Bruinier O2l Forms

To my parents
Table of Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1 Vector valued modular forms for the metaplectic group . . . 15
1.1 The Weil representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.2 Poincare series and Eisenstein series . . . . . . . . . . . . . . . . . . . . . . 19
1.2.1 Poincare series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.2.2 The Petersson scalar product . . . . . . . . . . . . . . . . . . . . . . 22
1.2.3 Eisenstein series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.3 Non-holomorphic Poincare series of negative weight . . . . . . . . . 27
2 The regularized theta lift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.1 Siegel theta functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.2 The theta integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.3 Unfolding against F
,m
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.4 Unfolding against
L
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3 The Fourier expansion of the theta lift . . . . . . . . . . . . . . . . . . . . 63
3.1 Lorentzian lattices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
3.1.1 The hyperbolic Laplacian. . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.2 Lattices of signature (2, l) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.3 Modular forms on orthogonal groups . . . . . . . . . . . . . . . . . . . . . . 84
3.4 Borcherds products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.4.1 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4 Some Riemann geometry on O(2, l) . . . . . . . . . . . . . . . . . . . . . . . 95
4.1 The invariant Laplacian . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4.2 Reduction theory and L
p
-estimates . . . . . . . . . . . . . . . . . . . . . . . 103
4.3 Modular forms with zeros and poles on Heegner divisors . . . . . 112
5 Chern classes of Heegner divisors . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.1 A lifting into the cohomology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
5.1.1 Comparison with the classical theta lift . . . . . . . . . . . . . 135
5.2 Modular forms with zeros and poles on Heegner divisors II . . . 137
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
VIII Table of Contents
Notation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
1 Vector valued modular forms for
the metaplectic group
In the following chapter we briey recall some facts about the metaplectic
cover of SL
2
(R), the Weil representation, and certain vector valued modular
forms. We essentially follow the terminology of Borcherds [Bo2]. Moreover,
we consider vector valued Poincare and Eisenstein series in some detail.
1.1 The Weil representation
As usual, we denote by H = C; () > 0 the upper complex half plane.
Throughout (unless otherwise specied) we will use as a standard variable
on H and write x for its real part and y for its imaginary part, respectively.
For z C we put e(z) = e
2iz
, and denote by
z = z
1/2
the principal branch
of the square root, so that arg(
z) (/2, /2]. For any integer k we put

z
k/2
= (z
1/2
)
k
. More generally if b C, then we dene z
b
= e
b Log(z)
, where
Log(z) denotes the principal branch of the logarithm. If x is a non-zero real
number, we let sgn(x) = x/[x[.
We write Mp
2
(R) for the metaplectic group, i.e. the double cover of
SL
2
(R), realized by the two choices of holomorphic square roots of c +d
for

a b
c d
SL
2
(R). Thus the elements of Mp
2
(R) are pairs
(M, ()),
where M =

a b
c d
SL
2
(R), and denotes a holomorphic function on H
with
()
2
= c +d.
The product of (M
1
,
1
()), (M
2
,
2
()) Mp
2
(R) is given by
M
1
,
1
()
M
2
,
2
()
M
1
M
2
,
1
(M
2
)
2
()
,
where M =
a+b
c+d
denotes the usual action of SL
2
(R) on H. The map
a b
c d
a b
c d
a b
c d
c +d
(1.1)
denes a locally isomorphic embedding of SL
2
(R) into Mp
2
(R).
J.H. Bruinier: LNM 1780, pp. 1538, 2002.
c Springer-Verlag Berlin Heidelberg 2002
16 1 Vector valued modular forms for the metaplectic group
Let Mp
2
(Z) be the inverse image of SL
2
(Z) under the covering map
Mp
2
(R) SL
2
(R). It is well known that Mp
2
(Z) is generated by
T =
1 1
0 1
, 1
,
S =
0 1
1 0
.
One has the relations S
2
= (ST)
3
= Z, where
Z =
1 0
0 1
, i
is the standard generator of the center of Mp

2
(Z). We will often use the
abbreviations
1
:= SL
2
(Z),
= (
1 n
0 1
) ; n Z
1
, and
:= 'T` =
1 n
0 1
, 1
; n Z
.
Throughout let L be an even lattice, i.e. a free Z-module of nite rank,
equipped with a symmetric Z-valued bilinear form (, ) such that the associ-
ated quadratic form
q(x) =
1
2
(x, x)
takes its values in Z. We assume that L is non-degenerated and denote its
signature by (b
+
, b
). We write
L
= x L Q; (x, y) Z for all y L

for the dual lattice. Then the quotient L
/L is a nite Abelian group, the so-

called discriminant group. The modulo 1 reduction of q(x) is a Q/Z-valued
quadratic form on L
/L whose associated bilinear form is the modulo 1 re-

duction of the bilinear form (, ) on L
.
Recall that there is a unitary representation
L
of Mp
2
(Z) on the group
algebra C[L
/L]. If we denote the standard basis of C[L
/L] by (e
)
L
/L
,
then
L
can be dened by the action of the generators S, T Mp
2
(Z) as
follows (cp. [Bo2]):
L
(T)e
= e(q())e
(1.2)
L
(S)e
i
b
b
+
[L
/L[
/L
e((, ))e
. (1.3)
This representation is essentially the Weil representation attached to the
quadratic module (L
/L, q) (cf. [No]). It factors through the nite group

1.1 The Weil representation 17
SL
2
(Z/NZ) if b
+
+ b
is even, and through a double cover of SL

2
(Z/NZ) if
b
+
+ b
is odd, where N is the smallest integer such that Nq() Z for all
L
. Note that
L
(Z)e
= i
b
b
+
e
. (1.4)
We write ', ` for the standard scalar product on C[L
/L], i.e.
/L
/L
/L
.
For , L
/L and (M, ) Mp
2
(Z) we dene the coecient
(M, ) of
the representation
L
by
(M, ) = '
L
(M, )e
, e
`.
Shintani proved the following explicit formula for
L
(cf. [Sn], Prop. 1.6).
Proposition 1.1. Let , L
/L and M =

a b
c d
SL
2
(Z). Then the
coecient
(

M) is given by
i
(b
b
+
)(1sgn(d))
,a
e(abq()), (1.5)
if c = 0, and by
i
(b
b
+
) sgn(c)
[c[
(b
+b
+
)/2
[L
/L[
rL/cL
e
a( +r, +r) 2(, +r) +d(, )

