You are on page 1of 18

arXiv:math/0507215v3 [math.

NT] 24 Jul 2007

INVARIANT METRICS AND LAPLACIANS ON SIEGEL-JACOBI


SPACE
JAE-HYUN YANG

Abstract. In this paper, we compute Riemannian metrics on the Siegel-Jacobi space


which are invariant under the natural action of the Jacobi group explicitly and also provide
the Laplacians of these invariant metrics. These are expressed in terms of the trace form.

1. Introduction
For a given fixed positive integer n, we let
Hn = { Z C(n,n) | Z = t Z,

Im Z > 0 }

be the Siegel upper half plane of degree n and let


Sp(n, R) = {M R(2n,2n) | tM Jn M = Jn }
be the symplectic group of degree n, where


0
En
Jn =
.
En 0
We see that Sp(n, R) acts on Hn transitively by
M Z = (AZ + B)(CZ + D)1 ,

(1.1)



A B
where M =
Sp(n, R) and Z Hn .
C D
For two positive integers n and m, we consider the Heisenberg group
(n,m)

HR

= { (, ; ) | , R(m,n) , R(m,m) , + t symmetric }

endowed with the following multiplication law


(, ; ) ( , ; ) = ( + , + ; + + t t ).
(n,m)

We define the semidirect product of Sp(n, R) and HR

(n,m)

GJ := Sp(n, R) HR

endowed with the following multiplication law



 



+ ,
t
M, (, ; ) M , ( , ; ) = M M , (
+ ; + +
t )
Key words and phrases. invariant metrics, Siegel-Jacobi space, spectral theory.
2000 Mathematics Subject Classification: Primary 32F45, 32M10.
This work was supported by Inha University Research Grant .
1

JAE-HYUN YANG

(n,m)

and (,
) = (, )M . We call this
with M, M Sp(n, R), (, ; ), ( , ; ) HR
J
group G the Jacobi group of degree n and index m. We have the natural action of GJ on
Hn C(m,n) defined by




(1.2)
M, (, ; ) (Z, W ) = M Z, (W + Z + )(CZ + D)1 ,


A B
(n,m)
where M =
and (Z, W ) Hn C(m,n) . The
Sp(n, R), (, ; ) HR
C D
homogeneous space Hn C(m,n) is called the Siegel-Jacobi space of degree n and index m.
We refer to [2-3], [6-7], [11], [14-21] for more details on materials related to the Siegel-Jacobi
space.

For brevity, we write Hn,m := Hn C(m,n) . For a coordinate (Z, W ) Hn,m with Z =
(z ) Hn and W = (wkl ) C(m,n) , we put
Z =
W =

X + iY,
U + iV,

X = (x ),
U = (ukl ),

Y = (y ) real,
V = (vkl ) real,

dZ =

(dz ),

dZ = (dz ),

dY = (dy ),

dW =

(dwkl ),

dW = (dw kl ),

dV = (dvkl ),

=
Z

=
X

w11

= ...
W

w1n

u11

1+
2
z
1+

2
x

...
..
.
...


wm1

..
.

,
,

wmn

um1

=
=

1+
2
z

1+
2
y

w11

= ...
W

w1n

v11

...

.. ,
..
= ...
.
.
V

u1n . . . umn
v1n
where ij denotes the Kronecker delta symbol.

= ...
U

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

,
,

w m1

..
.

w mn

vm1

... ,

vmn

C. L. Siegel [12] introduced the symplectic metric ds2n on Hn invariant under the action
(1.1) of Sp(n, R) given by


(1.3)
ds2n = Y 1 dZ Y 1 dZ
and H. Maass [8] proved that the differential operator

 


t
n = 4 Y
Y
(1.4)
Z Z

is the Laplacian of Hn for the symplectic metric ds2n . Here (A) denotes the trace of a
square matrix A.
In this paper, for arbitrary positive integers n and m, we express the GJ -invariant metrics
on Hn C(m,n) and their Laplacians explicitly.

INVARIANT METRICS AND LAPLACIANS ON SIEGEL-JACOBI SPACE

In fact, we prove the following theorems.


