptic curve
2
A(d) = c(d) , (17)
4Wl~ 2
L E,( 1 ) L E ' d ( 1 ) - i/d h(Pd) (18)
g e n e r a t e d by the group
.at+b. b
ftc--~) = (cT+d) 2f(T) (T ~ ~,(~ d ) 6 F) (22)
2 3 ~' (~)
q(T) - ~(T) = ~(T) - ~(T), -2~if(T) (23)
2q(T)-I
dx
(this gives a map as in (20) with the right p u l l - b a c k of 2--~L~; that
232
{(T) = q-2+2q-l+5+9q+18q2+29q3+51q4+82q5+131q 6+ . . . .
(24)
~(T) = q-3+3q-2+9g-t+21+46q+92q2+t80q3+329q4+593q5+-.-
So far we h a v e not used the fact t h a t f is a m o d u l a r f o r m on F;
we c o u l d h a v e t a k e n any p o w e r series f(T) = q+... and u n i q u e l y solved
(23) to get L a u r e n t series ~(T) = q-2+..., ~(T) = q-3+... However,
since ~ and ~ are s u p p o s e d to be F-invariant functions w i t h no
poles in ~, and since f is a m o d u l a r f o r m of w e i g h t 2, the two
functions f4 = f2~ and f6 = f3~ m u s t be h o l o m o r p h i c modular forms
on ? of w e i g h t 4 and 6, r e s p e c t i v e l y . B u t the s p a c e Mk(F) of
modular forms of w e i g h t k on F is f i n i t e - d i m e n s i o n a l for any k
and o n e can o b t a i n a b a s i s for it by an a l g o r i t h m i c procedure (e.g.,
u s i n g the E i c h l e r - S e l b e r g trace formulas, but we w i l l find a s h o r t c u t
here), so w e c a n i d e n t i f y f4 and f6 f r o m the b e g i n n i n g s of t h e i r
Fourier expansions. O n c e one has c a n d i d a t e s f4 and f6' one d e f i n e s
= f4/f 2 and ~ = f6/f3; t h e s e are then a u t o m a t i c a l l y modular func-
tions on F, and the v e r i f i c a t i o n of (23) reduces to the v e r i f i c a t i o n
of the two f o r m u l a e
= a 2 + 2b 2 + 5c 2 + 10d 2 + ac + a d + b c - 2bd
233
As to f6' we o b s e r v e t h a t the f u n c t i o n
.aT+b.
From @' = - 2 ~ i f and (22) it f o l l o w s t h a t the d i f f e r e n c e ¢~c--~-~J-%(T)
ab
is a c o n s t a n t for all Y = (c d ) 6 F. C a l l this c o n s t a n t C(y); clearly
C:F ÷ ¢ is a h o m o m o r p h i s m . The t h e o r y of E i c h l e r - S h i m u r a implies
that the image i = C(F) is a l a t t i c e in { with g2(i) and g3(A)
rational integers. S i n c e we can c a l c u l a t e @(T) and h e n c e C(y)
234
4. Modular forms a t t a c h e d to E
The most important c o n s e q u e n c e of the modular d e s c r i p t i o n of the
elliptic curve E is that the L-series of E equals the L-series of
the m o d u l a r form f, i.e. that the numbers a(n) in (7) are p r e c i s e l y
the Fourier c o e f f i c i e n t s in (21). This follows from the E i c h l e r -
Shimura theory (cf. [13]). As a consequence, the function LE defined
in (9) has the integral r e p r e s e n t a t i o n
foo oo
/~7
~-~ L{(1) = ~L~ (i) = 2 f( it) log t dt = 2 [ a(n)G(2zn),
3/77 n=l 3/~7
1
with
e-Xtlog t dt = i eiU d u ¢
G(x) =
I1
I x
U
and since there are w e l l - k n o w n expansions for G(x), this can be used
!
to c a l c u l a t e LE(1) = 0.30599... to any desired d e g r e e of accuracy,
as m e n t i o n e d in §i. Similarly, if -d is the d i s c r i m i n a n t of an
imaginary q u a d r a t i c field in w h i c h 37 splits, then the "twisted" form
f*(~) = [(--~d)a(n)qn-- is a cusp form of w e i g h t 2 and level 37d 2 satisfy-
n 2 22,
ing f * ( - i / 3 7 d T) = -37d T f (T), so
(.co .
LE,d(S) := (2~)-s37s/2dSF(s)LE d(S) = ] f*(~) ( t s - l + tl-s) dt,
' /
1
~ _ _ 2
a(n) c(d) = [ (~)c(~d 2 ) ( n e ~ , -d a f u n d a m e n t a l disorim- (27)
rln r inant).
r>0
2~idT
g(~) = ~ c(d)e
d>0
-d ~ 0 or l ( m o d 4)
(-d/37) = 0 or 1
d 3 4 7 ii 12 16 27 28 36 40 44 47 48 63 64 67 71 75 83 ... 148
(28)
c(d) 1 1 ~ 1 -i -2 -3 3 -2 2 -i -i 0 2 2 6 1 -i -i ... -3
where
dx dy = ~iw2/i
J
{/A
as c l a i m e d .
5. Heeqner points on E
In this section we d e s c r i b e a construction which associates to
each integer d > 0 a point Pd e E(~). These are the "modular
points" of the title, since their construction depends o n the m o d u l a r
description of E given in §3.
We assume first that -d is a f u n d a m e n t a l discriminant, i.e. the
discriminant of an imaginary quadratic field K. We consider points
b+i/d
T e .]( of the form '< - with
2a
-d
If (~7) = -i, there are no s u c h T and w e set Pd = 0; o t h e r w i s e
the set of ~ is i n v a r i a n t under F and there are h distinct points
71,...,Th modulo the a c t i o n of F, w h e r e h = h(-d) is the c l a s s
number of K. The theory of c o m p l e x multiplication shows that these
points are individually defined over a finite extension H of
(the H i l b e r t class field of K) and collectively over @ (i.e. their
images in ~6/F are p e r m u t e d by the a c t i o n of the G a l o i s group of H
over ~). Hence the sum ~(T I) + . . . + ~(Th) , w h e r e ~:~{/F ÷ E(~) is
the m a p constructed in §3, is in E(~). M o r e o v e r this s u m is d i v i s -
1
ible by u, w h e r e u is ~ the n u m b e r of u n i t s of K (= 1,2 or 3)
if 37~d and u = 2 if 371d ; this is b e c a u s e each point 7j ~
is the fixed point of an e l e m e n t of F of o r d e r u. We define
Pd e E(~) by
h
up d = [ ~(~j) ; (33)
j=l
so b(67) = 6; for d = 83 we h a v e h = 3,
so b(148) = -3. In this way one can make a table of the multiples
b(d). Such a table (up to d = 150) was computed b y B. Gross and
J. Buhler, while I was independently computing the Fourier coefficients
c(d) by the m e t h o d mentioned in §4; the letter with their data
arrived in G e r m a n y on the v e r y morning that I had completed my com-
putations and d r a f t e d a letter to them, and the p e r f e c t agreement of
the two t a b l e s gave ample reason to c o n j e c t u r e the followin~:
c
H2(Y ) = H2(Y ) ~ <Sl> ~ . . . ~ <Sr> , (34)
where SI,...,S r are (the h o m o l o g y classes of) the curves used in the
r e s o l u t i o n s of the cusp singularities and H~(Y) consists of h o m o l o g y
classes o r t h o g o n a l to the Sj; the homology groups in (34) are taken
with c o e f f i c i e n t s in Q.
For each integer N > 0 there is an a l g e b r a i c curve TN c y
defined as follows. C o n s i d e r all equations
ATIT2 + % %'
where
(h'(-d) = h(-d) for d > 4, h' (-3) = 1/3, h' (-4) = 1/2) and I (n)
P
is a c e r t a i n arithmetical function whose definition w e do n o t repeat.
The proof of (36) was geometrical: the p h y s i c a l intersection points
of TM and TN in X × ~/PSL2(~ ) are in i:I correspondence with
certain equivalence classes of b i n a r y quadratic forms and are counted
by the first term in (36), while the t e r m Ip(MN) counts the inter-
section points of TM and TN at i n f i n i t y and the intersection of
TM with the c o m b i n a t i o n of c u s p - r e s o l u t i o n curves S. which was
c 3
removed from TN to get T N.
T37 n TN = O P U D (37)
txl< 14/Y~Z~ (14SN-x2)/P
x2~148N (mod p)
N
with b(d) as in §5. Then ZB(N)q is a m o d u l a r form of weight 2,
level p and N e b e n t y p u s (~ .
rN(T~oX) = 0
N=I
for all R e R, and since this holds for all homology classes X, we
must have [r N [ T N ] = 0 in H 2(Y,@) . Since T Nc is a l i n e a r c o m b i n a t i o n
of T and curves S. coming f r o m the c u s p r e s o l u t i o n s , this m e a n s
n 3
co r
[ r N [ T N] + [ sj [S~] = 0 (40)
3
N=I j =i
divisor of ¢ T37 = [ rN [ P + d
N{I x2<148N (148N-x2/p)
x2{148N (mod p)
244
8. Completion of the p r o o f
We are now nearly done. For each N > 0 define
C(N) = ~ c(148N-x2),
P
x2<148N
x2~148N (mod p)
(pd+x2)/148
G(z) := ~ C(N)q N = [ c(d)q
N>0 d>0
xeZ
pd+x2~0 (mod148)
= g(pz) @ (z)IU148,
2 th
where @ = [qX and U is the m a p w h i c h picks out every m coef-
m
ficient of a F o u r i e r e x p a n s i o n , i.e.
1
%(z) IUm - m j(mod m)
integer n with
2
n E -pd (mod 148), InT ~ 37.
pd+n 2
Take N - 148 " Then in the equations
Theorem 6' The powe~ series d~0 pdqd __is_a modular form of weight
3/2 and level N.
~ ( ~ P(4NM_x2)/p )q M
M x2<4NM
x2~4NM (mod p)
b(d)q d
d>0
-d~square (mod 4N)
= ( q M - ~ -
(4NM-x2))
h(pT) 8 (T){U N [ I b
M>0 x2<4NM
X 2 ~ 4 N M ( m o d p)
Bibliography
Michael Atiyah
Mathematical Institute
O x f o r d O X I 3LB
recent preprint [3], Vafa and Witten have proved some striking results
about the eigenvalues of the Dirac operator, and this talk will present
DA : S 8 V + S ~ V.
In t e r m s of
an orthonormal basis e of tangent vectors DA is g i v e n
d 3
l o c a l l y by E e~V~, where V~ is t h e covariant derivative in the
3 3 3
j=l
e.-direction and e. a c t s on s p i n o r s b y Clifford multiplication. In
3 3
particular DA depends on A only in t h e O-order term, i.e. if
results is
value:
more precise.
of F e r m i o n - B o s o n duality.
namely:
0 A
(i.i)
so t h a t
DA :
I
D A2 = D A D A+ @ D A+ DA, and t h e non-zero eigenvalues of the t w o
factors +
DAD A and DA+ DA coincide. If V has large positive
D A2 will have a large gap between its O - e i g e n v a l u e and its first non-
the end of §i, the spinors decompose and DA takes the form given in
index D A+ = d i m N A+ - d i m NA
Theorem 1 holds.
those bundles V for which the index formula gives zero. To treat
these the key idea is now the following. Suppose we can find a
connection A on V so that
o
(ii) IIDA- DA II ~ C
o
this will do equally well since the only effect of taking multiple
so t h a t
V ~ (W' ~ W") = V ~ C N = NV
we find
(2.2) DA - DNA = B
o
Note that the simple formula (2.2), which is essential for the
independent of A.
but this time we pull back the bundle W' (and its complement W")
that +
DA, has index n dim V. Theorem 2 then follows easily if one
n follows by interpolation.
similar but more careful count of eigenvalues shows that Theorem 2 for
2 for d odd, they then follow for d even. In fact for d odd we
i* and 2*. The reason why the odd case yields stronger results is
roughly the following. In §2, for d even, we used the index theorem,
recall.
values 1 are now functions of t and when t goes once round the
3
circle the 1 have, as a set, to r e t u r n to their original position.