2c
,
(1.6)
if c = 0. Here,
,
denotes the Kronecker-delta.
Let
1
2
Z and f be a C[L
/L]-valued function on H. For (M, )

Mp
2
(Z) we dene the Petersson slash operator by
(f [
(M, )) () = ()
2
L
(M, )
1
f(M). (1.7)
As usual, by f f [
(M, ) an operation of Mp
2
(Z) on functions f : H
C[L
/L] is dened.
We denote by
L
the dual representation of
L
. If we think of
L
(M, )
((M, ) Mp
2
(Z)) as a matrix with entries in C, then
L
(M, ) is simply
the complex conjugate of
L
(M, ). We have a dual operation of Mp
2
(Z)
on functions f : H C[L
/L], given by
(f [
(M, )) () = ()
2
L
(M, )
1
f(M). (1.8)
Now assume that f : H C[L
/L] is a holomorphic function which is

invariant under the [
-operation of T Mp
2
(Z). We denote the components
of f by f
, so that f =

L
/L
e
. The invariance of f under T implies

that the functions e(q())f
() are periodic with period 1. Thus f has a

Fourier expansion
f() =
/L
nZq()
e
c(, n)e(n),
with Fourier coecients c(, n) =
1
0
f
()e(n) dx. Using the abbreviation

e
() := e
e() we may write

f() =
/L
nZq()
c(, n)e
(n). (1.9)
By means of the scalar product in C[L
/L] the coecients can be expressed

by
c(, n) =
1
0
'f(), e
(n )` dx. (1.10)
Denition 1.2. Let
1
2
Z. A function f : H C[L
/L] is called a modular

form of weight with respect to
L
and Mp
2
(Z) if
i) f [
(M, ) = f for all (M, ) Mp

2
(Z),
ii) f is holomorphic on H,
iii) f is holomorphic at the cusp .
Here condition (iii) means that f has a Fourier expansion of the form
f() =
/L
nZq()
n0
c(, n)e
(n).
Moreover, if all c(, n) with n = 0 vanish, then f is called a cusp form. The
C-vector space of modular forms of weight with respect to
L
and Mp
2
(Z)
is denoted by M
,L
, the subspace of cusp forms by S
,L
.
It is easily seen that M
,L
is nite dimensional.
Example 1.3. Fix a positive integer t. Let L be the 1-dimensional lattice Z,
equipped with the positive denite quadratic form q(a) = ta
2
. Then L
/L
=
Z/2tZ, and M
1/2,L
is isomorphic to the space J
,t
of Jacobi forms of weight
and index t (cf. [EZ] Theorem 5.1). If L denotes the lattice Z with the
negative denite quadratic form q(a) = ta
2
, then M
1/2,L
is isomorphic
to the space .
,t
of skew-holomorphic Jacobi forms of weight and index t
as dened in [Sk].
1.2 Poincare series and Eisenstein series 19
1.2 Poincare series and Eisenstein series
In this section we carry out the standard construction of Poincare and Eisen-
stein series for the space M
,L
. For simplicity we assume that (b
+
, b
) = (2, l),
(1, l 1), or (0, l 2) with l 3. Moreover, we assume that = 1 + l/2 (as
we will only need this case later).
1.2.1 Poincare series
Let L
/L and m Z q() with m > 0. Then e
(m) is a holomorphic
function H C[L
/L] which is invariant under the [
-operation of T
Mp
2
(Z). Moreover, by (1.4) we have
e
(m) [
Z = i
2
i
l2
e
(m) = e
(m).
Thus e
(m) is also invariant under the action of Z

2
.
We dene the Poincare series P
L
,m
of index (, m) by
P
L
,m
() =
1
2
(M,)
\ Mp
2
(Z)
e
(m) [
(M, ) (1.11)
(recall that

= 'T`). Since 5/2, the usual argument shows that

this series converges normally on H and thereby denes a Mp
2
(Z)-invariant
holomorphic function H C[L
/L]. We will often omit the superscript L, if

it is clear from the context.
Theorem 1.4. The Poincare series P
,m
has the Fourier expansion
P
,m
() =
/L
nZq()
n>0
p
,m
(, n)e
(n)
with
p
,m
(, n) =
m,n
(
,
+
,
)
+ 2
n
m
1
2

cZ{0}
H
c
(, m, , n)J
1
4
[c[
mn
. (1.12)
Here H
c
(, m, , n) denotes the generalized Kloosterman sum
H
c
(, m, , n) =
e
i sgn(c)/2
[c[
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
,
(1.13)
and J
the Bessel function of the rst kind ([AbSt] Chap. 9). In particular
P
,m
S
,L
.
Recall the abbreviations
1
= SL
2
(Z) and
= (
1 n
0 1
) ; n Z. The
sum in (1.13) runs over all primitive residues d modulo c and

a b
c d
is a
representative for the double coset in
`
1
/
with lower row (c d
) and
d
d (mod c). Observe that the expression
a b
c d
ma+nd
c
does not
depend on the choice of the coset representative.
The coecients
a b
c d
are universally bounded, because

L
factors
through a nite group. Hence there exists a constant C > 0 such that
H
c
(, m, , n) < C for all L
/L, n Z q(), and c Z 0.

This, together with the asymptotic behavior J
(t) = O(t
) (t 0) of the
J-Bessel function, implies that the series (1.12) converges absolutely. If we
also use the asymptotic property J
(t) = O(t
1/2
) as t , we nd that
p
,m
(, n) = O(n
1
) for n . (This is actually a very poor bound. For
even l Delignes theorem, previously the Ramanujan-Petersson conjecture,
states that p
,m
(, n) = O(n
(1)/2+
).)
Proof of Theorem 1.4. Let L
/L and n Z q(). According to (1.10)

the coecient p
,m
(, n) can be computed as
p
,m
(, n) =
1
2
1

(M,)
\ Mp
2
(Z)
e
(m) [
(M, ), e
(n )
dx.
Since e
(m) is invariant under the action of Z