Theorem 1.1. For any two positive real numbers A and B, the following metric


ds2n,m;A,B = A Y 1 dZ Y 1 dZ
 



+ B Y 1 t V V Y 1 dZ Y 1 dZ + Y 1 t (dW ) dW





1
1 t
1
1 t
(dW )
V Y dZ Y
(dW ) V Y dZ Y

is a Riemannian metric on Hn,m which is invariant under the action (1.2) of the Jacobi
group GJ .
Theorem 1.2. For any two positive real numbers A and B, the Laplacian n,m;A,B of
(Hn,m , ds2n,m;A,B ) is given by
 








t
1 t
t
Y
+ VY
V
Y
Y
n,m;A,B =
Z Z
W W









t
t
t
+ V
Y
+ V
Y
Z W
W Z


t
4

.
+ Y
B
W W
The following differential form
4
A

dv = ( det Y )(n+m+1) [dX] [dY ] [dU ] [dV ]


is a GJ -invariant volume element on Hn,m , where
[dX] = dx ,

[dY ] = dy ,

The point is that the invariant metric


in terms of the trace form.

[dU ] = k,l dukl

ds2n,m;A,B

and

[dV ] = k,l dvkl .

and its Laplacian n,m;A,B are expressed

For the case n = m = 1 and A = B = 1, Berndt proved in [1] (cf. [19]) that the metric
ds21,1 on H C defined by
ds21,1 := ds1,1;1,1 =

y + v2
y3

( dx2 + dy 2 ) +

1
y

( du2 + dv 2 )

( dx du + dy dv )
2v
y2
is a Riemannian metric on H C invariant under the action (1.2) of the Jacobi group and
its Laplacian 1,1 is given by
 2

 2


2
1,1 := 1,1;1,1 = y 2 x
+ ( y + v 2 ) u
2 + y 2
2 + v 2
 2


2
+ 2 y v xu
+ yv
.

It is a pleasure to thank Eberhard Freitag for his helpful advice and letting me know the
paper [8] of H. Maass.

JAE-HYUN YANG

Notations: We denote by R and C the field of real numbers, and the field of complex
numbers respectively. The symbol := means that the expression on the right is the
definition of that on the left. For two positive integers k and l, F (k,l) denotes the set of
all k l matrices with entries in a commutative ring F . For a square matrix A F (k,k)
of degree k, (A) denotes the trace of A. For any M F (k,l) , tM denotes the transpose
matrix of M . En denotes the identity matrix of degree n. For A F (k,l) and B F (k,k) ,
we set B[A] = t ABA. For a complex matrix A, A denotes the complex conjugate of A. For
A C(k,l) and B C(k,k), we use the abbreviation B{A} = t ABA.
2. Proof of Theorem 1.1


A B
J
Sp(n, R) and (Z, W )
Let g = (M, (, ; )) be an element of G with M =
C D
Hn,m with Z Hn and W C(m,n) . If we put (Z , W ) := g (Z, W ), then we have
Z = M Z = (AZ + B)(CZ + D)1 ,
W = (W + Z + )(CZ + D)1 .
Thus we obtain
(2.1)

dZ = dZ[(CZ + D)1 ] = t(CZ + D)1 dZ(CZ + D)1

and
(2.2)

dW = dW (CZ + D)1 + { (W + Z + )(CZ + D)1 C}dZ(CZ + D)1 .

Here we used the following facts that


d(CZ + D)1 = (CZ + D)1 C dZ(CZ + D)1
and that (CZ + D)1 C is symmetric.
We put
Z = X + iY ,

W = U + iV ,

X , Y , U , V real.

From [9], p.33 or [13], p.128, we know that


(2.3)

Y = Y {(CZ + D)1 } = t(CZ + D)1 Y (CZ + D)1 .

First of all, we recall that the following matrices




En b
t(b) =
, b = t b real,
0 En

t
h 0
, h GL(n, R),
g0 (h) =
0 h1


0 En
Jn =
En
0
generate the symplectic group Sp(n, R) (cf. [4], [5]). Therefore the following elements

INVARIANT METRICS AND LAPLACIANS ON SIEGEL-JACOBI SPACE

t(b; , , ), g(h) and n of GJ defined by





En b
(n,m)
t(b; , , ) =
,
, (, ; ) , b = t b real, (, ; ) HR
0 En


t
h 0
, (0, 0; 0) , h GL(n, R),
g(h) =
0 h1



0 En
n =
, (0, 0; 0)
En
0
generate the Jacobi group GJ . So it suffices to prove the invariance of the metric ds2n,m;A,B
under the action of the generators t(b; , , ), g(h) and n . For brevity, we write


(a) = Y 1 dZ Y 1 dZ ,


(b) = Y 1t V V Y 1 dZ Y 1 dZ ,


(c) = Y 1t (dW )dW ,


(d) = V Y 1 dZ Y 1 t (dW ) + V Y 1 dZ Y 1 t (dW )

and



(a) = Y1 dZ Y1 dZ ,


(b) = Y1 t V V Y1 dZ Y1 dZ ,


(c) = Y1 t (dW ) dW ,


(d) = V Y1 dZ Y1 t (dW ) + V Y1 dZ Y1 t (dW )
(n,m)

Case I. g = t(b; , , ) with b = t b real and (, ; ) HR

In this case, we have


Z = Z + b,

Y = Y,

W = W + Z + ,

V = V + Y

and
dZ = dZ,

dW = dW + dZ.