= - -
St + Dt
* : _ ~__
~t + D t
D t = -i ~ + t.
-ix ix
Dt+ 1 = e Dt e
on the torus S 1 × S 1.
fn(t) + (n + t ) f n ( t ) = 0
and so
2
fn(t) = C n ex~ (n+t) }
2
flow.
DNA act on different factors in the tensor product they commute, and
so
[DNA,F ]= X
locally given by
X = Ze F -1 $F.
1 1
At = (l - t ) h + t F(A)
of Dirac operators is
(3.4) D t = D A + tX.
a spectral flow. Moreover the general formula for the spectral flow
when M is a flat torus. For this they use the index t h e o r e m for
flow of a o n e - p a r a m e t e r family.
References
I. M.F. Atiyah and I.M. Singer, The index of elliptic operators III,
2. M.F. Atiyah, V.K. Patodi and I.M. Singer, Spectral asymmetry and
71-99.
W. Ballmann
Mathematisches Institut
WegelerstraBe 10
5300 B o n n I
not flat.
Vice versa, manifolds of rank one should resemble negatively curved mani-
folds.
is t h e following:
Note that rank (M) = I if M has a point p such that the sectional
curvatures of a l l tangent planes at p are negative. In p a r t i c u l a r ,
rank (M) = 1 if M is a c o m p a c t surface of n e g a t i v e Euler characteristic.
i) [BB] If M is c o m p a c t , then gt is e r g o d i c .
ii) [BI] If M has finite volume, then gt has a dense orbit.
iii) (Eberlein [B2]) If M has finite volume, then tangent vectors
to c l o s e d geodesics are dense in SM.
2
Theorem 8. Suppose that K M ~ -a and M has finite volume. Suppose
M* is an irreducible locally symmetric space of n o n c o m p a c t type and high-
er r a n k with finite volume. If t h e fundamental groups of M and M* are
isomorphic, then M and M* are isometric up to n o r m a l i z i n g constants.
References
[Mo] G.D. M o s t o w , S t r o n g R i g i d i t y of L o c a l l y S y m m e t r i c S p a c e s , A n n a l s
of M a t h . S t u d i e s No. 78, P r i n c e t o n U n i v e r s i t y Press, P r i n c e t o n ,
N e w J e r s e y , 1973.
[mr] A. P r e i s s m a n n , " Q u e l q u e s p r o p r i ~ t ~ s g l o b a l e s d e s e s p a c e s de R i e -
m a n n " , C o m m e n t . M a t h . H e l v e t i c i 15 (1943), 175 - 216.
[Y] S . - T . Yau, S e m i n a r o n D i f f e r e n t i a l G e o m e t r y , A n n a l s o f M a t h .
S t u d i e s No. 102, P r i n c e t o n U n i v e r s i t y P r e s s a n d U n i v e r s i t y o f
T o k y o Press, P r i n c e t o n , N e w J e r s e y , 1982.
Metrics with Holonomy G2 or Spin (7)
by
Robert L. Bryant
n2
example, if H preserves an orthogonal decomposition ~n = ~nl • ~ ,
n.
then g = gl + g2 locally where gi is a local metric on ~ ~. It
(i) SO(n)
(vi) G2 if n = ?
(vii) Spin(?) if n = 8
After noting that the above list is exactly the list of subgroups
(vi) and (vii) do not occur as s y m m e t r i c spaces [6]. This raises the
therein.
GL(7) is open. (In fact, there are exactly two o p e n GL(7) orbits
G
2
A = span{l,P,*F,*l = (i/7)~ /x *~}
G2
where A c A ( ~ 7) is the s u b r i n g of G2-invariant exterior forms.
the dual 4-form with respect to the underlying metric and orientation.
from such an ~.
valid for any positive /3 where the ~ is the Hodge star of the
the 4 - f o r m on ~8
0
~ = ~ A Io + ~ = *~
where ~ = *~ e A4(~7).
from such an ~.
appear elsewhere.
ly, that there exist local metrics on ~7 and ~8 which are not
factory.
is a c t u a l l y flat [II].
writing.
BIBLIOGRAPHY
I. K o b a y a s k i , S. a n d N o m i z u , K., F o u n d a t i o n s o_f_fD i f f e r e n t i a l
G e o m e t r y , W i l e y a n d Sons, N e w York, 1963 a n d 1969.
by
S,S. Chern
Mathematical Sciences Research Institute
1000 Centennial Drive
Berkeley, California 94720
R.S. Hamilton
Department of Mathematics
University of California, San Diego
La Jolla, California 92093
(I} A: B × B --* T M / B
d ~ f V l , W ) = dco(V0,W)
(5) dea~eavAev~s+Ra~vswvAws,l<a,~,v,~<dimM,
which has the property that, for any unit v e c t o r V, K{V,V) is the
Riemannian sectional curvature of the plane V ±. The structural
equations then take the following form.
d~ 1 = WzA~3 - W3A~2,
dw 3 = WlAW2 - ~2Aw1,
and
d~ 1 = ~2A~o3+K11~o2ALo3+K12Lo3A~o1+K13~o1A~02,
(9} W = (K11+K22+2K33+4)
Heisenberg group H 3.
(10) x2+y2+z2+w 2 = 1
in R 4. Differentiating we get
(11) ~ 0 = xdx + y dy + z dz + w dw = 0.
e I = x dy - y dx + z d w - w dz,
02) w2 = x dz - z dx + y dw - w dy,
= x dw - w dx + y dz - z dy.
~3
<Wl,Wl > = 1, <~1,w2 > = 0, etc., and that <¢~0,w0 > = 1, <w0,Wl>
= 0. e t c . T a k i n g e x t e r i o r d e r i v a t i v e we h a v e
(13) dw I = 2 w 2 A ~ 3, d ~ 2 = 2w3Aw 1, d ~ 3 = 2 ~ l A ~ 2,
and h e n c e m t h i s b a s i s
which makes
a n d t h e o t h e r e n t r i e s a r e zero. T h e W e b s t e r c u r v a t u r e W = 1.
P
K - c - ~ +I , when g-0,
(17)
t -I , when g>l .
1
(18) ~" (e 2I + 022 + 012).
Putting
we find
(20} d~ 1 = 2 ¢ w 2 A w 3, dw 2 = 2c~3A~01, dw 3 = 2 W l A W 2,
and
It follows that
W ~ c.
287
(23)
(24) ~+w+W=O
Z --.# Z + 8,
{25)
w'-'* w - K z + b.
{28) z = x + iy w = u + iv.
(29) x 2 + y2 + 2 u = 0
(30) du + x dx + y dy = 0.
and we compute
I
dwliO, dw2=O, d w 3 = 2 ~ i A w 2,
(32) ~ I = I ' ~ 2 = ~ 2 " ~3=-w3 '
K11 I , K 2 2 = I , K33=-3,
and the other entries are zero. By (9) we have W=0. All these
examples give metrics adapted to a c o n t a c t form w=w3, since in an
orthonormal basis =w3 = wlAw 2.
In general, given a metric adapted to a c o n t a c t form ~, we
shall r e s t r i c t our attention to orthonorma] bases of 1-forms Wl' w2'
~3 with w3=co. Considering the dual basis of vectors, we only need to
choose a unit v e c t o r in B. These form a principal circle bundle, and
all of our structural equations will live naturally on this circle bundle.
I t turns out to be advantageous to compare the general situation to
that on the Heisenberg group. Therefore, we introduce the forms
q/l' ¢/2' ¢3 and the matrix L l l , L12 ..... L33 defined by
d~ 1 = ¢2A~3-~3Aw2,
(34) dw 2 = @ 3 A W l - C l A W 3 ,
d~ 3 = 2~1A~2,
and
{ ~lAw2 - W2A~I = o,
(35)
¢IA~I + ~2Aw2 = o,
and
de 1 = @2A@3+LllW2Aw3+L12w3A~I+L13WlAW2 ,
(36) de 2 = ¢3A¢l+L21w2A~3+L22~3A~I+L23w1Aw2 ,
de 3 = ~lA¢2+L31~2A~3+L32~3A~l+L33wIA~2 .
@I = aWl+bW2'
(37)
~2 = b W l - a ~ 2 "
Q = wl+i~ 2, W = W 3,
t = @1+i~2 , ~ = @3'
Z = a+ib,
p - (LII +L22 ) , q
~l . ~1 (LII-L22), r = L 12'
(38) s q+ir,
=
1
z - ~ (LI3+iL23) ,
t=L
33'
w = ~l ( t _ a 2 b 2 ),
(39) dO = i(@AO-zQA~),
and
{ d@ = i[2wQA~+(zO-iQ)A~],
(4O) dz ~ i(2Z@+zO-sw) mod O,
p+l~l 2 = O.
(41) df = D a f • w a,
(43) f = T(Vv),
then,
then
{471 df = a ¢ . O + ~ f - O ÷ D v f . W
292
~f = ~ (Dlf - iD2f),
(48) ~f - ~ (Dlf + i D 2 f ) ,
Dvf " D3f ,
3.9. [,emma.
w ~ = cos.8 Wl-sin, e w 2,
(51)
~ = sin.e Wl+COS.e co2
(52) Q* = e iO O.
4.1. Lemma.
¢= = ¢ + dO,
(53)
z" = z e 2i~.
2 2
a{~l-W 2) + 2b ca)-]t~ 2 ,
(54}
- 2°
w~ = Au)I+B~2,
(55) ,~ = C¢~I+DW2,
w ~ = t~3
~ i = gWl+hC~ 2,
~ 2 = kwl+lw2'
w3=0
~ 1 = gc°l+h~2'
(56) ~2 = h~l-g~2'
to3 = 0.
{57) f~ = fQ and ~o = O.
and that
4.2. Lemma.
z = -i Dvf,
(59)
J
= i(Df-CY-~f-Q) - {zf+~f)w
w~ = f- ~l-D2f- uJ3,
(60) ~ = f . w 2 + D l f . w 3,
uJ~ = f2u>3,
In c o m p l e x n o t a t i o n
O" = fO + 2i ~ f ' w ,
(61)
~, = f2~.
F o r an i n f i n i t e s i m a l v a r i a t i o n w e d i f f e r e n t i a t e t o o b t a i n
/ / /
0 = f 0+2i ~ f -~,
{62)
/ /
w = 2f w.
(4.3. L,emma.)
296
4.3.
(65)
z* = z - 2 " ~ ' - 6 2
4.4 [,emma.
/' s s
(67) W = -of - 2 f W.
(69) EW =I MW~t,
which is analogous to the energy
(70) E = f R
J M
so nothing is lost.
Next we observe that we can integrate by parts.
~p d(fQA~A~} = 0
and the second follows from
~ p d{fOA~q~} = O,
S
Proof. W e compute the infinitesimal variation E W . Fixing the
CR s t r u c t u r e and varying the potential f of the c o n t a c t form with w s
= f2w gives ~ = 4f ~ and
EW = {-Elf + 2f W) v = 2 f W y,
P
5.3 Lemma.
(75) ~ = ~ Ipd¢AwA¢.
dv~A~ = 2iWOAQA~a
EW = A~A¢ + dCA~A¢
EW" = - ½ I p (Lf+~f) v.
Next we c o n s i d e r t h e e n e r g y E L .
Proof. T h e e n e r g y E L is g i v e n by
Ez
_-f p
so its f i r s t v a r i a t i o n is
E z = (L ~ + L 2 )p + z y .
P
5.5. Lemma.
E~
-= 2 Im
f P {Dv~ ~ .
S i n c e f is a n y r e a l f u n c t i o n on M, we s e e Ez = 0 w h e n DVL - 0.
T h e n s = 0 by Lemma 3.9 and K l l = K22 a n d K12 = 0 by s u b s t i t u t i o n
{38).
This c o n d i t i o n says that, at each point of M, the sectional
c u r v a t u r e of all p l a n e s p e r p e n d i c u l a r t o t h e c o n t a c t p l a n e B a r e equal.