2
, this can be written in the
form
1
0
'e
(m), e
(n )` dx +
1
0
'e
(m) [
Z, e
(n )` dx
+
1
a b
c d
\
1
c=0
e
(m) [
a b
c d
, e
(n )
dx
=
m,n
(
,
+
,
)
+
c=0
a b
c d
\
1
/
(c +d)
a b
c d
1
e
(m
a b
c d
), e
(n )
dx.
Since
L
is unitary we have
a b
c d
1
e
(m
a b
c d
), e
(n )
a b
c d
a b
c d
e(n),
and thereby
p
,m
(, n) =
m,n
(
,
+
,
)
+
c=0
a b
c d
\
1
/
a b
c d
(c +d)
a b
c d
dx.
(1.14)
We now use the elementary identities

c +d = sgn(c)
+d/c and
a b
c d
=
a
c

1
c
2
( +d/c)
. (1.15)
We nd that the latter sum in (1.14) equals
c=0
a b
c d
\
1
/
[c[
sgn(c)
a b
c d
( +d/c)
ma
c

m
c
2
( +d/c)
n
dx
=
c=0
a b
c d
\
1
/
[c[
sgn(c)
a b
c d
ma
c
m
c
2
n( d/c)
dx
=
c=0
[c[
sgn(c)
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
m
c
2
dx.
We substitute = iw in the integral and obtain for the above expression
2i
c=0
[c[
sgn(c)
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
1
2i
C+i
Ci
w
exp
2m
c
2
w
+ 2nw
dw,
where C is a positive real constant. We now use i
sgn(c)
= e
i sgn(c)/2
and the denition of the generalized Kloosterman sum (1.13) and nd
p
,m
(, n) =
m,n
(
,
+
,
) + 2
c=0
[c[
1
H
c
(, m, , n)
1
2i
C+i
Ci
w
exp
2m
c
2
w
+ 2nw
dw. (1.16)
If n 0, we consider the limit C and deduce that p
,m
(, n) = 0. If
n > 0, then the latter integral is an inverse Laplace transform that can be
found in any standard reference on integral transforms. According to [E2]
p. 245 (40) one has
1
2i
C+i
Ci
w
exp
2m
c
2
w
+ 2nw
dw =
c
2
n
m
1
2
J
1
4
[c[
mn
.
If we insert this into (1.16) we obtain the assertion. .
1.2.2 The Petersson scalar product
Let f, g be modular forms in M
,L
. Then it can be easily checked that
'f(), g()`y
is Mp
2
(Z)-invariant. (As before, ', ` denotes the scalar product on C[L
/L].)
If f or g is a cusp form, then 'f(), g()`y
is bounded on the standard

fundamental domain
T = = x +iy H; [x[ 1/2, [[ 1
for the action of
1
on H. (Hence the invariance property implies that
'f(), g()`y
is bounded on the whole upper half plane H.) We dene the

Petersson scalar product of f and g by
(f, g) =
F
'f(), g()`y
dxdy
y
2
. (1.17)
It is easily seen that the integral does not depend on the choice of the fun-
damental domain. The space S
,L
endowed with the scalar product (, ) is a
nite dimensional Hilbert space.
The Petersson coecient formula for S
,L
has the following form.
Proposition 1.5. Let L
/L and m Z q() with m > 0. Let f be a

cusp form in S
,L
and denote its Fourier coecients by c(, n) (as in (1.9)).
Then the scalar product of f with the Poincare series P
,m
is given by
(f, P
,m
) = 2
( 1)
(4m)
1
c(, m).
Proof. Let
1
=
1
/1. We use the usual unfolding argument and nd
(f, P
,m
) =
1
2
1
\H
f(),
(M,)
\ Mp
2
(Z)
e
(m) [
(M, )
dxdy
y
2
=
1
\H
\
1
'f(M), e
(mM)`(M)
dxdy
y
2
= 2
\H
'f(), e
(m)`y
2
dxdy.
We insert the Fourier expansion of f and infer:
(f, P
,m
) = 2
0
1
nZq()
n>0
c(, n)e(n)e(mx +miy)y
2
dxdy
= 2
y=0
c(, m)e
4my
y
2
dy
= 2
( 1)
(4m)
1
c(, m).
.
As a corollary we obtain that the Poincare series P
,m
generate the space
S
,L
.
1.2.3 Eisenstein series
Let L
/L with q() Z. Then the vector e
= e
(0) C[L
/L], consid-
ered as a constant function H C[L
/L], is invariant under the [
-action of
T, Z
2
Mp
2
(Z). The Eisenstein series
E
L
() =
1
2
(M,)
\ Mp
2
(Z)
e
(M, ) (1.18)
converges normally on H and therefore denes a Mp
2
(Z)-invariant holomor-
phic function on H.
Theorem 1.6. The Eisenstein series E

E
() =
/L
nZq()
n0
q
(, n)e
(n)
with
q
(, n) =
,
+
,
, if n = 0,
(2)
n
1
()
cZ{0}
[c[
1
H
c
(, 0, , n), if n > 0.
(1.19)
Here, H
c
(, 0, , n) denotes the generalized Kloosterman sum dened in
(1.13). In particular E
is an element of M
,L
.
Proof. We proceed in the same way as in the proof of Theorem 1.4. Let
L
/L and n Z q(). By (1.10) we have

q
(, n) =
1
2
1

(M,)
\ Mp
2
(Z)
e
(M, ), e
(n )
dx.
Because e
is invariant under the action of Z

2
, this equals
0,n
(
,
+
,
) +
c=0
a b
c d
\
1
/
(c +d)
a b
c d
1
e
, e
(n )
dx
=
0,n
(
,
+
,
) +
c=0
a b
c d
\
1
/
a b
c d
(c +d)
e(n) dx.
We now evaluate the integral. Since