Therefore




(a) = Y1 dZ Y1 dZ = Y 1 dZ Y 1 dZ = (a),




(b) = Y 1 t V V Y 1 dZ Y 1 dZ + Y 1 t V dZ Y 1 dZ




+ t V Y 1 dZ Y 1 dZ + t dZ Y 1 dZ ,




(c) = Y 1 t (dW ) dW + Y 1 t (dW ) dZ




+ Y 1 dZ t dW + Y 1 dZ t dZ

JAE-HYUN YANG

and




(d) = V Y 1 dZ Y 1 t (dW ) dZ Y 1 t (dW )




V Y 1 dZ Y 1 dZ t dZ Y 1 dZ t




V Y 1 dZ Y 1 t (dW ) dZ Y 1 t (dW )




V Y 1 dZ Y 1 dZ t dZ Y 1 dZ t .

Thus we see that

(a) = (a)

and

(b) + (c) + (d) = (b) + (c) + (d) .

Hence
n
o
ds2n,m;A,B = A (a) + B (b) + (c) + (d)

is invariant under the action of t(B; , , ).


Case II. g = g(h) with h GL(n, R).
In this case, we have
Z = t h Z h,

Y = t h Y h,

W = W h,

V = V h

and
dZ = t h dZ h, dW = dW h.
Therefore by an easy computation, we see that each of (a), (b), (c) and (d) is invariant
under the action of all g(h) with h GL(n, R). Hence the metric ds2n,m;A,B is invariant
under the action of all g(h) with h GL(n, R).



0 En
Case III. g = n =
, (0, 0; 0) .
En
0
In this case, we have
Z = Z 1

(2.4)

and

W = W Z 1 .

We set
1 := Re Z 1
and
2 := Im Z 1 .
Then 1 and 2 are symmetric matrices and we have
(2.5)

Y = 2

and

V := Im W = V 1 + U 2 .

It is easy to see that


(2.6)

Y = Z2 Z = Z2 Z,

(2.7)

1 Y + 2 X = 0

and
(2.8)

1 X 2 Y = En .

According to (2.6) and (2.7), we obtain


(2.9)

X = (2 )1 1 Y

and

Y 1 = 1 (2 )1 1 2 .

INVARIANT METRICS AND LAPLACIANS ON SIEGEL-JACOBI SPACE

From (2.1) and (2.2), we have


dZ = Z 1 dZZ 1

(2.10)
and
(2.11)



dW = dW Z 1 W Z 1 dZZ 1 = dW W Z 1 dZ Z 1 .

Therefore we have, according to (2.6) and (2.10),




1
1
(a) = (2 )1 Z 1 dZ Z 1 (2 )1 Z dZ Z


= Y 1 dZ Y 1 dZ = (a).
According to (2.5)-(2.10), we have


(b) = (2 ) (1 V + 2 U )(V 1 + U 2 )(2 ) Z dZ Y dZ Z




1
= { t U (2 )1 1 t V }{U V 1 (2 )1 }Z 1 dZ Y 1 dZ Z

= { t W + (iEn (2 )1 1 ) t V }{W V (iEn + 1 (2 )1 )}

1
1
1
Z dZ Y dZ Z
,




1
1
1
1 t
1
1 t
Z
(dW ) Z dZZ
W dW Z W Z dZ Z

(c) = (2 )

1
1
1
= (2 )1 Z 1 t (dW ) dW Z (2 )1 Z 1 t (dW )W Z dZ Z
(2 )1 Z 1 dZZ 1 t W dW Z


1
1
+(2 )1 Z 1 dZZ 1 t W W Z dZ Z




1
= Y 1 t (dW ) dW Y 1 t (dW )W Z dZ




1
Y 1 dZ Z 1 t W dW + Y 1 dZ Z 1 t W W Z dZ

JAE-HYUN YANG

and


1 t

1 t

o

Z
(dW ) Z dZ Z
W
(d) = (V 1 + U 2 )(2 ) Z dZZ (2 )