If, on t h e other hand, we fix t h e CR s t r u c t u r e and v a r y t h e
c o n t a c t form by a p o t e n t i a l f, we have from Lemma 4.3
300
Z'= ~: - 2 -6 2.
Ez
"
=
fP
(-2(~ a~f
"
+ ~f'} + 3
IIL 2 f ' } ~,
2( a, + 3 l,j 2
a) C- is not empty.
b) P is locally invertible.
I S I
f + 2CNf = -W ,
1 1
(78) /W)min
1 ~< f min ~< f max ~< /W) max
1
Notice that the estimate fails ff W and ~N are positive. Having
control of the m a x i m u m and minimum of f, it is easy to control the
303
.__,
this shows wn & in S kP + 2, and proves P is proper.
The assertion (d} that ~- is simpb" connected follows by
shrinking along straight line paths to W = - I . This completes the
proof of the theorem.
(81) Ez P
o = i Dr,-5,
2
z = DvL.
metric is obtained as the limit of the heat equation flow as t --* **.
REFERENCES
J s
APPENDIX
by
Alan Weinstein
contact form. Since the flow of V' leaves the metric invariant, it
leaves the invariant the form to , from which it follows that V' is the
P
contact vector field associated with ~o .
Having made the changes described in the previous paragraph,
we may revert to our original notation, dropping primes, and assume
that the flow of V is periodic. A rescaling of • will even permit us
to assume that the least period of V is 1. {Note that, by Gray's
theorem [2], w e could actually assume that the n e w contact structure
equals the one which was originally given.)
Suppose for the moment that the action of S] = IR/Z
generated by V is free. Then M is a principal S 1 bundle over the
surface M / S 1. The form w is a connection on this bundle; since
is a contact form, the corresponding curvature form on M / S I is
nowhere vanishing. Thus M / S 1 is an orientable surface, and the
Chern class of the flbration M --* M/S I is non-zero. By the
classification of surfaces, bl(M/SI) is even; by the Gysin s e q u e n c e ,
b1(M) = bI{M/SI) and is therefore even as well.
We are left to consider the case where the action of S I,
although locally free, is not free. The procedure which w e will follow
is that of [8]. Let r~s I be the {finite) subgroup generated by the
isotropy groups of all the elements of M. Then M is a branched
cover of M/r, and M / r is a principal bundle over M / S I with fiber
the circle S1/r. The branched covering map M --, M / r induces
isomorphisms on real cohomology, so it suffices to show that b1(M/r)
is even. To see this, we consider the fibration
s 1 / r - - . M / r - - , M / S 1. The quotient spaces M/r and M/S I are
V-manifolds in the sense of [4], and w e have a fibre bundle in that
category. The base M / S I is actually a topological surface which is
orientable since it carries a nownere-zero 2-form on the complement
of its singular points. Now the contact form may once again be
considered as a connection on our V-fibration, and so, just as in the
preceding paragraph, w e m a y conclude that b1(M/r) is even.
[5] has shown that the first Betti number of a compact Sasakian
manifold is even. On the other hand, since every almost complex
structure on a surface is integrable, every 3-dimensional K-contact
mainfold is Sasakian, and hence our result follows from Tachibana's
theorem. In higher dimensions, compact symplectic manifolds with odd
Betti numbers in even dimension are known to exist [3] [7], and circle
bundles over them will carry K-contact structures, while having odd
Betti numbers in even dimension.
The paper [6] contains a study of which Self err fiber manifolds
over surfaces actually admit sl-invariant contact structures.
Acknowledgments. This research was partially supported by
NSF Grant DMS84-03201. I would like to thank Geoff Mess for his
helpful advice.
R E F E R E N C E S
S.K. Donaldson
All Souls College
Oxford, England
Topology of b u n d l e s .
This is standard material that may be found in [2] for example. Con-
sider a fixed manifold X and a family of b u n d l e s over X parametrised
by some auxiliary space T , so w e have a bundle P over the product
X × T with structure group G (compact and connected, say). Take first
the case when T is a point so we h a v e a single bundle over X , deter-
mined up to equivalence by a homotopy class of m a p s from X to BG .
This may be non-trivial, detected for example by characteristic classes
in t h e cohomology of X . If w e choose a connection A on the bundle
the real characteristic classes can be represented by explicit differen-
tial forms built from the curvature of the connection. Equally if D is
an elliptic differential operator over X then using a connection it m a y
be extended to act on objects (functions, forms, spinors etc.) twisted by
a vector bundle associated to P . This has an integer valued index:
~(~) = c2 (P)/~
Here I only want to say enough to fit into our overall theme; more
details and r e f e r e n c e s may be found in [4], but I learnt the p o i n t of
v i e w we are adopting now from lectures of Quillen.
acted upon by G~ , and realising via the first Chern class the cohomo-
logy class obtained under our map ~ from the fundamental cycle of the
curve C , as in S e c t i o n I.
n Es+m <0 11
lic" form (~
\, !h is d i s t i n g u i s h e d by the fact that the symmetric power:
U/
k+1
(H 1 @ ... @ H k)
316
References:
[2] Atiyah, M.F. and Singer, I.M. "The index of elliptic operators
IV". A n n a l s of Math. 93 (1971) 119-138.
G. F a l t i n g s
Fachbereich Mathematik
Universit~t-Gesamthochschule Wuppertal
GauSstr. 20
5600 Wuppertal I
INHALTSVERZEICHNIS
§ I EINLEITUNG
§2 DEGENERIERENDE ABELSCHE VARIET~TEN
§3 MUMFORD'S KONSTRUKTION
§4 KONSTRUKTION VON AG
§ 5 LEVEL-N-STRUKTUREN
§ 6 MODULFORMEN UND MINIMALE KOMPAKTIFIZIERUNG
§ 7 ETALE GARBEN
§8 DIE TORELLI-ABBILDUNC
§ 9 DIE KOMPLEXE THEORIE
§ I EINLEITUNG
Z u n ~ c h s t b e t r a c h t e n wir d e g e n e r i e r e n d e a b e l s c h e V a r i e t ~ t e n und o r d n e n
ihnen q u a d r a t i s c h e F o r m e n zu. Dies w i r d zum einen benutzt, um sp~ter
die K o m p a k t h e i t zu zeigen, und m o t i v i e r t zum a n d e r e n die Wahl der Daten,
w e l c h e bei der v e r a l l g e m e i n e r t e n M u m f o r d - K o n s t r u k t i o n eingehen.
Fall erfordert keine neuen Ideen, sondern nut e i n e Reihe von Notationen
und Definitionen. Wir setzen voraus, dab G eine abelsche Variet~t
q
ist, und dab der maximale Torus yon Gs zerfNllt. Dann ist die formale
A
Kemptettierung G eine Erweiterung einer formalen abelschen Varietfit
A
A (entsprechend einem A ~ber R ) durch einen formalen Torus
~ ~ G~ . Es gibt eine Gruppe G , mit ~ ~ , so dag G eine Erwei-
terung von A durch T = Gr ist.
m
0~T~G~A~0
~F--> 0 ,
0 ®0 ---0
w Z+~
A
Das formale Geradenb~ndel L ist d a n n samt seiner kubischen Struktur
A A
Pullback eines M auf A , welches eine prinzipale Polarisation f~r
A A
A definiert. M kommt von einem M auf A , und ~ auf ~ sei das
Pullback von M . Dann ist L isomorph zu L , wobei der Isomorphismus
die k u b i s c h e Struktur respektiert. Allerdings ist d i e s e r Isomorphismus
nicht eindeutig, sondern kann mit einem Charakter X : ~ ~G
modifi-
m
ziert werden. Es sei n o c h b e m e r k t , dab e i n solcher Charakter eindeutig
bestim~t ist durch seine Einschr~nkung ~ C X auf T , und dab man auf
diese Weise genau alle ~'s erh~it, welche im K e r n der Abbildung
X ~ Pic0(A) liegen.
C : X~A(R) ,
Definition:
i) c(~)*(0v) ~ 0w
ii) M ~ c(~)*(M) , so dab
a) Die Isomorphismen in i) sind linear in b und
b) FOr ~,w 6 X kommutiert das D i a g r a m m
M
-~+w
<---~ -M~ ~0 v
<---~-C(~)*(~)
-
ec(~)*(0
--
v)
A A A
Den formalen Schnitt (~r, 6 I" ( G , L ) kann m a n nun nach T-Eigenfunktionen
entwickeln :
@L = ~EX
Z a(~). @M ~
Satz I:
Bemerkung:
~>G : G × G - ~ G x G
ein P r o d u k t ist°
Wir w~hlen nun ein H , welches T[2] umfaBt. Dann ist H/T[2] = HI~A[2]
maximal isotrop, und jedes solche HI kann man auf dies® Weise erhalten.
M a n l~Bt n u n H so auf L ® 2 operieren, dab m a n den ~ b e r b l i c k [ber
die z u l ~ s s i g e n Isomorphismen nicht verliert. D a z u gehe m a n f o l g e n d e r -
m a B e n vor:
c
c I : X:A+A I
0+T : T I +~-1 + A 1 + 0
Sei p 6 X . Wihle ~ : H(R) ÷ {±I} mit sIT[2] : piT[2] . Dann ist der
folgende Schnitt von £2 @ £ 2 ein Produkt:
A A
Z 8(z) OL(x+y+z) @ 8L(x-y+z)
z6HIR)
A~ A
zCH(R)~ (z) a(R)a(v) 0 M ( x + y + z ) @ OM (x-y+z)
~,vOX
Bei festem b,v verschwindet die Summe ~ber H(R) (sogar schon ~ber
T[2]) , auBer wenn es ~,5 6 X gibt mit b = p+~+B , v = ~-B • Also
ergibt sich
P+~+[{(x+y+z) ® eMa-5(x-y-z)
~, BOX a(p+~+~) a(~-~) z~H(R)~(Z)UM
mit
¢
A : AxA+A×A
Wir wissen schon, da6 a(~) = 0 <=> a ( 2 #0-~) = 0 . Wit behaupten zu-
n[chst, da6 eine Untergruppe y cX existiert mit
iii) Y : X :
Andernfalls g~be es ein endlich flaches Untergruppenschema NcT,
N #(0) , so d a b a l l e ~ 6Y auf N identisch den Wert I annehmen,
A A
und 8G ist ein Eigenvektor fir die Aktion von N auf L . F~r
n
Elemente x I ..... x n £ G(R) mit jZI
= xj = 0 ist j~1= x j * ( 0 G ) ein
331
* ®n
globaler Schnitt von ® xj (L) --L , welcher ein Eigenvektor fur N
ist. G(R) ist Zariski-dicht in G , und es ist w o h l b e k a n n t , dab fur
@n
n_> 3 die o b e n definierten Schnitte von L eine projektive Einbet-
tung von G liefern. Andererseits muB diese Einbettung ~ber (G/N)
n n
f a k t o r i s i e r e n , w a s e i n W i d e r s p r u c h ist.
a(l+}~+~) . a ( l + ~ ) - 1 . a ( l + ~ ) - 1 . a ( b + v ) - 1
-I
-a(l) .a(~) .a(m) a(O) : I
A A
Wie bisher kann man 0G, auf jeder Komponente yon G* nach T-Eigen-
funktionen entwickeln. Man erh~it dann Koeffizienten { ~ ( ~ ) , b 6 X} ,
welche vonder Komponente abh~ngen. Wir mUssen zeigen, dab sie a l l e
verschieden yon Null sind, u n d daft ~(~+~) ~(0)/(a(~)a(v)) = ~(~,m)
bilinear und unabhingig vonder Komponente ist. Wir wissen s chon, daft
nicht alle ~(~) verschwinden. Wir schlieBen mit unserem alten Trick:
W~hle HcGcG* wie vorher. Dann ist fur jeden Charakter s :H(R) + {±I ]
und jedes x 0 6 G*(R)
332
Z s(z)(~G.(x+Xo+Y+Z)
z6H ® GG.(X-y+z )
f) SchlieBlich ben~tigen wir noch ein Resultat, nach dem b(~,~) und
die P o l a r i s a t i o n auf A die P o l a r i s a t i o n
von G bestimmen. Es seien da-
A An
ZU g e g e b e n zwei G's,G I und G 2 , so dab G I = G2 und damit
AI ~ A2 ~ A . Weiter nehmen wir an, dab Geradenb~ndel ~I und ~2 auf
GI bzw. G2 existieren, welche prinzipale Polarisationen auf den ge-
nerischen Fasern liefern und a u c h dieselbe Polarisation auf A ergeben.