c +d = sgn(c)
+d/c, we have
(c +d)
e(n) dx = [c[
sgn(c)
( +d/c)
e(n) dx
= [c[
sgn(c)
nd
c
e(n) dx.
We substitute = iw and nd
(c +d)
e(n) dx
= 2i
[c[
sgn(c)
nd
c
1
2i
C+i
Ci
w
e
2nw
dw,
where C is a positive real constant. If n 0, one can consider the limit C
to infer that the integral is 0. It follows that q
(, n) =
0,n
(
,
+
,
)
in this case. Now let n > 0. Then the integral is an inverse Laplace transform.
According to [E2] p. 238 (1) we have
1
2i
C+i
Ci
w
e
2nw
dw =
(2n)
1
()
.
(In fact, one can deform the path of integration and essentially obtains the
Hankel integral for 1/().) Hence we get
q
(, n) =
(2)
n
1
()
c=0
a b
c d
\
1
/
[c[
sgn(c)
a b
c d
nd
c
.
If we use i
sgn(c)
= e
i sgn(c)/2
and substitute in the denition of
H
c
(, 0, , n), we obtain the assertion. .
For computational purposes the formula (1.19) for the coecients q
(, n)
is not very useful. An explicit formula, involving special values of Dirichlet
L-series and nitely many representation numbers modulo prime powers at-
tached to the lattice L, is derived in [BK].
To give an example we briey consider the case that L is unimodular. Then
L
/L is trivial and M
,L
is just the space of elliptic modular forms of weight
1 + l/2. We have one Eisenstein series E with Fourier coecients q(n) and
q(0) = 2. For n > 0 the above expression for the coecients can be simplied
as follows. First note that l 2 (mod 8). This implies e
i sgn(c)/2
= 1
and
H
c
(, 0, , n) =
1
[c[
d(c)
nd
c
=
1
[c[
d|(c,n)
d>0
([c[/d)d.
Here we have used the evaluation of the Ramanujan sum

d(c)
e(
nd
c
) by
means of the Moebius function . Thus we nd
q(n) = 2
(2)
n
1
()
c=1
c
d|(c,n)
d>0
(c/d)d
= 2
(2)
n
1
()()
d|n
d
1
= 2
(2)
()()
1
(n),
where
1
(n) denotes the sum of the (1)-th powers of the positive divisors
of n and (s) the Riemann zeta function. Using () =
(2i)
2!
B
with the
-th Bernoulli number B
, and the fact that 2 (mod 4), we get

q(n) =
4
B
1
(n).
Thus in this case the Eisenstein series E(z) is the classical Eisenstein series
for
1
, normalized such that its constant term equals 2.
Let us now turn back to the general case and determine the number of
linearly independent Eisenstein series E
. The [
-invariance of E
under Z
and the identity e
Z = e
imply
E
= E
Z =
1
2
(M,)
\ Mp
2
(Z)
e
Z [
(M, ) = E
.
Thus the space M
Eis
,L
spanned by all Eisenstein series E
equals the space

spanned by the E
where runs through a set of representatives of

L
/L; q() Z modulo the action of 1. Comparing constant terms one

nds that the set of Eisenstein series E
with as above is already linearly

independent. We obtain
dim(M
Eis
,L
) = # L
/L; q() Z, 2 = 0 +L
+
1
2
# L
/L; q() Z, 2 = 0 +L. (1.20)

Now let f M
,L
and write the Fourier expansion of f in the form
f() =
/L
nZq()
c(, n)e
(n).
The invariance of f under Z and (1.4) imply that c(, n) = c(, n). Hence
f
1
2
/L
q()Z
c(, 0)E
1.3 Non-holomorphic Poincare series of negative weight 27

lies in S
,L
. This shows that M
,L
= M
Eis
,L
S
,L
and (1.20) is a formula for
the codimension of S
,L
in M
,L
.
In later applications we will mainly be interested in the Eisenstein series
E
0
() which we simply denote by E(). In the same way we write q(, n) for
the Fourier coecients q
0
(, n) of E().
Let us nally cite the following result of [BK]:
Proposition 1.7. The coecients q(, n) of E() are rational numbers.
1.3 Non-holomorphic Poincare series of negative weight
As in the previous section we assume that L is an even lattice of signature
(b
+
, b
) = (2, l), (1, l 1), or (0, l 2) with l 3. Put k = 1 l/2 and

= 1 + l/2. We now construct certain vector valued Maass-Poincare series
for Mp
2
(Z) of weight k. Series of a similar type are well known and appear
in many places in the literature (see for instance [He, Ni, Fa]).
Let M
,
(z) and W
,
(z) be the usual Whittaker functions as dened
in [AbSt] Chap. 13 p. 190 or [E1] Vol. I Chap. 6 p. 264. They are linearly
independent solutions of the Whittaker dierential equation
d
2
w
dz
2
+
1
4
+

z

2
1/4
z
2
w = 0. (1.21)
The functions M
,
(z) and M
,
(z) are related by the identity
W
,
(z) =
(2)
(
1
2
)
M
,
(z) +
(2)
(
1
2
+ )
M
,
(z) (1.22)
([AbSt] p. 190 (13.1.34)). This implies in particular W
,
(z) = W
,
(z). As
z 0 one has the asymptotic behavior
M
,
(z) z
+1/2
( /
1
2
N), (1.23)
W
,
(z)
(2)
( + 1/2)
z
+1/2
( 1/2). (1.24)
If y R and y one has
M
,
(y) =
(1 + 2)
( + 1/2)
e
y/2
y
(1 +O(y
1
)), (1.25)
W
,
(y) = e
y/2
y
(1 +O(y
1
)). (1.26)
For convenience we put for s C and y R
>0
:
s
(y) = y
k/2
M
k/2, s1/2
(y). (1.27)
In the same way we dene for s C and y R 0:
)
s
(y) = [y[
k/2
W
k/2 sgn(y), s1/2
([y[). (1.28)
If y < 0 then equation (1.22) implies that
s
([y[) =
(1 +k/2 s)
(1 2s)
)
s
(y)
(1 +k/2 s)(2s 1)
(1 2s)(s +k/2)

1s
([y[).
(1.29)
The functions
s
(y) and )
s
(y) are holomorphic in s. Later we will be
interested in certain special s-values. For y > 0 we have
k/2
(y) = y
k/2
M
k/2, k/21/2
(y) = e
y/2
, (1.30)
)
1k/2
(y) = y
k/2
W
k/2, 1/2k/2
(y) = e
y/2
. (1.31)
Using the standard integral representation
(1/2 +)W
,
(z) = e
z/2
z
+1/2
0
e
tz
t
1/2+
(1 +t)
1/2++
dt
(1() > 1/2, 1(z) > 0) of the W-Whittaker function ([E1] Vol. I p. 274
(18)), we nd for y < 0:
)
1k/2
(y) = [y[
k/2
W
k/2, 1/2k/2
([y[)
= e
|y|/2
[y[
1k
0
e
t|y|
(1 +t)
k
dt
= e
|y|/2
[y[
1k
1
e
t|y|
t
k
dt.
If we insert the denition of the incomplete Gamma function (cf. [AbSt] p. 81)
(a, x) =
x
e
t
t
a1
dt, (1.32)
we obtain for y < 0 the identity
)
1k/2
(y) = e
y/2
(1 k, [y[). (1.33)
The usual Laplace operator of weight k (cf. [Ma1])
k
= y
2