 1 t

1
1 1
1
1
1 t
(V 1 + U 2 )(2 ) Z dZ Z (2 )
Z
(dW ) Z dZZ
W


1 1
1 t
= (V 1 + U 2 )(2 ) Z dZ Y
(dW )


1 t
1 1
1
W
+ (V 1 + U 2 )(2 ) Z dZ Y dZ Z


1
(V 1 + U 2 )(2 )1 Z dZ Y 1 t (dW )


1
1 t
1 1
W .
+ (V 1 + U 2 )(2 ) Z dZ Y dZZ
Taking the (dZ, dW )-part (Z, W ) in (b) + (c) + (d) , we have




(Z, W ) = V Y 1 dZ Y 1 t (dW ) + Y 1 dZ ( t W Z 1 t W ) dW


because W = W Z 1 (cf. (2.4)).
= V Y 1 dZ Y 1t (dW )

Similiarly, if we take the (dZ, dW )-part (Z, W ) in (b) + (c) + (d) , we have




1
(Z, W ) = V Y 1 dZ Y 1 t (dW ) + dZ Y 1 t (dW )(W W Z )


because W = W Z 1 .
= V Y 1 dZ Y 1t (dW )

If we take the (dW, dW )-part (W, W ) in (b) + (c) + (d) , we have




(W, W ) = Y 1 t (dW ) dW .

Finally, if we take the (dZ, dZ)-part (Z, Z) in (b) + (c) + (d) , we have

t


(Z, Z) =
W + (iEn (2 )1 1 ) t V
W V (iEn + 1 (2 )1 )

1
1
1
Z dZ Y dZ Z


1
+ Z 1 t W W Z dZ Y 1 dZ


1
1 1
1
t
+ W (V 1 + U 2 )(2 ) Z dZ Y dZ Z


t
1 1
1
1
+ W (V 1 + U 2 )(2 ) Z dZ Y dZZ
.
Since

(V 1 + U 2 )(2 )1 = U + V 1 (2 )1
= W + V {iEn + 1 (2 )1 }
= W V {iEn 1 (2 )1 },

INVARIANT METRICS AND LAPLACIANS ON SIEGEL-JACOBI SPACE

we have


1 



iEn (2 )1 1 t V W V (iEn + 1 (2 )1 )

1
1
Z dZ Y dZ


t
1 
1
1
1
Z
iEn (2 ) 1 V W Z dZ Y dZ




1 
iEn (2 )1 1 t V V iEn + 1 (2 )1
= Z

1
1
Z dZ Y dZ .

(Z, Z) = Z

By the way, according to (2.9), we obtain


o
o
n
n
1
= (1 i2 ) iEn (2 )1 1
iEn (2 )1 1
Z
= 2 1 (2 )1 1 = Y 1

and

Therefore

o
n
iEn + 1 (2 )1 Z 1 = 1 (2 )1 1 2 = Y 1 .

Hence (a) = (a) and



(Z, Z) = Y 1 t V V Y 1 dZ Y 1 dZ .

(b) + (c) + (d) =


=

(Z, W ) + (Z, W ) + (W, W ) + (Z, Z)


(b) + (c) + (d).

This implies that the metric


n
o
ds2n,m;A,B = A (a) + B (b) + (c) + (d)

is invariant under the action (1.2) of n .


Consequently ds2n,m;A,B is invariant under the action (1.2) of the Jacobi group GJ . In
particular, for (Z, W ) = (iEn , 0), we have




ds2n,m;A,B = A dZ dZ + B t (dW )dW
( n
)
X
X
2
2
= A
(dx2 + dy
)+2
(dx2 + dy
)
=1

+B

1<n

(du2kl
1km, 1ln

2
dvkl
)

which is clearly positive definite. Since GJ acts on Hn,m transitively, ds2n,m;A,B is positive
definite everywhere in Hn,m . This completes the proof of Theorem 1.1.

Remark 2.1. The scalar curvature of the Siegel-Jacobi space (Hn,m , ds2n,m;A,B ) is constant
because of the transitive group action of GJ on Hn,m . In the special case n = m = 1 and

10

JAE-HYUN YANG

A = B = 1, by a direct computation, we see that the scalar curvature of (H1,1 , ds21,1;1,1 ) is


3.