(d. h. ~I und ~2 unterscheiden sich um eine Translation). SchlieBlich
sollen ~I und ~2 dieselbe Bilinearform b liefern.
Satz 2:
A A
Unter diesen Umst[nden ist der formale Isomorphismus GI --~> G2 alge-
braisch, d. h., er w i r d induziert yon e i n e m Isomorphismus polarisierter
abelscher Variet~ten G1,B ~_T_> G2,n
A A A A
L I ~ [-I]*_L I ---L2@ [-I ]*L 2 = Pullback
von M@ [-I]*M
Wir zeigen, dab sich bei diesem Isomorphismus die algebraischen Schnitte
F(GI,LI @ [-I]*~i ) und F(G2,~2 @ [-I]*L2 ) entsprechen. Genauer gesagt
333
zeigen wir, dab man ein Erzeugendensystem der algebraischen Schnitte von
~I @ [ - I ] * L oder von [2 ~ [ - I ] * ~ 2 erh~it durch die Reihen
Z
zCH(R) E ( z ) ^~L.(x+y+z)OL. (-x+y+z)
--3 --3
-c(8)*(8MP+C~(x+y+z) @ QM(-x+y+z))
z 6 H (R)
334
Die innere Summe ist e i n Schnitt yon M ® [-I]*M , (welcher noch von
y abh~ngt), der sich unter H (bei e i n e r geeignet zu d e f i n i e r e n d e n
Operation von H auf d i e s e m B@ndel) als E i g e n v e k t o r transformiert. Der
zugeh6rige Charakter ist u n a b h ~ n g i g von j = 1,2 , und somit sind die
inneren Summen Vielfache voneinander, f~r j = 1,2 . Es folgt, da~ der
A ~ A
formale Isomorphismus ~1 ® [ - I ] * L ] ~ ~ 2 ® [ - I ] * ~ 2 einen Isomorphismus
§ 3 Mumford's Konstruktion
Y --> ~(K)
E(~+v)*(M) -~>E(~)*(c(~)*(_M))~c(~)*(M v)
Die Abbildungen in ii) , iv) und vi) sollen kompatibel sein in d e m fol-
genden Sinne: b entspricht einer Abbildung y b> T(K) . Dann sei f~r
6 Y i ( y ) b ( y ) -I 6 G(R) , und die Isomorphismen in iv) werden definiert
durch Translation mit diesem Element.
Wir definieren auf den Daten noch eine ~quivalenzrelation, wie folgt:
Sei c : Y x X~R* eine Bilinearform, so dab eine Funktion q : Y ~R*
existiert mit
Dann erlauben wir, dab man die Isomorphismen in v) mit q(~) , die in
iv) mit c(z,v) multipliziert, und schlieBlich b durch b - c ersetzt.
Aus der Definitheit der Form b folgt sofort, dab i : Y ~G(K) eine
Injektion ist.
336
Sb*
Tg *S ~ *(L)
_ ~ > T g *(L_ @ 0_~) = Tg* (L)
_ @ Tg* (0 - ~)
II Tg
Sg*Tg*(L)_ _L®Tg*(0-~)
S *(L)
_ > _L @ 0 - P < L@x*(0
_ -~ )
Bemerkung:
Tg*S~x* : ~ (g) ~B * T g*
S~*(_M~=LI~r~} ~ S~*(_Mffnl---->s(~)*(_MIZnl
----> (M_~0~]Zn)~ (M~Le0~)IZn-~-> (_Me_~)IZn
Es ist a(0) : I , a(b+v) : a(~) a(v) b(b,v) (Man b e a c h t e , dab auf
338
}1 : 0A ® n>1@ 0~ 8 n : 0 A • n ~ 1 0 ~ • 8 n
_ "O n
~2 =
0 A e n91 0 ~ _Nn-0 n = _
0 A ® n>~1 (~n
± @ 0))
vCX
= a(p)n_c ( ~ ) * ( % ) i ( b ) * ( f j 0 n) 6 M n ® 0 v + n ®RK'8 n
{ S *(M®
_ Oc~'O ) I ~6 Y,c~6 E}
[M4], 3.1:
~ * : S ~ *(L)
-- ~ > _L ® 0 -~ und Ti(~) * : Ti(~) *(L)
-- ~ > _L
340
[M4], 3.2:
q q
[M4], 3.3:
[M4], 3,5:
[M4], 3.6:
s (U0) n s (U 0) :
Beweis:
[M4], 3.7:
[M4], 3.8:
P0 ist zusammenhangend
Beweis:
[M4], 3.10:
A
abgeschlossenes Unterschema B~P , so d a b B : Quotient von
A ~ A
(P-~Yt S (G))A~ . Sei G : P-B . Es ist G = G Q
[M4], 4.2:
[M4], 4.3:
P ist irreduzibel
Es gilt:
c) Sei W3 = W2/Yc P
A
d) Sei W3 E P definiert durch W3 = W3 .
e) GI = W3 D G .
Nur Schritt b) ist nicht trivial. Er folgt aus einer Variante yon [M4];
343
Es folgt:
Z (n) - - > G*
I ~n
{~} --> G*
[M4], 4.10:
Beweis:
Wie in [M4]
344
[M4], 4.11:
0 ~ t ° r sI ~ Gt°rs
I ~ YI ~(~/2Z) ~ 0
[M4], 4,12:
G ist s e m i a b e l s c h
A A A A
C) Das ample Geradenb~ndel 0(I) auf P erf~llt 0(]) IG
Da ~ auf L[~ operiert, ist L[G kanonisch trivial, und somit er-
A A A
h~it L®M~ M eine k u b i s c h e Struktur. Wir zeigen, dab diese m i t der
A
kubischen Struktur auf 0(I) Obereinstimmt. Dies ist der Fall, wenn
A
die k u b i s c h e Struktur auf 0(I) vertr~glich ist m i t der T - O p e r a t i o n
A A
(T operiert auf L und (trivial) auf M ) . Dies ergibt sich aus den
nun folgenden 0berlegungen:
: ~mj : GI = ~3 ~ 8
m * ( ®j0 ( 1 )
--
-+I) ~ m- -* ( ® j (^L ® M^) ~:I) ~ m- -* ( ® Lj -+I)
345
Z( Z ~j(gk)) ±I
J j,k6J = I
A
Da die kubische Struktur auf 0(I) durch ihre Y-Invarianz eindeutig
bestimmt ist, folgt die Behauptung.
Weiter k6nnen wir den Grad der durch 0(I) auf Gq definierten Pola-
risation berechnen:
A
O = Z O (~)
'~EX
e(#+,J) = s *(0(v))
A A A A A A
F (~,n ® M) Vc I'(G,n ® M) v = F (A,M ® % )
Insgesamt folgt :
dual
Abbildung A ~ A )
Es f o l g t insgesamt (f~r b e l i e b i g e s Y ):
= ~[Y S * ( r . 8 ( v ) )
Wenn ~ ein Vertretersystem f~r X/Y durchl~uft, und 9(u) eine Basis
yon F(A,M® 0 ) , so e r h ~ I t man auf diese Weise eine Basis yon
r(Gn,0(1))
A
8 = ~ y a(~) b(~,'o) c (~)* (r.9(~)) ,
Kleinigkeit:
Wir haben bis jetzt ein relativ komplettes Modell nur unter der Annahme
konstruiert, dab
6 Y, ~ 6 ~ . Wir wollen noch aufzeigen, wie man dies fallen lassen kann:
(*) ist dann immer erfNllt, wenn man for ein genOgend groBes n
ersetzt durch Mn t Y dutch n.Y t i : Y~G(K dutch i -~ : nY~(K),
-- n
G--G 1 ~G 2
n n2
mit Geradenb~ndel N, N = N I,N2 = N Wir ben~tigen noch Information
iiber d i e globalen Schnitte F(G,N) . Wir kennen schon die globalen
nI n2
Schnitte von ~I = ~ und ~2 = ~ : Man erh[it zum Beispiel eine
n
Basis von F ( G q , N I) wie folgt: Durchlaufe m ein Vertretersystem von
-- nI
X/nIY , und 8(~) eine Basis von F(A,M ® 0 ) . Dann gibt ~? f~r ein
passendes r 6 R , r # 0 , eine Basis aus Elementen 9 6 F(G,N ) mit
A
0 = r- Z a ~)b(~,~) c(~)*(8(~))
~6niY
Es liegt dann nahe, dab man eine Basis von F(G,}j) erh~it aus 8's mit
348
A
0 = r- ~ cyZ a(p)b(~,v) c (p)*(0(v))
-- ,
~6X/Y, % (v) 6 F ( A , M ® 0 )
A
Und in d e r T a t r e c h n e t man nach, da6 diese 0 folgende Bedingung er-
f~llen: FOr ml,m 2 ganz und positiv mit mlnl-m2n 2 > 0 ist
Bemerkung:
Die Funktion a(p) h~ngt yon der Wahl des relativ kompletten Modells
ab (siehe die Bemerkung nach der D e f i n i t i o n eines solchen). Wirklieh
wichtig ist n u t die Bilinearform b mit a(~+m) = a(p)a(v)b(~,m)
Der Leser wird sich leicht ~berzeugen, dab w i t in d e r T a t g e z e i g t haben,
dab m a n b e i passender Wahl des kompletten Modells alle Funktionen a( )
erh~it, welche dieser Gleichung gen~gen, und fur d i e e i n r 6 R, r # 0
existiert mit r.a(p) 6 R. Zwei verschiedene unterscheiden sich um
einen Homomorphismus Y ~ K* .
Wir formulieren nun das Hauptergebnis dieses Kapitels. Der Einfachheit
halber betrachten wir nun prinzipale Polarisationen.
Satz 3:
0 ~ T ~ A ~ 0 ,
Es gilt:
Dann ist
A
8N = p6XZ a(~) _c (Z)*(%M) , mit a(z) C R , a(~) # 0 , und
G : ~/i(X) , wobei
i : X~G(K) wie folgt zu e r k l ~ r e n ist:
I I (~ 1
0 ~[~ROR 0G~Q~ ~G/R~0 : ~R/~)
I
W e i t e r gibt die e r s t e C h e r n - K l a s s e c(N) 6 HI (G,~G/R) einen Morphismus
I I
! G ® ~R ÷ H I ( G ' ~ R @ R 0 ~ ) , w e l c h e r im g e n e r i s c h e n P u n k t ein Isomorphismus
wird. Man kann dann K auffassen als B i l i n e a r f o r m
I
350
Wir nehmen n u n an, da~ (~,M) und das vertr~gliche System von Isomor-
phismen schon 0ber einem Unterring R0 ~R definiert sind, und betrach-
ten,statt der absoluten Differentiale, Differentiale relativ R0
K : ~ × ~*G ~RK/R 0 verschwindet dann auf ~A x ~*G ' und definiert eine
Lemma :
K(~,v) = d log(b(~,v))
Beweis:
~ T ~ ! G ~ HI (G'0G)
A A
Sei 1 : X~R eine Linearform. Da P : ~/X , definiert 1 eine Klasse
in HI (P,0~)-
= = HI (P,0p) . Deren Einschr~nkung auf G ist d a s Bild von
1 . Andererseits sei ~ 6 X .
351
F(P-~/R~) @R K .