2
x
2
+

2
y
2
+iky

x
+i

y
(1.34)
acts on smooth functions f : H C[L
/L] component-wise. Let L
/L
and m Z+q() with m < 0. Then it can be easily checked that the function
s
(4[m[y)e
(mx)
is invariant under the [
k
-operation of T Mp
2
(Z) (as dened in (1.7)) and
an eigenfunction of
k
with eigenvalue
s(1 s) + (k
2
2k)/4.
Denition 1.8. With the above notation we dene the Poincare series F
L
,m
of index (, m) by
F
L
,m
(, s) =
1
2(2s)
(M,)
\ Mp
2
(Z)
[
s
(4[m[y)e
(mx)] [
k
(M, ),
(1.35)
where = x + iy H and s = + it C with > 1. If it is clear from the
context, to which lattice L the series (1.35) refers, we simply write F
,m
(, s).
By (1.23) and (1.27) the above series (1.35) has the local majorant
(M,)
\ Mp
2
(Z)
(M)
.
Hence (1.35) converges normally for H, s C and > 1 and thereby
denes a Mp
2
(Z)-invariant function on H. As in [Ma1] it can be seen that
k
commutes with the action of Mp
2
(Z), i.e. that
k
(f [
k
(M, )) = (
k
f) [
k
(M, )
for any smooth function f : H C[L
/L] and any (M, ) Mp

2
(Z). We may
infer that F
,m
is an eigenfunction of
k
:
k
F
,m
=
s(1 s) + (k
2
2k)/4
F
,m
.
(Thus F
,m
is automatically real analytic as a solution of an elliptic dier-
ential equation.) The invariance of F
,m
under the action of Z Mp
2
(Z)
implies that F
,m
= F
,m
.
Let f : H C[L
/L] be a continuous function with the property f [

k
T =
f. By virtue of the same argument as in section 1.1 one nds that f has a
Fourier expansion of the form
f() =
/L
nZ+q()
c(, n, y)e
(nx), (1.36)
where the coecients are given by
c(, n, y) =
1
0
'f(), e
(nx)` dx. (1.37)

Theorem 1.9. The Poincare series F
,m
F
,m
(, s) =
(1 +k/2 s)
(2 2s)(s +k/2)
1s
(4[m[y)(e
(mx) + e
(mx))
+
/L
q()Z
b(, 0, s)y
1sk/2
e
/L
nZ+q()
n>0
b(, n, s))
s
(4ny)e
(nx) +

F
,m
(, s),
where

F
,m
(, s) is the T-invariant function
F
,m
(, s) =
/L
nZ+q()
n<0
b(, n, s))
s
(4ny)e
(nx),
and the Fourier coecients b(, n, s) are given by
2[n/m[
k1
2
(s +k/2)
cZ{0}
H
c
(, m, , n)I
2s1
4
[c[
[mn[
, if n > 0,
4
1k/2
1+sk/2
[m[
sk/2
(2s 1)(s +k/2)(s k/2)
cZ{0}
[c[
12s
H
c
(, m, , 0), if n = 0,
(1 +k/2 s)
(2s)(1 2s)
m,n
(
,
+
,
)
+
2[n/m[
k1
2
(s k/2)
cZ{0}
H
c
(, m, , n)J
2s1
4
[c[
[mn[
, if n < 0.
Here H
c
H
c
(, m, , n) =
e
i sgn(c)k/2
[c[
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
(1.38)
and J
(z), I
(z) the usual Bessel functions as dened in [AbSt] Chap. 9.

The sum in (1.38) runs over all primitive residues d modulo c and

a b
c d
is a
`
1
/
with lower row (c d
) and
d

1
a b
c d
ma+nd
c
does not
depend on the choice of the coset representative. The fact that H
c
(, m, , n)
is universally bounded implies that the series for the coecients b(, n, s)
converge normally in s for > 1.
Proof of Theorem 1.9. We split the sum in equation (1.35) into the sum over
1, Z, Z
2
, Z
3
` Mp
2
(Z) and the sum over (M, )

` Mp
2
(Z), where
M =

a b
c d
with c = 0. The latter sum will be denoted by H(, s). Since

e
(mx) [
k
Z = e
(mx), we nd that the rst term equals

1
(2s)
s
(4[m[y)(e
(mx) + e
(mx)).
According to (1.29) this can be rewritten as
(1 +k/2 s)
(2 2s)(s +k/2)
1s
(4[m[y)(e
(mx) + e
(mx))
+
(1 +k/2 s)
(2s)(1 2s)
)
s
(4my)(e
(mx) + e
(mx)). (1.39)
We now calculate the Fourier expansion of H(, s). Since e
(mx) is in-
variant under the action of Z
2
, we may write H(, s) in the form
1
(2s)
a b
c d
\
1
c=0
[
s
(4[m[y)e
(mx)] [
k
a b
c d
.
Let L
/L and n Z+q() and denote the (, n)-th Fourier coecient of