3. Proof of Theorem 1.2





A B
If (Z , W ) = g (Z, W ) with g =
, (, ; ) GJ , we can see easily that
C D



t
= (CZ + D) (CZ + D)
(3.1)
Z
Z

  
+ (CZ + D) t C t W + C t D t t
W

and

(3.2)

= (CZ + D)
.
W
W

For brevity, we put






() := 4 Y
Y
,
Z Z




t
,
() := 4 Y
W
W





1 t
t
() := 4 V Y
V
,
Y
W W

 


t
() := 4 V
Y
Z W


and

() := 4
We also set





V
Y
.
W Z
t




,
() := 4 Y Y
Z Z



t
,
() := 4 Y
W
W





1 t
t
,
() := 4 V Y V Y
W W





t
() := 4 V Y
Z W


and

() := 4

We need the following lemma for the proof of Theorem 1.2. H. Maass [8] observed the
following useful fact.

INVARIANT METRICS AND LAPLACIANS ON SIEGEL-JACOBI SPACE

11

Lemma 3.1. (a) Let A be an n k matrix and let B be a k n matrix. Assume that the
entries of A commute with the entries of B. Then (AB) = (BA).
(b) Let A be an m n matrix and B an n l matrix. Assume that the entries of A commute
with the entries of B. Then t (AB) = t B t A.
(c) Let A, B and C be a k l, an n m and an m l matrix respectively. Assume that the
entries of A commute with the entries of B. Then
t

(A t (BC)) = B t (A t C).

Proof. The proof follows immediately from the direct computation.

Now we are ready to prove Theorem 1.2. First of all, we shall prove that n,m;A,B is
invariant under the action of the generators t(b; , , ), g(h) and n .



En b
, (, ; ) with b = t b real.
Case I. g = t(b; , , ) =
0 En
In this case, we have
Y = Y,
V = V + Y
and





t t t
=

=
.
and
Z
Z
W
W
W
Using Lemma 3.1, we obtain





t
() = () Y
Y
Z W









t t
t t
Y Y
+ Y Y
W Z
W W


t

= (),
() = Y
W
W





t
t
() = () + V Y
W W









t t
t t
+ V Y
+ Y Y
,
W W
W W





() = () + Y t Y
Z W









t t
t t
+ Y Y
V Y
W W
W W
and
()




= () + Y Y
W Z









t
t
t t
V Y
Y Y
.
W W
W W


Thus () = () and
() + () + () + () = () + () + () + () .

12

JAE-HYUN YANG

Hence
o
4 n
4
() + () + () + ()
() +
B
A
is invariant under the action of all t(b; , , ).


t
h 0
, (0, 0; 0) with h GL(n, R).
Case II. g = g(h) =
0 h1
In this case, we have
Y = t h Y h V = V h
and



= h1 t h1
= h1
.
,
Z
Z
W
W
According to Lemma 3.1, we see that each of (), (), (), () and () is invariant under
the action of all g(h) with h GL(n, R). Therefore n,m;A,B is invariant under the action
of all g(h) with h GL(n, R).



0 En
Case III. g = n =
, (0, 0; 0) .
En
0
In this case, we have
n,m;A,B =

Z = Z 1

and

W = W Z 1 .

We set
1 := Re Z 1
and
2 := Im Z 1 .
Then we obtain the relations (2.5)-(2.9). From (2.6), we have the relation
2 Z = Z 1 Y.

(3.3)
It follows from the relation (2.3) that
(3.4)

Y = Z

Y Z 1 = Z 1 Y Z

= 2 .

From (2.9), we obtain


(3.5)

1 21 1 = Y 1 2 .

According to (3.1) and (3.2), we have








t
t t
t
(3.6)
=Z Z
+Z
W
Z
Z
W
and

=Z
.
W
W
From (2.6), (3.3) and Lemma 3.1, we obtain









t
t
i V 2 Z Y
() = () U 2 Z Y
Z W
Z W









t
t
t
t
Z2 U Y
+ i Z2 V Y
W Z
W Z









+ W 2 t V t Y
.
W 2 t U t Y
W W
W W

(3.7)

INVARIANT METRICS AND LAPLACIANS ON SIEGEL-JACOBI SPACE

13

From the relation (3.4), we see () = (). According to (3.3), (3.5) and Lemma 3.1, we
otain





t
t
t
t
t
.
() = () + (V 2 V V 1 U U 1 V U 2 U ) Y
W W
Using the relation (3.3) and Lemma 3.1, we finally obtain