§ 4 KONSTRUKTION VON AG
a) Wir kommen nun zum Hauptziel unserer Bem~hungen, n~mlich der Kon-
struktion eines Ober ~ eigentlichen algebraischen Feldes, welches
A als offene Teilmenge enth[it. Wie schon in d e r E i n l e i t u n g erw~hnt,
g
ist fur uns e i n algebraisches Feld eine Art Quotient S/R , wobei
ein Schema von endlichem Typ Ober Z ist, und R~S x ~S eine endliche
Abbildung, welche R zu e i n e m Gruppoid macht ~ber S . AuBerdem wird
vorausgesetzt, dab die P r o j e k t i o n e n von R auf S @tale sind. Man
Oberzeugt sieh leicht yon der ~quivalenz dieser Definition mit der in
[DM] :
+
B (X) = U~
352
+
mit rationalen Kegeln ~ ~B (X)~ (Die d sind d i e konvexe H~lle
endlich vieler rationaler Halbgeraden in B+(X)~ )
Diese erf~lle
iv) Jedes u wird aufgespannt von einer Teilmenge einer Basis yon
B(X)
I
adIz) = ~ I z ® b+id(~) 6~. [S2(X) ]
Xd x X d ~ B ( X d ) * : S2(X )
b : X ×X ~K*
0 o @ d
s{a) a : ~ [ B { X a ) * n o ~ ] ) ,
b :X xX ~K* ,
G O (7 C
R 0 •
A
Satz 3 liefert dann ~ber R (:m-adische Komplettierung yon R ) eine
A A A
abelsche Variet~t G mit einem Geradenb~ndel N , so dab (G,N) --- ( ,M) ,
und dab die zugeh~rige Bilinearform (nach Satz I) gleich b ist. Nach
dem Approximationssatz yon M. Artin (siehe z. B. [A]) kann man annehmen,
dab G und N schon [ber R definiert sind.
^I ^1
O _ , T I _,~I _,GI -*0 ,
0 * ;2 0 ,
und
0~ (9 2 / 912) ~ ~1 2 ~ AA~ 0
i) huf ~I :
~1 = E
# E X I @I (#)
A
Dabei sind die 8 1 (~) ~-Eigenfunktionen unter T 1 , und konvergieren
gegen Null in der I-adischen Topologie
A
ii) Auf G2 :
~.2 = Z a(N) c ( ~ ) * ( G M)
~EX -- "
O
A
012(v) : p,+vZ a(#) _c(P')*(0 M)
Andererseits ist b(~,~) = b' (b,#) = Einheit mod I . Also ist tats~ch-
lich XI = X
wobei:
ii) a ' (0) = 1, a' (p+v) = a' (#) a ' (~) b' (p,v)
Damit folgt:
A2
e 1 (v) = a 1 (v) Z
I/+~0
a ' (;z) --
c (#)*(6 M) , v E XT
Man Oberlegt sich ~brigens leicht, dab man erreichen kann, dab
A A
a' (g) E K regulfir ist auf dem P u l l b a c k yon U
c Spek(R) in Spek(R)
o T--
(Dies gilt schon, falls # C Kern(X ~ X) , und sonst modifiziere man
A
a3 und a' mit einer linearen Abbildung X ~ K* . )
Z e 1 (#)
Y
A
induziert dann die entsprechende Zerlegung ~ber RI . Da
A2
e t(v) = a~,(v) p÷vZ a ' ( ~ . ) _c ( ~ ) * ( e M) ,
357
A
und a'(~) = Einheit in R I , folgt dab 01(v) = a1(~) (regul~rer
Schnitt, % 0 in s I ) . Es folgt, dab bI und die Bilinearform zu
aI (dies ist b7 ) sich nur um E i n h e i t e n unterscheiden.
x t* ÷ ~/R T ® A
~*G -- G I R RI
Beweis:
A I A
Bild(t* G ®~*A ) ~> ~ R / R o ® RRI
I A
S2 (~*(T/T]) ) ÷ ~ R / R O ® RRI
fNr ~, v6 X ( T / T I) = Kern(X ° + X )
I
Die d log(b' (~,v)) bilden aber eine Basis von 9Z[s,]/Z oder auch
I A
~R / R T ®RR1 , und es folgt alles.
e) Lemma:
T T T
und ~T ist Pullback eines Geradenb~ndels ~T auf ~T
-- - 7 7
~6X
T
Welter zerlege man b = b -b' Dann existiert ein
T
A7
ao 6R7 = F(U ,00,[) , ao % 0 ,
At ^
SO dab 8y(V)87(-v) = b (v,v).a "f , wobei der Schnitt
T ) auf U in keinem Punkt identisch auf der Faser
-- T
verschwindet.
Beweis:
Punkt i) ist klar. F~r ii) w~hlt man a0 so, dab die Bedingung f~r
= 0 erf~llt ist. Dann gilt sie auch in allen Punkten s] 6 U ,
welche s in ihrem AbschluB enthalten (nach den vorherigen ~berlegun-
gen ~ber Bilinearformen). AuSerdem reicht es, sie f~r eine gewisse
endliche Anzahl von ~'s zu v e r i f i z i e r e n . (Wegen des Zusammenhangs
359
mit Bilinearformen).
Korollar:
Beweis:
A
Sei RI der lokale Ring in s1,R I seine Komplettierung. Betrachte die
A
Zerlegung in G in s 1 :
A A A A A
0~T I ~G~A~0 , T I : TT .
A
Wenn R2 eine verselle Deformation ist v o n A (mit P o l a r i s a t i o n ) und
der Erweiterung durch T I , so ist d i e A b b i l d u n g R2 ~ R7 ~ RI 6tale,
nach Teil iii) des Lemmas. Weiter gilt fur die zur D e g e n e r a t i o n geh~-
A
rige Bilinearform bI : X ×X ~K~ , dab bl/b T Werte in d e n E i n -
A T • A
heiten RI* annimmt. Wenn man die Abbildung R T ~R I mit einem geeig-
A
neten Element aus S(T) (RI) twister, darf man annehmen, dab b I = bT ,
und alles hat seine Ordnung.
Satz 4:
Beweis:
A
R endlich ~ber R I ®R 2 (das k o m p l e t t e Tensorprodukt ist zu n e h m e n ~ber
der Komplettierung der strikten Hense]ierung yon ~ in e i n e m Primideal
p~, p % 0 ), n o r m a l , enthilt RI und R2 , und es ist
GI ® R R = G 2 @R R = G12 . Zu d e n degenerierenden polarisierten abelschen
I 2
Variet~ten geh6ren Gitter X I = X 2 : X12 und symmetrische Bilinear-
formen.
b I : X I x X I ~ KI*
b2 : X 2 × X 2 ~ K2*
°I : ~2 :
Angenommen o12 = oIRO 2 ist eine echte Seite von 01 (oder analog
von d 2 ). D a n n gibt es endlich viele Elemente ~ j , v j 6 X I , so d a b
fir jede Bilinearform bE0]2 E b(~j,vj) = 0 , abet dab diese Summe
positiv wird fur jedes b in oi-d. 2 , und negativ fur b £ o2-o12
Es sind dann [ b 1 ( ~ j , ~ j) und Hb2-1 (Zj,v j) Elemente aus mI bzw.
R2 , wobei das erste Produkt keine Einheit ist. Ihr Produkt ist aber
eine Einheit in R , und das geht nicht.
%1 : Spek(K)+
~2 : Spek(V') ÷S ,
§ 5 LEVEL-N-STRUKTUREN
Es g i l t nun:
Satz 5:
Beweis:
0 +~(n) ÷ G(n) + X o / n . X ÷ 0
@
Dabei ist ~(n) ~tale vom Rang n 2g-r@ , u n d die Faser ~ber der Klasse
modulo n.X von ~ C X ist isomorph zu
o
ii) Wie bisher sind die Projektionen yon R auf S 6tale, und
--n --n
somit ist --nR~ Sn x ~ [ ] / n , e 2 ~ i / n ] ~n unverzweigt.
Korollar:
a) Auf A e r h ~ I t m a n in n a t ~ r l i c h e r W e i s e e i n e R e i h e v o n V e k t o r -
g
b~ndeln. D a b e i ist e i n V e k t o r b ~ n d e l auf A gegeben durch ein Vektor-
g
b~ndel auf S , d e s s e n b e i d e P u l l b a c k s zu R i s o m o r p h sind (unter
ErfOllung geeigneter Bedingungen).
I
2.) Im a l l g e m e i n e n ist 9S/~ nicht lokal frei. Als Ersatz dient
besser das B~ndel @ , definiert wie folgt: F~r jede Torus-Einbettung
S o--
S o sei 0So das Unterb~ndel des direkten Bildes von ~ SI/ ~ , welches
von den d~/b erzeugt wird (b = C h a r a k t e r von S ). Da S lokal in
der 6talen Topologie isomorph ist zu e i n e m S O , erh~it man 0 auf
Wenn die Kegelzerlegung glatt ist, so ist @ die Garbe der D i f f e r e n -
tialformen mit logarithmischen P o l e n in ~ . D i e K o d a i r a - S p e n c e r K l a s s e
liefert einen Isomorphismus K : S 2 (~*G) ---~> @ (Dies f o l g t aus d e r
ziemlich expliziten Bestimmung von < ).
Auf jeden Fall ist p,(H) lokal frei auf S0 . Wenn wir Level-2-Struk-
turen einfNhren, also zu S~ ~bergehen, so e r g i b t sich eine irreduzible
Darstellung der @-Gruppe auf p,(I~) , und p,(H) ~ K 2 g , f~r e i n Ge-
radenbNndel K auf S~ oder auf Ag,2_i Aus dem Satz von miemann-Roch
folgt, dab in P i C ( A g , 2 ) @~ K und ~ dasselbe Bild haben. Eine
-I
Potenz von K stimmt somit mit einer Potenz yon ~ ~berein.
0+T ÷~ ~A÷0
A
Sei R die K o m p l e t t i e r u n g des R i n g e s R~ zu S(~)u in der I - a d i s c h e n
Topologie, wobei das Ideal I den a b g e s c h l o s s e n e n S(0)-Orbit in S(U)
beschreibt. Dann l i e f e r t die M u m f o r d - K o n s t r u k t i o n eine semiabelsche
Variet~t G fiber Spek(Rj) x (AV) ~ , (zun~chst ~ b e r dem f o r m a l e n
Schema, aber man m a c h t a l l e s algebraisch, da die K o n s t r u k t i o n gleich
eine Kompaktifizierung dieser Variet[t liefert), welche gute R e d u k t i o n
hat auf d e m U r b i l d des o f f e n e n S(o)-0rbits S(J) ~ S ( o )
o
fG = 71CB
Z (X u ) ,7.fx. (d l o g ( ~ 1 ) A . . . A d log(~ r a ))k
mit
f 6 (Ag-ro(t*A) )k
×
367
k
Die f liefern globale Schnitte yon ~ ~ber A u n d es g i b t
X g-r o
v
ein 7
_o , so dab fx verschwindet fur X { Xo + ~
d iog(~1)A...Ad 1og(~ r ) !)
O
mit f 6
X
Man kann statt Z auch andere Grundringe w~hlen, wie {, Z/nZ u.s.w.
Beim G r u n d r i n g { erh~it man bis auf einen Faktor (2~i) gk die klas-
sisehe F o u r i e r e n t w i c k l u n g einer Modulform, indem man etwa B+(X) v ]
identifiziert mit den h a l b g a n z e n s y m m e t r i s c h e n p o s i t i v d e f i n i e r t e n
Matrizen. Es folgt zum Beispiel, dab der Raum der M o d u l f o r m e n Hber
eine Basis besitzt, deren Elemente ganze F o u r i e r k o e f f i z i e n t e n haben,
und dab eine ~ - M o d u l f o r m genau dann ~ber ~ d e f i n i e r t ist, wenn alle
Fourierkoeffizienten in (2~i)gk-z liegen.
formen ~ber diesem Ring definiert sind. (Bis auf 2~i's ). Die F o u r i e r -
koeffizienten von 8(Zja,b) werden parametrisiert durch
X 61 B(X)* = I S2(X) Sie sind v e r s c h i e d e n y o n Null nur f~r X
2n 2 2n 2 "
vonder Form
1
{~(_m+a) ® (m+a), mEx : zg].