H(, s) by c(, n, y) (cp. (1.36)). Then we have
c(, n, y) =
1
0
'H(, s), e
(nx)` dx
=
c=0
a b
c d
\
1
/
(c +d)
k
(2s)

s
4[m[
y
|c+d|
2
1
L
a b
c d
m1
a b
c d
, e
(nx)
dx.
By (1.15) we nd that
1
L
a b
c d
m1
a b
c d
, e
(nx)
=
1
a b
c d
ma
c
1
m
c
2
( +d/c)
e(nx),
and therefore
c(, n, y) =
1
(2s)
c=0
a b
c d
\
1
/
a b
c d
ma
c
(c +d)
k
4[m[
y
[c +d[
2
m
c
2
( +d/c)
nx
dx.
If we use

c +d = sgn(c)
+d/c and substitute x by x d/c in the

integral, we obtain
c(, n, y) =
1
(2s)
c=0
a b
c d
\
1
/
a b
c d
ma +nd
c
[c[
k
sgn(c)
k
4[m[
y
c
2
[[
2
mx
c
2
[[
2
nx
dx
=
1
(2s)
c=0
[c[
1k
i
k
H
c
(, m, , n)
4[m[
y
c
2
[[
2
mx
c
2
[[
2
nx
dx.
By the denition of
s
this equals
(4[m[y)
k/2
(2s)
c=0
[c[H
c
(, m, , n)
k/2
M
k/2, s1/2
4[m[
y
c
2
[[
2
mx
c
2
[[
2
nx
dx
=
(4[m[y)
k/2
(2s)
c=0
[c[H
c
(, m, , n)
y ix
y +ix
k/2
M
k/2, s1/2
4[m[y
c
2
(x
2
+y
2
)
mx
c
2
(x
2
+y
2
)
nx
dx. (1.40)
We abbreviate the latter integral by I and substitute x = yu. Then I is equal
to
y
1 iu
1 +iu
k/2
M
k/2, s1/2
4[m[
c
2
y(u
2
+ 1)
[m[u
c
2
y(u
2
+ 1)
nyu
du;
and if we set A = ny and B =
|m|
c
2
y
, we get
I = y
1 iu
1 +iu
k/2
M
k/2, s1/2
4B
u
2
+ 1
exp
2iBu
u
2
+ 1
+ 2iAu
du.
The integral I/(y(2s)) was evaluated by Hejhal in [He] p. 357. (Notice that
M
k/2, s1/2
(y)/(2s) is the same as Hejhals function G(y) with = k/2.
Moreover, W
,
(y) equals e
y/2
y
+1/2
( + 1/2 , 1 + 2, y) in Hejhals
notation.) We nd
I
y(2s)
=
[B/A[
(s k/2)
W
k/2, s1/2
(4[A[)J
2s1
[AB[
, if A > 0,
4
1+s
(2s 1)(s +k/2)(s k/2)
B
s
, if A = 0,
2
[B/A[
(s +k/2)
W
k/2, s1/2
(4[A[)I
2s1
[AB[
, if A < 0.
We resubstitute for A and B and obtain
[c[I
(2s)
=
[m/n[
(s k/2)
(4[n[y)
k/2
J
2s1
4
[c[
[mn[
)
s
(4ny), if n < 0,
4
1+s
[m[
s
[c[
12s
y
1s
(2s 1)(s +k/2)(s k/2)
, if n = 0,
2
[m/n[
(s +k/2)
(4[n[y)
k/2
I
2s1
4
[c[
[mn[
)
s
(4ny), if n > 0.
If we insert this into (1.40) we get c(, n, y) is given by
2[n/m[
k1
2
(s k/2)
c=0
H
c
(, m, , n)J
2s1
4
[c[
[mn[
)
s
(4ny), n < 0,
4
1k/2
1+sk/2
[m[
sk/2
(2s 1)(s +k/2)(s k/2)
c=0
[c[
12s
H
c
(, m, , 0)y
1sk/2
, n = 0,
2[n/m[
k1
2
(s +k/2)
c=0
H
c
(, m, , n)I
2s1
4
[c[
[mn[
)
s
(4ny), n > 0.
Combining this with (1.39) we obtain the assertion. .
Since
k
F
,m
(, 1 k/2) = 0, the Poincare series F
,m
(, s) are in par-
ticular interesting in the special case s = 1 k/2.
Proposition 1.10. For s = 1k/2 the Fourier expansion of F
,m
(, s) given
in Theorem 1.9 can be simplied as follows:
F
,m
(, 1 k/2) = e
(m) + e
(m) +
/L
nZ+q()
n0
b(, n, 1 k/2)e
(n)
+

F
,m
(, 1 k/2).
The function

F
,m
(, 1 k/2) has the expansion
F
,m
(, 1 k/2) =
/L
nZ+q()
n<0
b(, n, 1 k/2))
1k/2
(4ny)e
(nx),
and the Fourier coecients b(, n, 1 k/2) are given by
n
m
k1
2

cZ{0}
H
c
(, m, , n)I
1k
4
[c[
[mn[
, if n > 0,
(2)
2k
[m[
1k
(2 k)
cZ{0}
[c[
k1
H
c
(, m, , 0), if n = 0,
1
(1 k)
m,n
(
,
+
,
)
+
2
(1 k)
n
m
k1
2

cZ{0}
H
c
(, m, , n)J
1k
4
[c[
[mn[
, if n < 0.
Proof. By virtue of (1.30) and (1.31) this immediately follows from Theorem
1.9. .
Denition 1.11. A function f : H C[L
/L] is called a nearly holomorphic

modular form of weight k (with respect to
L
and Mp
2
(Z)), if
i) f [
k
2
(Z),
iii) f has a pole in , i.e. f has a Fourier expansion of the form
f() =
/L
nZ+q()
n
c(, n)e
(n).
The space of these nearly holomorphic modular forms is denoted by M
!
k,L
.
The Fourier polynomial
/L
nZ+q()
n<0
c(, n)e
(n)
is called the principal part of f.
The following proposition is an analogue of the theorem in [He] on p. 660
(see also [Ni] Theorem 6).
Proposition 1.12. Let f() be a nearly holomorphic modular form of weight
k and denote its Fourier coecients by c(, n) ( L
/L, n Z+q()). Then

f() =
1
2
/L
nZ+q()
n<0
c(, n)F
,n
(, 1 k/2).
Proof. Consider the function
g() = f()
1
2
/L
nZ+q()
n<0
c(, n)F
,n
(, 1 k/2).
The invariance of f under Z Mp
2
(Z) implies that c(, n) = c(, n).
According to Proposition 1.10 we have F
,n
(, 1k/2) = e
(n) +e
(n) +
O(1) for all , n. Hence g() is bounded as y . But g also satises
g [
k
(M, ) = g ((M, ) Mp
2
(Z)),
k
g = 0.
Thus g has to vanish identically (see [He] Chap. 9 Prop. 5.14c). .
Lemma 1.13. Let c Z0, , L
/L, m Z+q() and n Z+q().