() = V 1 Z t Y
+ U 2 Z t Y
Z W
Z W









t
t
t
t
+ U 2 W Y
+ V 1 W Y
W W
W W
and

() =








t
t
Z1 V Y
+ Z2 U Y
W Z
W Z









t
t
t
t
+ W 2 U Y
.
+ W 1 V Y
W W
W W


Using the fact Z 1 = 1 + i2 , we can show that

() + () + () + () = () + () + () + () .
Hence
o
4
4 n
() + () + () + ()
() +
B
A
is invariant under the action of n .
Consequently n,m;A,B is invariant under the action (1.2) of GJ . In particular, for
(Z, W ) = (iEn , 0), the differential operator n,m;A,B coincides with the Laplacian for the
metric ds2n,m;A,B . It follows from the invariance of n,m;A,B under the action (1.2) and the
transitivity of the action of GJ on Hn,m that n,m;A,B is the Laplacian of (Hn,m , ds2n,m;A,B ).
The invariance of the differential form dv follows from the fact that the following differential
form
( det Y )(n+1) [dX] [dY ]
is invariant under the action (1.1) of Sp(n, R) (cf. [13], p. 130).

n,m;A,B =

4. Remark on Spectral Theory of n,m;A,B on Siegel-Jacobi Space


Before we describe a fundamental domain for the Siegel-Jacobi space, we review the
Siegels fundamental domain for the Siegel upper half plane.
We let

n
o
Pn = Y R(n,n) | Y = t Y > 0

be an open cone in Rn(n+1)/2 . The general linear group GL(n, R) acts on Pn transitively by
h Y := h Y t h,

h GL(n, R), Y Pn .

Thus Pn is a symmetric space diffeomorphic to GL(n, R)/O(n). We let



n
o

GL(n, Z) = h GL(n, R) h is integral

14

JAE-HYUN YANG

be the discrete subgroup of GL(n, R).


The fundamental domain Rn for GL(n, Z)\Pn which was found by H. Minkowski [10] is
defined as a subset of Pn consisting of Y = (yij ) Pn satisfying the following conditions
(M.1)-(M.2) (cf. [9] p. 123):
(M.1) aY t a ykk for every a = (ai ) Zn in which ak , , an are relatively prime for
k = 1, 2, , n.
(M.2) yk,k+1 0 for k = 1, , n 1.
We say that a point of Rn is Minkowski reduced or simply M-reduced.
Siegel [12] determined a fundamental domain Fn for n \Hn , where n = Sp(n, Z) is the
Siegel modular group of degree n. We say that = X + iY Hn with X, Y real is Siegel
reduced or S-reduced if it has the following three properties:
(S.1) det(Im ( )) det(Im ())
for all n ;
(S.2) Y = Im is M-reduced, that is, Y Rn ;
(S.3) |xij | 12 for 1 i, j n, where X = (xij ).
Fn is defined as the set of all Siegel reduced points in Hn . Using the highest point method,
Siegel [12] proved the following (F1)-(F3) (cf. [9], p. 169):
(F1) n Fn = Hn , i.e., Hn = n Fn .
(F2) Fn is closed in Hn .
(F3) Fn is connected and the boundary of Fn consists of a finite number of hyperplanes.
The metric ds2n given by (1.3) induces a metric ds2Fn on Fn . Siegel [12] computed the
volume of Fn
n
Y
k (k)(2k),
vol (Fn ) = 2
k=1

where (s) denotes the Gamma function and (s) denotes the Riemann zeta function. For
instance,
vol (F1 ) =

,
3

vol (F2 ) =

3
,
270

vol (F3 ) =

6
,
127575

vol (F4 ) =

10
.
200930625

Let fkl (1 k m, 1 l n) be the m n matrix with entry 1 where the k-th row and
the l-th column meet, and all other entries 0. For an element Hn , we set for brevity
hkl () := fkl ,

1 k m, 1 l n.

For each Fn , we define a subset P of C(m,n) by

n
m X
n
m X

X
X

kl hkl () 0 kl , kl 1 .
kl fkl +
P =

k=1 j=1

k=1 j=1

For each Fn , we define the subset D of Hn C(m,n) by


o
n
D := (, Z) Hn C(m,n) | Z P .
We define

Fn,m := Fn D .

INVARIANT METRICS AND LAPLACIANS ON SIEGEL-JACOBI SPACE

Theorem 4.1. Let

15

(n,m)

n,m := Sp(n, Z) HZ
J
be the discrete subgroup of G , where
n
o
(n,m)
(n,m)
HZ
= (, ; ) HR
| , , are integral .
Then Fn,m is a fundamental domain for n,m \Hn,m .

Proof. The proof can be found in [20].