Bilder p , ( H 21)
Satz 6:
§ 7 ETALE GARBEN
Auf Ag,n ist die Garbe R1p,(~ I) lokal k o n s t a n t und besitzt eine
nicht a u s g e a r t e t e symplektische Form mit Werten in 91(-I) ((-I) =
Tate-Twist). Am Rand ist sie zahm verzweigt: A -A ist ein Divisor
g,n g,n
mit n o r m a l e n ~berkreuzungen. Seine i r r e d u z i b l e n K o m p o n e n t e n sind die
Strata Ag,n,o f~r ocB+(X)~ ein e i n d i m e n s i o n a l e r Kegel der Zerlegung.
(FHr n = I e n t s p r e c h e n sie sogar e i n d e u t i g den K o n j u g a t i o n s k l a s s e n
dieser o unter GL(X)) . Die O p e r a t i o n der zugeh~rigen M o n o d r o m i e
e r h ~ i t man aus der B e s c h r e i b u n g der l - T o r s i o n s p u n k t e einer d e g e n e r i e r e n -
den a b e l s c h e n Variet~t, die wir in § 3 (entsprechend [M4], 4.11) g e g e b e n
haben: Sei so ein E r z e u g e n d e s der H a l b g r u p p e <o> A B+(X). so ist eine
p o s i t i v definite symmetrische B i l i n e a r f o r m s o : X~ x X ÷~ und
371
Satz 7:
Bemerkung:
§ 8 DIE TORELLI-ABBILDUNG
Wir nehmen weiter an, dab alle irreduziblen Komponenten yon Cs geome-
trisch irreduzibel sind und alle Doppelpunkte rational 0bet dem Rest-
klassenk~rper k von R . Dies gilt immer, wenn k algebraisch ab-
geschlossen ist.
A
T(R) "~ > K e r n ( P i e (C) - - > P i c (~))
0~T~G~A~0
373
Es ist G(R) = G(R) . Man erh~It auch eine Abbildung T(K)G(R) ~Pic0(CQ) :
Es r e i c h t , die R e s t r i k t i o n T(K) ÷ P i c 0 ( C q ) anzugeben. Ein Element aus
T(K) wird gegeben durch einen Homomorphismus F ÷K* , welcher eine
~quivariante Operation yon F auf 0~ ® R K definiert. W~hle ein F-in-
C
variantes gebrochenes koh~rentes Ideal Jc0~®R K . Dies ergibt eine
-- C
koh~rente torsionsfreie Garbe auf ~ oder auch C , u n d auf d e r g e n e -
rischen Faser ein Geradenb~ndel vom Grad 0
Wir definieren nun eine Abbildung yon X in d e n Kern des obigen Mor-
phismus: FUr e 6 [ sei 0C, e der zugeh~rige lokale Ring, und fe
ein erzeugendes Element von Ie . Ein X 6 X w£rd gegeben durch ganze
Zahlen x 6Z , f~r a l l e e 6 E , so d a 6 f~r alle Z ±x p E V
= 0 .
e e÷p e
Dabei w~hlt man eine Orientierung aller e 6 [ , und die Summe geht ~ber
alle e's mit Anfangs-oder Endpunkt p , wobei das V o r z e i c h e n je n a c h
Orientierung zu w [ h l e n ist. Wir definieren dann eine symmetrische Bi-
linearform
b : XxXcK*
dutch
XeY e
b(x,y) = e~[ fe
b : X~T(K)
Es gilt nun
Satz 8:
Es sei C h a r (K) % 2
Beweis:
04c,e ~ R [ [ S ' T ] ] / ( S T - f e )
A A
und [I und P2 werden durch S bzw. T erzeugt.
dab R s c h o n ein solcher ist. Dann ist das P r o b l e m fur den Fall e i n e r
voll d e g e n e r i e r e n d e n Kurve schon in [MD] b e h a n d e l t worden, und wit
f o l g e n den d o r t i g e n A u s f ~ h r u n g e n : Wit k~nnen C,~ , G und G als
rigid-analytische Objekte ~ber K auffassen. Es ist d a n n C = ~/F ,
und G/i(X) ist r i g i d - a n a l y t i s c h e abelsche Variet~t
(i = b.c : X~G(K)) . Wit definieren zun~chst eine rigid-analytische
Abbildung ¢ : C x C~G/i(X) , via ~ : C x C~G . D a z u m ~ s s e n wir f0r
jeden endlichen Erweiterungsk6rper L von K eine A b b i l d u n g
: ~(L) × ~(L) ~ G ( L ) definieren. Da alle u n s e r e Konstruktionen inva-
riant unter Grundk~rpererweiterung sein w e r d e n , r e i c h t es, ~ auf
~(K) x ~(K) zu d e f i n i e r e n . Dazu ersetzen wir zun~chst C und ~ durch
ihre r e g u l ~ r e n semistabiien Modelle ~ber R . D i e s ~ n d e r t n i c h t s an
allen Definitionen und Behauptungen. Dann ist ~(K) = ~(R) . Seien also
Zl,Z 2 E C(R) zwei Punkte. Der D i v i s o r D = y ~ F y ( ( z I) - (z2)) ist d a n n
F-invariant auf ~ , d o c h hat die E i n s c h r ~ n k u n g von 0(D) auf die
irreduziblen Komponenten C im a l l g e m e i n e n n i c h t den G r a d Null. Dies
P
w i r d nun k o r r i g i e r t :
e) W~hle ein P0 , und definiere eine A b b i l d u n g
f : p ~ - > f(p0,p)
V --> K*
i) f(p0,P0 ) = I
i±) W e n n der k O r z e s t e Weg in G yon p nach P0 der K o m p o n e n t e als
n~chste Ecke Pl trifft, so sei
Definiere
Es ist g(pl,P3;p) = g ( p l , P 2 ; p) g ( p 2 , P 3 ; p )
i) Sei 6 6 F , entsprechend x : (x e) E X .
Dann ist
- - >
Der E x p o n e n t ist + I , w e n n die O r i e n t i e r u n g e n v o n e und plP2 Ober-
einstimmen, sonst - 1 .
grad(J(pl,P2) / Cp) =
±x e
6*(L(zl,z2) = ( ~--> fe (PI'P2))~(z1'z2)
e 6 plP2
W e n n m a n also e i n e n M o r p h i s m u s
p : F+X÷K*
+x e
definiert durch p(6) = H > f - , entsprechend einem p 6 T(K) ,
eEPlP 2 e
so k a n n m a n F via p ~quivariant operieren lassen und erh[it ein
Geradenb~ndel aus G(R) = Pic0(C) . Wenn Pl = P2 ist, so w i r d d i e s e s
GeradenbQndel d u r c h den D i v i s o r Zl-Z 2 auf C gegeben. Wenn
z 2 = 7(z I) mit y 6 F , entsprechend y = (ye) £ X , so e r h ~ I t m a n das
Geradenb~ndel zu c(y) 6 G ( R ) . W e n n m a n das o b i g e E l e m e n t aus G(R)
noch mit p 6 T(K) multipliziert, ergibt sich s c h l i e B l i c h eine Abbildung
mit
i) ~ ( Z l , Z 2) + ~ ( z 2 , z 3) = ~ ( Z l , Z 3)
ii) ~(zl,7(zl)) = i(y) (y = Bild(7) E X )
: Crl x C q ~ G q = P i c 0 (C n)
CxC~G
I i*(x)
C x C----~> G
g) Alles in a l l e m haben wir den Satz 8 bewiesen bis auf die Tatsache,
dab m a n fur ii) und iii) i durch einen Automorphismus von X ab~ndern
muB. Wir wollen zeigen, dab dieser Automorphismus die Identit~t ist.
x e : 0 <=> l(x) e = 0
Teilmenge ~ g/B(Z g )
2~i S p u r (MZ)
ZM(Z) : e
b : ~g x~zg~0
s
b(x,y) : XM ,
I
wobei M die Matrix ist mit Eintrigen mjk~ = ~ ( x iky~ + X k Y j) . b(x,x)
setzt sich fort zu e i n e r regul~ren Funktion auf S
d
b : xxzg~0 * ,
-- S
o
2ziz I 2~iz
8(e ..... e g) =
g t
= ~ ei~(mZtm+j~imjZjj)e2~im z
= (ml, .... mg) 6 ~ g
381
]]~]t2(~) = f I1~II 2
G
s
dz I dz 2
-- A . . . A
z I Zg
ANHANG:
Dann gilt:
: A ÷ A*
g,n g,n
-I
ii) %IAg,n ist eine offene Einbettung, und Ag,n = ¢ (%(Ag,n))
iil) A* -A hat K o d i m e n s i o n g in A*
g,n g,n g,n
Dies liefert globale E r z e u g t h e i t ~ber A g . Man muB nun noch den Rand
betrachten,sowie zeigen, dab die Fasern yon ¢ : A g ÷ A *g endlich ~ber
A sind.
g
383
Literatur
[M4 ] D. Mumford, An A n a l y t i c C o n s t r u c t i o n of D e g e n e r a t i n g A b e l i a n
V a r i e t i e s over Complete Rings, Comp. Math. 24 (1972),
2 3 9 - 272.
c l o s e d , it is o n l y closed inside Ag
°, the s e t of p o i n t s of Ag that
(i.e. that are not products). For g=1,2,3 the c l o s u r e of the image
the a p p r o a c h e s have been worked out more completely, while new and
this corner of n a t u r e are subtle and worthy of h i s time still very much
holds true.
386
The ingredients.
varieties of d i m e n s i o n g over ~,
functions
squares of
i
CX: X ~N N = 2g-1
+
Z ÷ ( .... @2[0] ( ~ , z ) , . . . ) = @2(~,z)
¢: Ag(2,4) ~ ~N
+
+ ( .... e 2 [ o ] ( T , 0 ) .... )= @2(s,0)
~: A (2,4) ~ ~ M M = 2g-l(2g+])-i
g
÷ ( .... e2E ~ 3(~,0) .... ).
E
and ~ are c o n n e c t e d by the special Veronese V defined by (i) :
~pN
V
Ag (2,4) ~ +
]pM
gives us a diagram
¢ t $ 6 = mm (2)
J ~ J
21
where % = ~*, - denotes transpose and ~,I are the principal polari-
= J x p / { (~,~ (~)) : ~ ~ H I}
2~ 0)
Let o: Jxp + J be the natural isogeny. Then the polarization (0 p
One finds (using that for Jacobians the theta divisor in J a c g-I con-
0 - 1 ( 0 o ) n (Jac(C) x (0)) = 0 + @
0 O,~
(~.)-i~ = 0 + e -~ . (3)
389
6: P ~ 12oji
p + (~*) -i ( ~ j ,-p) -
inclusion
~ 12%1
formula
0(u+v)0(u-v) : Z c~Bs~(u)s~(v)
gives rise to a d i a g r a m
~X ®2 v)
~ ]P(H°(X,Lx )
X ~ ~ ~ B '
~ 12°x[
where %i(x) = 0X, x + O X , _ x and B' is i n d u c e d by B.
that P can be w r i t t e n as
is a t r a n s l a t i o n of (4).
Let
by s u b s t i t u t i n g 8[~,0]
8 0 [~, 01] (Tg,0) for X ~,] The g r o u p Fg/Fg(4,8)
(4,8).
g
It is a r e c u r r i n g phenomenon in the h i s t o r y of the S c h o t t k y problem
APPROACH 2 : TRISECANTSo
following.
~: (pl),N --~ M.
We put as u s u a l
Proof. By the lemma, the rank of this 2gxn-matrix is less than n ¢=~
is g i v e n by a local homomorphism
{[c]/(smJ)
0X,%(xj)
~ N }.
V = {2t c X : t + Y c #~i(/) for some line lc
A
v
(2.5) The property of having flexes is closely related to the Kadomcev-
To get the link, note that an inclusion Spec ~[~]/(a N+I) ÷(X,0)
is given by a local h o m o m o r p h i s m
CEa]/(c N+I )
0X'0 N
i
f + [ A i (f)
i=l
A = id ,
O Ai(gh) = k+~:iAk(g)~/(h)
or formally
eJ~imj ~j
~ Ak ak (mod aN+l).
k=0
some point. If we assume that this point is the origin and that Y :
395
13+
(V) 3 = S p e c ~ [ ~ ] / ( ~ 4) ~=~ 3 m 3 such that (~m I D I D 2 + D 3 ) f =0 (all v)
~=~ rank(( 1 D1
24 DI+3
D - 3DID3)_2) (~,0) ~ 2.