Then
H
c
(, m, , n) = H
c
(, m, , n).
Proof. From Proposition 1.1 it can be easily deduced that
a b
c d
a b
c d
and
1
a b
c d
=
1
a b
c d
.
Now the assertion immediately follows from the denition of H
c
(, m, , n).
.
Remark 1.14. This implies that the Fourier coecients b(, n, 1k/2) of the
function F
,m
(, 1 k/2) are real numbers.
We now compare the Fourier coecients of F
,m
(, 1 k/2) with the
coecients of the holomorphic Poincare series P
,n
S
,L
constructed in the
previous section.

Then we have
H
c
(, m, , n) = H
c
(, n, , m).
Proof. We use the fact that (

M)
1
=

M
1
for any M
1
, and that =
2 k. We nd
H
c
(, m, , n) =
e
i sgn(c)k/2
[c[
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
=
e
i sgn(c)(2k)/2
[c[
d(c)
a b
c d
\
1
/
d b
c a
md +na
c
=
e
i sgn(c)/2
[c[
d(c)
a b
c d
\
1
/
a b
c d
na md
c
= H
c
(, n, , m).
.
/L, n Z+q(), m Z+q(), and m < 0.

Denote the (n, )-th Fourier coecient b(, n, 1 k/2) of F
,m
(, 1 k/2)
(see Proposition 1.10) by b
,m
(, n). If n < 0, then
b
,m
(, n) =
1
(1 k)
p
,n
(, m),
where p
,n
(, m) is the (, m)-th coecient of the Poincare series
P
,n
() S
,L
(see 1.12). If n = 0, then
b
,m
(, 0) = q
(, m),
where q
(, m) is the (, m)-th coecient of the Eisenstein series E
()
M
,L
(see 1.19).
Proof. We compare the formulas for the Fourier coecients given in Propo-
sition 1.10 and Theorem 1.4 (resp. Proposition 1.10 and Theorem 1.6) and
apply Lemma 1.15. Note that k 1 = 1 . .
For L
/L and n Zq() with n > 0 let a

,n
: S
,L
C denote the
functional in the dual space S
,L
of S
,L
which maps a cusp form f to its
(, n)-th Fourier coecient a
,n
(f). According to Proposition 1.5, a
,n
can
be described by means of the Petersson scalar product as
a
,n
=
(4n)
1
2( 1)
( , P
,n
). (1.41)
The following theorem can also be proved using Serre duality (see [Bo3]
Thm. 3.1). We deduce it from Proposition 1.16.
Theorem 1.17. There exists a nearly holomorphic modular form f M
!
k,L
with prescribed principal part
/L
mZ+q()
m<0
c(, m)e
(m)
(c(, m) C with c(, m) = c(, m)), if and only if the functional
/L
mZ+q()
m<0
c(, m)a
,m
,
equals 0 in S
,L
.
Proof. First, suppose that there exists such a nearly holomorphic modular
form f. Then by Proposition 1.12
f() =
1
2
/L
mZ+q()
m<0
c(, m)F
,m
(, 1 k/2),
and in particular
/L
mZ+q()
m<0
c(, m)
F
,m
(, 1 k/2) = 0.
Since

)
1k/2
(4ny)e
(nx) = 0,
this implies that
/L
mZ+q()
m<0
c(, m)b
,m
(, n) = 0
for all L
/L and n Z + q() with n < 0. (Here we have used the same

notation as in Proposition 1.16.) Applying Proposition 1.16 we nd that
/L
mZ+q()
m<0
c(, m)p
,n
(, m) =
/L
mZ+q()
m<0
c(, m)a
,m
(P
,n
)
= 0
for all L
/L and n Z+q() with n < 0. Since the Poincare series P

,n
generate the space S
,L
we obtain
/L
mZ+q()
m<0
c(, m)a
,m
= 0. (1.42)
Now assume that (1.42) holds. Then we may reverse the above argument
to infer that
/L
mZ+q()
m<0
c(, m)
F
,m
(, 1 k/2) = 0.
According to Proposition 1.10 this implies that
1
2
/L
mZ+q()
m<0
c(, m)F
,m
(, 1 k/2)
is a nearly holomorphic modular form with the right principal part. .
1 Vector valued modular forms for
the metaplectic group
In the following chapter we briey recall some facts about the metaplectic
cover of SL
2
(R), the Weil representation, and certain vector valued modular
forms. We essentially follow the terminology of Borcherds [Bo2]. Moreover,
we consider vector valued Poincare and Eisenstein series in some detail.
1.1 The Weil representation
As usual, we denote by H = C; () > 0 the upper complex half plane.
Throughout (unless otherwise specied) we will use as a standard variable
on H and write x for its real part and y for its imaginary part, respectively.
For z C we put e(z) = e
2iz
, and denote by
z = z
1/2
the principal branch
of the square root, so that arg(
z) (/2, /2]. For any integer k we put

z
k/2
= (z
1/2
)
k
. More generally if b C, then we dene z
b
= e
b Log(z)
, where
Log(z) denotes the principal branch of the logarithm. If x is a non-zero real
number, we let sgn(x) = x/[x[.
We write Mp
2
(R) for the metaplectic group, i.e. the double cover of
SL
2
(R), realized by the two choices of holomorphic square roots of c +d
for

a b
c d
SL
2
(R). Thus the elements of Mp
2
(R) are pairs
(M, ()),
where M =

a b
c d
SL
2
(R), and denotes a holomorphic function on H
with
()
2
= c +d.
The product of (M
1
,
1
()), (M
2
,
2
()) Mp
2
(R) is given by
M
1
,
1
()
M
2
,
2
()
M
1
M
2
,
1
(M
2
)
2
()
,
where M =
a+b
c+d
denotes the usual action of SL
2
(R) on H. The map
a b
c d
a b
c d
a b
c d
c +d
(1.1)
denes a locally isomorphic embedding of SL
2
(R) into Mp
2
(R).
Let Mp
2
(Z) be the inverse image of SL
2
(Z) under the covering map
Mp
2
(R) SL
2
(R). It is well known that Mp
2
(Z) is generated by
T =
1 1
0 1
, 1
,
S =
0 1
1 0
.
One has the relations S
2
= (ST)
3
= Z, where
Z =
1 0
0 1
, i
is the standard generator of the center of Mp