In the case n = m = 1, R. Berndt [2] introduced the notion of Maass-Jacobi forms. Now
we generalize this notion to the general case.
Definition 4.1. For brevity, we set n,m := n,m;1,1 (cf. Theorem 1.2). Let
(n,m)

n,m := Sp(n, Z) HZ

be the discrete subgroup of GJ , where


o
n
(n,m)
(n,m)
| , , are integral .
= (, ; ) HR
HZ

A smooth function f : Hn,m C is called a Maass-Jacobi form on Hn,m if f satisfies the


following conditions (MJ1)-(MJ3) :
(MJ1) f is invariant under n,m .
(MJ2) f is an eigenfunction of the Laplacian n,m.
(MJ3) f has a polynomial growth, that is, there exist a constant C > 0 and a positive
integer N such that
|f (X + iY, Z)| C |p(Y )|N

as det Y ,

where p(Y ) is a polynomial in Y = (yij ).


It is natural to propose the following problems.
Problem A : Construct Maass-Jacobi forms.
Problem B : Find all the eigenfunctions of n,m .
We consider the simple case n = m = 1. A metric ds21,1 on H1 C given by
ds21,1 =

1
y + v2
( dx2 + dy 2 ) + ( du2 + dv 2 )
y3
y
2v
2 ( dx du + dy dv )
y

is a GJ -invariant K
ahler metric on H1 C. Its Laplacian 1,1 is given by

 2
2

2
+
1,1 = y
x2
y 2
 2


2
2
+
+ (y + v )
u2
v 2


2
2
+ 2yv
+
.
xu
yv

16

JAE-HYUN YANG

We provide some examples of eigenfunctions of 1,1 .


1

(1) h(x, y) = y 2 Ks 1 (2|a|y) e2iax (s C, a 6= 0 ) with eigenvalue s(s 1). Here


2
Z
o
n z
1
exp (t + t1 ) ts1 dt,
Ks (z) :=
2 0
2

where Re z > 0.
(2) y s , y s x, y s u (s C) with eigenvalue s(s 1).
(3) y s v, y s uv, y s xv with eigenvalue s(s + 1).
(4) x, y, u, v, xv, uv with eigenvalue 0.
(5) All Maass wave forms.

We fix two positive integers m and n throughout this section.


For an element Hn , we set
L := Z(m,n) + Z(m,n)
It follows from the positivity of Im that the elements fkl , hkl () (1 k m, 1 l n)
of L are linearly independent over R. Therefore L is a lattice in C(m,n) and the set
{ fkl , hkl () | 1 k m, 1 l n } forms an integral basis of L . We see easily that if
is an element of Hn , the period matrix := (In , ) satisfies the Riemann conditions
(RC.1) and (RC.2) :
(RC.1) Jn t = 0 ;
(RC.2) 1i Jn t > 0.
Thus the complex torus A := C(m,n) /L is an abelian variety.
It might be interesting to investigate the spectral theory of the Laplacian n,m on a
fundamental domain Fn,m . But this work is very complicated and difficult at this moment.
It may be that the first step is to develop the spectral theory of the Laplacian on the
abelian variety A . The second step will be to study the spectral theory of the Laplacian
n (see (1.4)) on the moduli space n \Hn of principally polarized abelian varieties of dimension n. The final step would be to combine the above steps and more works to develop
the spectral theory of the Lapalcian n,m on Fn,m . Maass-Jacobi forms play an important
role in the spectral theory of n,m on Fn,m . Here we deal only with the spectral theory
on L2 (A ).
We fix an element = X + iY of Hn with X = Re and Y = Im . For a pair (A, B)
with A, B Z(m,n) , we define the function E;A,B : C(m,n) C by
E;A,B (Z) = e2i((

t AU

)+ ((BAX)Y 1 t V ))

where Z = U + iV is a variable in C(m,n) with real U, V .


Lemma 4.1. For any A, B Z(m,n) , the function E;A,B satisfies the following functional
equation
E;A,B (Z + + ) = E;A,B (Z),

Z C(m,n)

for all , Z(m,n) . Thus E;A,B can be regarded as a function on A .

INVARIANT METRICS AND LAPLACIANS ON SIEGEL-JACOBI SPACE

17

Proof. We write = X + iY with real X, Y. For any , Z(m,n) , we have


E;A,B (Z + + ) = E;A,B ((U + X + ) + i(V + Y ))
= e2i{ (

tA(U +X+))+ ((BAX)Y 1 t(V

+Y )) }

= e2i{ (

tAU + tAX+ tA)+ ((BAX)Y 1 t V

= e2i{ (

tAU ) + ((BAX)Y 1 t V

+B t AX t )}

)}

= E;A,B (Z).
Here we used the fact that

tA

and B t are integral.