D 1 ÷ aD I, D 2 ÷ a 2 D 2 + b D I, D 3 ÷ a 3 D 3 + a 2 b D 2 + c D I, a # 0,b,c, h e n c e we c a n re-
write this as
32 ÷
((D - DID3 + ~D 2 + d) ~ ) (~,0) = 0 (8)
U y y + ( U t + U x x x + U U x ) x = 0. It is s a t i s f i e d on a J a c o b i a n by
2
U = D l l O g 8 ( z + x a 1 + Y a 2 + t a 3) + c for some a l , a 2 , a 3 c c g , c ~ {, see [15 ].
acterize Jacobians :
32 ÷
((D - DID 3 + ~D 2 + d ) 82 ) (T,0) = 0. (9)
an i r r e d u c i b l e component.
rize the Jacobian locus, see [ i ]. Using the version of (2.4) given
i.e. we look at t h e h y p e r p l a n e s
[ ~ 02[a](~,0) = 0 ( x = x )T (10)
O
satisfying
i.e. define
( [ ~~f
0 (.... 0 2 [ 0 ] (~,O) .... ) @ 2 [ o ] (~,z) : f ~ Ig } .
a relation
the relation
4 4 4 22 22 22
r I + r 2 + r 3 - 2rlr 2 - 2rlr 3 - 2r2r 3 : 0
between the squares of the even thetas. Using (i) this gives an equa-
tion
¢= ~ ~F
(" "" /92[°] (T'0) .... ) @ 2 [ o ] (<,z)
O
Proof. Use (2.1) and put a=b, c=d there. One finds a relation
0 = {x e J a c ( C ) : e-x e@ } .
o
Then obviously,
F X ~_ n (e uG ). ( K: c a n o n i c a l divisor)
C~ K--C~
c Sing 0
If C is h y p e r e l l i p t i c then
1 W o
Sing 00 = g2 + g-3 '
finds
Fx = { (x-y) ~ J a c ( C ) : x,y c C } .
N
Ak ~ 2 ( T , 0 ) c k = ~2(~,0) + ~ ~ij?i~j~2(~,0).
k=0
1 4 1 2 ÷ ÷
(( ~-~DI + ~ D 2 - DID 3) @ 2 ) ( ~ , 0 ) = d ~ (~,0) + ~ eij~i~j@2(~,0) ,
+
where eij is the coefficient of 4 in ~ij" If ~ eij~i~j@2(~,0)
2+
is a multiple of DI@2(~,0), then we can change coordinates such that
zero, cf [ii]
of the moduli space we can also intersect the Kummer variety with the
~X({¼(x-y) : x , y c C }).
1
%(~(x-y)) = O([P]),
1 1 1
cf. [13],p.340, where ~(x-y) 6 Jac(C) is s u c h that 2(~(x-y)) =~(x-a) .
1
By symmetry it t h e n follows that all of ~ X ( ~ (x-y)) lies in ~ (Ag(2,4))
we conjectured that
and we proved this for g=3. Secondly, we hope that this characterizes
Jacobians :
zzg+ ~ZZ g . Using the Heat Equations the question becomes whether the
rank of
at (7,z) is m a x i m a l .
401
~x(Z).
A Jacobian is known to possess a lot of such flexes : if X =
Jac(C), then applying (2.3) with Zl:Z2:Z3= p we find that all points
1
of { ~(x-y) E X : x,y ~ C } are flex points. Hence the rank of z is
1
is not of maximal rank are those corresponding to {~(x-y) : x,y< C }
a technical assumption.
ments A,B with A= sn+ ... (... = lower order terms), B = ~m+ ... w i t h
for n >> 0.
B as above.
V : F*(ZP.) ~ F*((Z+I)P)
d 1 d
by taking ~ on C-P and z + ~ on U. A non-zero section so
n-I
as O = ~ a i(x) v l s .
i=0 o
403
This gives us a m a p
F (C-P,0 C) ~ D
n-i
a ÷ ~ ai(x) ~i.
i=0
V as a b o v e and d e f i n e
F* F*(nP)
n
F (C-P,0 C) ~ D
~: T t ~ N ~ D
+ B (t)
~t n
n
dL : Tt{ ~ Cn~ t + [[ C n B n , L ] ,
n
a finite dimensional L
RL = ¢(ker dL) •
RL = {P ~ D : [P,L] = 0}
(4.2). Moreover, it turns out that ker dL can be identified with the
405
ed s u c h a reduction.
A FINAL REMARK.
References.
Serge Lang
Department of Mathematics
Yale University
New Haven, CT 06520
U.S.A.
§ I. Nevanlinna theory
m(D,r,f) : S -logl f * s I 2 @ .
B d B (r)
r
N(D,r,f) = S [N(D,r,f)-N(D,0,f)] dr + N(D,0,f)log r .
0 r
§ 2. Weil functions
v(a) = -loglal
v
~D = v ° f +
In the sequel, metrics will not be used as such; only the associated
Weil functions and the above properties will play a role. Note that
these Well-Green functions need not be harmonic. In some cases, they
may be, for instance in the case of divisors of degree 0 on a curve.
But if the divisor has non-zero degree, then the Green function is not
harmonic.
In the sequel, we shall deal with global objects, and then the Weil
functions and others must be indexed by v , such as ID,v' ~v' etc.
§ 3. Heights (Cf.[La])
Let K be a number field, and let {v} be its set of absolute values
extending either the ordinary absolute value on ~ , or the p-adic ab-
solute values such that Ipl v = I/p . We let Kv be the completion, and
Ka its algebraic closure. Then we have the product formula
v
Z dvv(a) : 0
v
dv
where dv - [ K[K
v :
: ~v]
~] and a 6 K , a ~ 0 . We let llalIv = ialv
Let
: X --> ]pn
H 1. h = h + 0(I)
c <0
H 2. If D is e f f e c t i v e , then h D ~ -O(1)
411
h D : O(hE)
h D = E dv ID,v + O(I)
V
m(D,S) = Z d v 1 D ,v
v6S
Then
412
h D = m(D,S) + N(D,S) ,
Or in o t h e r words,
EXAMPLES
f(t) = ~ (d ~ - t )
d
I~_ ~ >__/__I
= 2+g
q
(~i) = Li-L °
TT
i v6S H(p) n + l + C
for all P outside the closed set Z(e) . This is S c h m i d t ' s theorem,
except that Schmidt arrives at t h e conclusion that the e x c e p t i o n a l set
is a f i n i t e union of h y p e r p l a n e s . In o r d e r to m a k e Vojta's conjecture
imply Schmidt strictly, one would have to r e f i n e it so as to give a
bound on the degrees of the components of the exceptional set, which
should turn out to be I if the original data is linear.
hK ~ s hE on X(K) ,
which shows that h°(mD) Sh°(mD + mE') = x(m(D + E')) for m large
because the h i g h e r cohomology groups vanish for m large, so
h°(mD)>> md
E d v ~D,v = hD + O(])
v6S
I
Ix 3 y2 I > max(Ix31 , ly21)
2 2
f : ]PI > ]P2
3 2
~(x,y) = X-y
418
Then the d i v i s o r of ~ is g i v e n by
(~) = C - 3L .
I
(I) vo<0 (x,y) = - 2~ Uh L
< of(x,y) +error term.
2 > 2
]Pl ]P2 "
BIBLIOGRAPHY
H e n r y C. W e n t e
D e p a r t m e n t of M a t h e m a t i c s
The U n i v e r s i t y of T o l e d o
T o l e d o , O h i o 4 3 6 0 6 , U. S. A.
be the first and second fundamental forms for the surface. We shall set
the m e a n curvature H = ½. The Gauss and Codazzi-Mainardi equations in
this case become (see [4] for details)
~v = -k2 ~v
Here k I = L/E = e-~sinh w, k 2 = e-~cosh ~ so w e see t h a t the lines
of curvature correspond to l i n e s parallel t o the coordinate axes in R2°
Furthermore, the surface is f r e e o f u m b i l i c points.
If x(u,v) is to b e a doubly periodic mapping then so must ~(u,v).
However the converse n e e d n o t b e true. Suppose that ~(u,v) is a p o s -
itive solution t o the d i f f e r e n t i a l equation (3) o n a r e c t a n g u l a r domain
~AB lying in t h e first quadrant with t w o of its sides on the c o o r d i n -
ate a x e s a n d t h e v e r t e x opposite the origin at (A,B). Suppose also
that the solution ~(u,v) vanishes on the boundary o f the r e c t a n g l e .
FolLowing the argument used in [3] , one can show that ~ (u,v} satisfies
the following symmetry properties.
v w-plane (A , B ) /~-plane
v
!- J ~u' >u
S(~,X)
I //~ (A(~),0,~,~)I
r / I
I s
~i ~*
iso
e2
Av + yv = 0 on ~ , v = 0 on boundary ~ (9)
1 1 ,Xu 2 3
9 > ~/2
2 3
1 2
3
/
Figure ~: Case 3. W s o m e w h a t larger, the Planes ~o' ~i s t i l l separated.
, / ,( /
References:
i. A.D. Alexandroff, Uniquenes s Theorems for Surfaces in the Large,
V. Vestnik, L e n i n g r a d Univ. No. 19 (1958) 5-8: Am. Math. Soc. Transl.
(Series 2) 21, 412-416.
2. L.P. Eisenhart, A T r e a t i s e on the D i f f e r e n t i a l Geometry of Curves
and Surfaces, Dover Reprint (1960)o
3. B. Gidas, W. Ni, L. Nirenberg, S y m m e t r y and Related P r o p e r t i e s via
the M a x i m u m P r i n c i p l e , Comm. Math. Physics 68 (1979) No. 3, 209-243.
4. H. Hopf, D i f f e r e n t i a l Geometry in the L a r g e , ( S e m i n a r Lectures New
York Univ. 1946 and Stanford Univ. 1956) Lecture Notes in M a t h e m a t i c s
No. 1000, S p r i n g e r Verlag, 1983.
5. Wu-Yi Hsiang, G e n e r a l i z e d R o t a t i o n a l H y p e r s u r f a c e s of C o n s t a n t Mean
Curvature in the E u c l i d e a n Space ~, Jour. Diff. Geometry 17(1982)337-356.
6. J.L. Moseley, O__nnA s y m t o t i c Solutions for a D i r i c h l e t P r o b l e m with an
e x p o n e n t i a l Singularity, Rep Amr I, West V i r g i n i a U n i v e r s i t y (1981)
7. V.H. Weston, On the A s y m t o t i c Solution of a Partial D i f f e r e n t i a l
E q u a t i o n w i t h an E x p o n e n t i a l Nonlinearity, SIAM J. Math Anal 9(1978)
1030-1053.
8. H. C. Wente, c o u n t e r e x a m p l e to a c o n j e c t u r e of H. Hopf, (to appear)
Pac. Jour. of Math.
THE T O P O L O G Y AND G E O M E T R Y OF THE M O D U L I SPACE
OF RIEMANN SURFACES
Scott A. W o l p e r t *
D e p a r t m e n t of M a t h e m a t i c s
U n i v e r s i t y of M a r y l a n d
C o l l e g e Park, MD 20742
i. Definitions and N o t a t i o n .
f0
(F,S) ~ (R0,P 0)
(RI,P 1 )
Briefly Ts is a c o m p l e x m a n i f o l d and is h o m e o m o r p h i c to
~46g-6+2s. Theg m a p p i n g class group Fs acts n a t u r a l l y on Ts : to
the e q u i v a l e n c e classes {h} 6 F s and g { ( R , f , P ) } 6 Ts assig~ the
g Fs g
class { (R,f o h,P)} 6 T s. The action represents as b i h o l o m o r -
phisms of Ts . g g
g
Ts is s u m m a r i z e d in the d i a g r a m
g
T l'0TSg x ( T l ' 0 ) * T S g / i - ~ i
2.3. The cut system complex CS, [ll], A cut system <Cl,...,Cg>
on F is the isotopy class of a c o l l e & t i o n of d i s j o i n t simple
closed curves Cl,...,Cg such that F - (C 1 U...U Cg) is connected.