Lemma 4.2. The metric


ds2 = (Im )1 t (dZ) dZ)

(m,n)

is a K
ahler metric on A invariant under the action (5.15) of J = Sp(n, Z) HZ
(, Z) with fixed. Its Laplacian of ds2 is given by



t
= (Im )
.
Z
Z

on

Proof. The proof can be found [20].



We let L2 (A ) be the space of all functions f : A C such that
Z
|f (Z)|2 dv ,
||f || :=
A

R
where dv is the volume element on A normalized so that A dv = 1. The inner product
( , ) on the Hilbert space L2 (A ) is given by
Z
f (Z) g(Z) dv , f, g L2 (A ).
(f, g) :=
A


Theorem 4.2. The set E;A,B | A, B Z(m,n) is a complete orthonormal basis for
L2 (A ). Moreover we have the following spectral decomposition of :


L2 (A ) = A,BZ(m,n) C E;A,B .

Proof. The complete proof can be found in [20]

References
[1] R. Berndt, Sur larithmetique du corps des fonctions elliptiques de niveau N , Seminaire de Theorie de
nombres de Paris 1982-83, Birkh
auser, Boston (1984), 2132.
[2] R. Berndt and R. Schmidt, Elements of the Representation Theory of the Jacobi Group, Birkh
auser,
1998.
[3] M. Eichler and D. Zagier, The Theory of Jacobi Forms, Progress in Math., 55, Birkh
auser, Boston, Basel
and Stuttgart, 1985.
[4] E. Freitag, Siegelsche Modulfunktionen, Grundlehren der Math. Wissenschaften, Springer-Verlag, 1983.
[5] M. Kashiwara and M. Vergne, On the Segal-Shale-Weil Representations and Harmonic Polynomials,
Invent. Math. 44 (1978), 147.

18

JAE-HYUN YANG

[6] J. Kramer, A geometrical approach to the theory of Jacobi forms, Compositio Math. 79 (1991), 119.
[7] J. Kramer, An arithmetical theory of Jacobi forms in higher dimensions, J. reine angew. Math. 458
(1995), 157182.
[8] H. Maass, Die Differentialgleichungen in der Theorie der Siegelschen Modulfunktionen, Math. Ann. 126
(1953), 4468.
[9] H. Maass, Siegel modular functions and Dirichlet series, Lecture Notes in Math. 216, Springer-Verlag,
Berlin and New York, 1971.
[10] Minkowski, H.: Gesammelte Abhandlungen. Chelsea, New York (1967).
[11] B. Runge, Theta functions and Siegel-Jacobi functions, Acta Math. 175 (1995), 165196.
[12] C. L. Siegel, Symplectic Geometry, Gesammelte Abhandlungen, Band II, Springer-Verlag (1966), 274
359.
[13] C. L. Siegel, Topics in Complex Function Theory : Abelian Functions and Modular Functions of Several
Variables, vol. III, Wiley-Interscience, 1973.
[14] J.-H. Yang, The Siegel-Jacobi Operator, Abh. Math. Sem. Univ. Hamburg 63 (1993), 135146.
[15] J.-H. Yang, Remarks on Jacobi forms of higher degree, Proc. of the 1993 Workshop on Automorphic
Forms and Related Topics, the Pyungsan Institute for Mathematical Sciences, Seoul (1993), 3358.
[16] J.-H. Yang, Singular Jacobi Forms, Trans. of American Math. Soc. 347, No. 6 (1995), 20412049.
[17] J.-H. Yang, Construction of Modular Forms from Jacobi Forms, Canadian J. of Math. 47 (1995), 1329
1339.
[18] J.-H. Yang, A geometrical theory of Jacobi forms of higher degree, Proceedings of Symposium on Hodge
Theory and Algebraic Geometry ( edited by Tadao Oda ), Sendai, Japan (1996), 125-147 or Kyungpook
Math. J. 40 (2), 209237 (2000) or arXiv:math.NT/0602267.
[19] J.-H. Yang, A Note on Maass-Jacobi Forms, Kyungpook Math. J. 43 (2003), 547566.
[20] J.-H. Yang, A note on a fundamental domain for Siegel-Jacobi space, Houston Journal of Mathematics,
Vol. 32, No. 3 (2006), 701712.
[21] C. Ziegler, Jacobi Forms of Higher Degree, Abh. Math. Sem. Hamburg 59 (1989), 191224.
Department of Mathematics, Inha University, Incheon 402-751, Korea
E-mail address: jhyang@inha.ac.kr