A 6imple mov~ of cut systems is the r e p l a c e m e n t of <C.> by <C~>
1 1
where Cj = C[3 for j ~ k and Ck intersects C~ once (all
434
|>
<Ci> <C i
<
C," >
1
<C.,C.> <C.,C'.>
z 3 i ]
I I I 21
<C~,C.> <C~,C',>
l 3 i ]
Figure 1
C~ ;/
¢i
/
C~~C, Cj-
(R3)
~.__J C
Figure 2
435
¢: Fs ÷ Fs , r { 1
g,r g,r+l
~: Fs ÷ Fs r > 2 and
g,r g+l,r-l' -
Z: Es ÷ F s r a 2
g,r g+l,r-2'
7{
Figure 3
437
L--/
Figure 4
Theorem 2.2.
~,: H k ( F Sg, r) ÷ Hk s
(Fg+l ,r + l ) is an i s o m o r p h i s m
-#
h '
Figure 5
Figure 6
440
above by -3(3g-2)
4~ In fact the a r g u m e n t s show that the c u r v a t u r e s
are g o v e r n e d by the s p e c t r u m of the Laplacian: the n e g a t i v e curva-
ture is a m a n i f e s t a t i o n of the n o n p o s i t i v i t y of the Laplacian. These
last results have also been o b t a i n e d by Royden.
As a further application of the t e c h n i q u e s I wish to c o n s i d e r the
characteristic classes of the T e i c h m u l l e r curve T . T is the
^ g g
natural fibre space over T ; the fibre above R ~ T is a c o m p a c t
g g
submanifold isomorphic to R. If ~ :T ÷ T is the p r o j e c t i o n
g g
then the kernel of the d i f f e r e n t i a l d~ : TI, 0 T + TI,0T defines a
g g
line b u n d l e (v) on T , the v e r t i c a l b u n d l e of the fibration.
g
The r e s t r i c t i o n of (v) to a fibre of z is simply the tangent
bundle of the fibre; by the u n i f o r m i z a t i o n theorem the h y p e r b o l i c
metric induces a metric on (v). I have c o m p u t e d the c u r v a t u r e
2-form for this m e t r i c and found that it is negative, a pointwise
version of A r a k e l o v ' s result that the dual (v)* is
numerically effective, [4]. Once again the c u r v a t u r e is g o v e r n e d
by the s p e c t r u m of the Laplacian. By i n t e g r a t i n g powers of the
Chern class Cl(V) over the fibers of ~ one obtains Fg invariant
characteristic classes <n(P) = f -l(p) Cl(V)n+l defined on Tg.
dS~p = dZld~l
lZll2(iog1/l~ll)3+ O(iZli(logi/Izli)l31, [17].
In p a r t i c u l a r the W-P length of a d i f f e r e n t i a b l e curve in ~ is
g
finite and the metric extends to a c o m p l e t e metric on M . Starting
g
from Masur's results it follows that the K'ahler form ~ extends
WP
to a closed, positive, (i,i) current ~I4P on ~ , [26]. In fact
_ g
the above s i n g u l a r i t y is s u f f i c i e n t l y m i l d for ~WP to be w r i t t e n
as ~F, F a continuous function (as an example c o n s i d e r
1
F= log i/[zr),~
~ [28]. Standard a p p r o x i m a t i o n techniques may then be
applied to c o n c l u d e that ~WP is the limit of smooth Kahler forms
in its c o h o m o l o g y class. Thus even though the Kahler form is not
smooth it is suitable for applications in p a r t i c u l a r the Kodaira
theorem may be quoted to c o n c l u d e that there is a p r o j e c t i v e embedding.
N o w the d i s c u s s i o n will continue w i t h the results on d e s c r i b i n g the
c o h o m o l o g y class of
WP"
So E
Figure 7
The 2-cycle E is p a r a m e t r i z e d by 1
M I. Now to state the d e s i r e d
formal property of ~WP let s be the W-P K~hler form for Ms .
g g
Then briefly 0
~glE = ~ l1l M--;
~ the r e s t r i c t i o n of __~0 g to E is
1 -- : 71
272 ~WP
41 Z2
t t~ = ~ e +e sin @ sin @ (3.2)
P q
Y (P'q) E~#Yx~#7 2(e ~-i)
mI m2
_[ e +e
sin ~ sin 8 .
r s
(r,s)(~#Bx~#Y 2 (e~_l)
23
Figure 8
447
Figure 9
Figure i0
448
In p a r t i c u l a r T is a cell. F u r t h e r m o r e the D e l i g n e - M u m f o r d
_ g
compactification M can be c o n s t r u c t e d from M by simply allowing
g g
the length p a r a m e t e r s Z, to vanish, [1,26]. The d i s c u s s i o n of the
p r e v i o u s section already suggests a r e l a t i o n s h i p between the Kahler
form WWp and the F-N coordinates. Recall that ~WP is i n v a r i a n t
under the F-N vector fields and observe that the coordinate vector
fields ~ are indeed F-N vector fields. C o n s e q u e n t l y the coeffi-
3
cients of ~WP in F-N coordinates (Tj,Zj) must be i n d e p e n d e n t of
the twist variables T.. In fact much more is true, [26].
3
T h e o r e m 3.3. ~WP = [
j d Z .3 A d T .3
.
Figure ii
449
Figure 12
REFERENCES
24 S. A. Wolpert, On the s y m p l e c t i c g e o m e t r y of d e f o r m a t i o n s of a
h y p e r b o l i c surface, Ann. of Math., 117 (1983), 207-234.
25 S. A. Wolpert, On the K a h l e r form of the m o d u l i space of o n c e
p u n c t u r e d tori, Comment. Math. Helv., 58 (1983), 246-256.
26 S. A. W o l p e r t , On the W e i l - P e t e r s s o n g e o m e t r y of the m o d u l i
space of curves, Amer. J. Math., to appear.
27 S. A. W o l p e r t , On the h o m o l o g y of the m o d u l i space of stable
curves, Ann. of Math., 118 (1983), 491-523.
28 S. A. W o l p e r t , On o b t a i n i n g a p o s i t i v e line b u n d l e from the W e i l -
P e t e r s s o n class, Amer. J. Math., to appear.
29 S. A. W o l p e r t , C h e r n forms and the R i e m a n n t e n s o r for the m o d u l i
space of curves, p r e p r i n t .
30 S. A. W o l p e r t , Geodesic length functions and the N i e l s e n problem,
preprint.
31 A. Borel, S t a b l e real c o h o m o l o g y of a r i t h m e t i c groups, Ann. Sci.
Ecole Norm. Sup. 4e(7) (1974), 235-272.
Lecture Notes in Mathematics Vol. 1111
ISBN 978-3-540-15195-1
c Springer-Verlag Berlin Heidelberg 2008
F. Hirzebruch
Arbeitstagungen
Addendum
Lecture Notes Vol. 1111 (Springer) includes the programmes of all 25
Arbeitstagungen held between 1957 and 1984. The archives for AT 57 and
AT 58 were very incomplete. However N. Kuiper had made a note for himself
of the topics of the talks and communicated these to me in a letter of June
22nd 1991.
Arbeitstagung 1957
Arbeitstagung 1958
Programme
der
I. M a t h e m ~ c h e ~be~%ta~!u~~ 1957
Sonntag, 14.7.:
10.15 Uhr Ausflug mit D a m p f e r f a h r t
Montag, 15.~±:
9.30 Uhr A. G r o t h e n d i e c k
15.15 Uhr M.F. Atiyah
17.15 Uhr N. Kuiper
Dienstag, 16.7.:
11.30 Uhr A. Grothendieck
15.15 Uhr F. H i r z e b r u c h
Mittwoch, 17.7.:
11.30 Uhr A. Grothendieck
15.15 Uhr M.F. At iyah
S. A b h y a n k a r
R. Bott
H. Grauert
A. Grothendieck
M. Kervaire
J. Milnor
D. Puppe
R. Remmert
J.-P. Serre
K. Stein
R. Thom
456
A. A n d r e o t t i : Vanishing theorems
M. Kneser: Approximation in a l g e b r a i c g r o u p s
M. Kneser: S i e g e l ' s t h e o r e m and T a m a g a w a n u m b e r s
A. G r o t h e n d i e c k : Schemes of m o d u l i
A. Haefliger: Links
R. Abraham: Transversality of m a p p i n g s
H. Grauert: Rigid singularities
J. Eells -
Harmonic maps
J.H. Sampson:
S. Lang: T r a n s c e n d e n t a l numbers
F. Oort: On a u t o m o r p h i s m s of v a r i e t i e s
H. Grauert: On S u p e r - C o c y c l e s
S. Lang: On Manin's theorem
A. Borel: I n t r o d u c t i o n to a u t o m o r p h i c forms
E. Thomas: E n u m e r a t i o n of v e c t o r b u n d l e s
J.F. Adams: Some a p p l i c a t i o n s of K - t h e o r y to h o m o t o p y t h e o r y
D. Anderson: S e v e r a l a s p e c t s of K - t h e o r y
R. Wood: Pre-Palais
R. Palais: Index t h e o r e m for e l l i p t i c b o u n d a r y p r o b l e m s
A. Borel: Pseudo-concavity for a r i t h m e t i c groups
A. Douady: An hour of c o u n t e r e x a m p l e s
M.F. Atiyah: Elliptic operators on m a n i f o l d s and g e n e r a l i z e d
Lefschetz fixed point theorem II
S. Lang: Division points on curves
A. Grothendieck: Riemann-Roch I
J.P. Serre: Formal groups
A. Grothendieck: Riemann-Roch II
J. Milnor: Projective class groups in t o p o l o g y
N. Kuiper: Piecewise linear m i c r o b u n d l e s are bundles
v v
I. Safarevic: Algebraic analogue of u n i f o r m i s a t i o n
D.V. Anosov: Asymptotic theory of some partial differential
equations
M. Karoubi: Real and complex K - t h e o r y
S. H i l d e b r a n d t : Boundary values in c a p i l l a r i t i e s
D. B u r g h e l e a : H o m o t o p y g r o u p s of spaces of d i f f e o m o r p h i s m s
T. t o m Dieck: B o r d i s m of c o m m u t i n g involutions
R. Gardner: R i g i d i t y and u n i q u e n e s s of c o n v e x h y p e r s u r f a c e s
U. Koschorke: Pseudo-kompakte Teilmengen unendlich dimensionaler
Mannigfaltigkeiten
467
M. Ar t i n : L~roth's theorem
E. Bombier i : On p l u r i c a n o n i c a l models of a l g e b r a i c surfaces
A.I. Kostr ikin : Deformation of Lie algebras
J. Eells: Introduction to s t o c h a s t i c R i e m a n n i a n g e o m e t r y
M.F. Atiyah: A s y m p t o t i c p r o p e r t i e s of e i g e n v a l u e s in R i e m a n n i a n
geometry
C.T.C. Wall: N o r m s of units in g r o u p rings
S. Lang: F e r m a t curves and units in the m o d u l a r function
field
470
M. Berger: Wiedersehensmannigfaltigkeiten
( C o n j e c t u r e of B l a s c h k e )
A. V a n de Ven: Inequalities for C h e r n numbers of surfaces
G. Zuckerman: Representations of semi simple lie g r o u p s
D. Vogan: Size of r e p r e s e n t a t i o n s
L. Berard-Bergery: A n e w example of E i n s t e i n manifolds
M.F. Atiyah: V e c t o r B u n d l e s on R i e m a n n S u r f a c e s
A. Borel : L 2 - c o h o m o l o g y of a r i t h m e t i c g r o u p s
I. Bake iman" Topological methods in the t h e o r y of
Monge-Amp@re equations
H. King : T o p o l o g y of real a l g e b r a i c v a r i e t i e s
D. Quillen: Determinants of Y - o p e r a t o r s
J. Coates: Heights on elliptic curves
D. V o g a n : Representations with cohomology
J. B e r n s t e i n : Beilinson-Bernstein construction