2
(Z). We will often use the
abbreviations
1
:= SL
2
(Z),
= (
1 n
0 1
) ; n Z
1
, and
:= 'T` =
1 n
0 1
, 1
; n Z
.
Throughout let L be an even lattice, i.e. a free Z-module of nite rank,
equipped with a symmetric Z-valued bilinear form (, ) such that the associ-
ated quadratic form
q(x) =
1
2
(x, x)
takes its values in Z. We assume that L is non-degenerated and denote its
signature by (b
+
, b
). We write
L
= x L Q; (x, y) Z for all y L

for the dual lattice. Then the quotient L
/L is a nite Abelian group, the so-

called discriminant group. The modulo 1 reduction of q(x) is a Q/Z-valued
quadratic form on L
/L whose associated bilinear form is the modulo 1 re-

duction of the bilinear form (, ) on L
.
Recall that there is a unitary representation
L
of Mp
2
(Z) on the group
algebra C[L
/L]. If we denote the standard basis of C[L
/L] by (e
)
L
/L
,
then
L
can be dened by the action of the generators S, T Mp
2
(Z) as
follows (cp. [Bo2]):
L
(T)e
= e(q())e
(1.2)
L
(S)e
i
b
b
+
[L
/L[
/L
e((, ))e
. (1.3)
This representation is essentially the Weil representation attached to the
quadratic module (L
/L, q) (cf. [No]). It factors through the nite group

1.1 The Weil representation 17
SL
2
(Z/NZ) if b
+
+ b
is even, and through a double cover of SL

2
(Z/NZ) if
b
+
+ b
is odd, where N is the smallest integer such that Nq() Z for all
L
. Note that
L
(Z)e
= i
b
b
+
e
. (1.4)
We write ', ` for the standard scalar product on C[L
/L], i.e.
/L
/L
/L
.
For , L
/L and (M, ) Mp
2
(Z) we dene the coecient
(M, ) of
the representation
L
by
(M, ) = '
L
(M, )e
, e
`.
Shintani proved the following explicit formula for
L
(cf. [Sn], Prop. 1.6).
/L and M =

a b
c d
SL
2
(Z). Then the
coecient
(

M) is given by
i
(b
b
+
)(1sgn(d))
,a
e(abq()), (1.5)
if c = 0, and by
i
(b
b
+
) sgn(c)
[c[
(b
+b
+
)/2
[L
/L[
rL/cL
e
a( +r, +r) 2(, +r) +d(, )

2c
,
(1.6)
if c = 0. Here,
,
denotes the Kronecker-delta.
Let
1
2
Z and f be a C[L
/L]-valued function on H. For (M, )

Mp
2
(Z) we dene the Petersson slash operator by
(f [
(M, )) () = ()
2
L
(M, )
1
f(M). (1.7)
As usual, by f f [
(M, ) an operation of Mp
2
(Z) on functions f : H
C[L
/L] is dened.
We denote by
L
the dual representation of
L
. If we think of
L
(M, )
((M, ) Mp
2
(Z)) as a matrix with entries in C, then
L
(M, ) is simply
the complex conjugate of
L
(M, ). We have a dual operation of Mp
2
(Z)
on functions f : H C[L
/L], given by
(f [
(M, )) () = ()
2
L
(M, )
1
f(M). (1.8)
Now assume that f : H C[L
/L] is a holomorphic function which is

invariant under the [
-operation of T Mp
2
(Z). We denote the components
of f by f
, so that f =

L
/L
e
. The invariance of f under T implies

that the functions e(q())f
() are periodic with period 1. Thus f has a

Fourier expansion
f() =
/L
nZq()
e
c(, n)e(n),
with Fourier coecients c(, n) =
1
0
f
()e(n) dx. Using the abbreviation

e
() := e
e() we may write

f() =
/L
nZq()
c(, n)e
(n). (1.9)
By means of the scalar product in C[L
/L] the coecients can be expressed

by
c(, n) =
1
0
'f(), e
(n )` dx. (1.10)
Denition 1.2. Let
1
2
Z. A function f : H C[L
/L] is called a modular

form of weight with respect to
L
and Mp
2
(Z) if
i) f [

2
(Z),
iii) f is holomorphic at the cusp .
Here condition (iii) means that f has a Fourier expansion of the form
f() =
/L
nZq()
n0
c(, n)e
(n).
Moreover, if all c(, n) with n = 0 vanish, then f is called a cusp form. The
C-vector space of modular forms of weight with respect to
L
and Mp
2
(Z)
is denoted by M
,L
, the subspace of cusp forms by S
,L
.
It is easily seen that M
,L
Example 1.3. Fix a positive integer t. Let L be the 1-dimensional lattice Z,
equipped with the positive denite quadratic form q(a) = ta
2
. Then L
/L
=
Z/2tZ, and M
1/2,L
is isomorphic to the space J
,t
of Jacobi forms of weight
and index t (cf. [EZ] Theorem 5.1). If L denotes the lattice Z with the
2
, then M
1/2,L
is isomorphic
to the space .
,t
of skew-holomorphic Jacobi forms of weight and index t
as dened in [Sk].
1.2 Poincare series and Eisenstein series
In this section we carry out the standard construction of Poincare and Eisen-
stein series for the space M
,L
. For simplicity we assume that (b
+
, b
) = (2, l),
(1, l 1), or (0, l 2) with l 3. Moreover, we assume that = 1 + l/2 (as
we will only need this case later).
1.2.1 Poincare series
Let L
/L and m Z q() with m > 0. Then e
(m) is a holomorphic
function H C[L
/L] which is invariant under the [
-operation of T
Mp
2
(Z). Moreover, by (1.4) we have
e
(m) [
Z = i
2
i
l2
e
(m) = e
(m).
Thus e
(m) is also invariant under the action of Z

2
.
We dene the Poincare series P
L
,m
of index (, m) by
P
L
,m
() =
1
2
(M,)
\ Mp
2
(Z)
e
(m) [
(M, ) (1.11)
(recall that

= 'T`). Since 5/2, the usual argument shows that

this series converges normally on H and thereby denes a Mp
2
(Z)-invariant
holomorphic function H C[L
/L]. We will often omit the superscript L, if

it is clear from the context.
Theorem 1.4. The Poincare series P
,m
P
,m
() =
/L
nZq()
n>0
p
,m
(, n)e
(n)
with
p
,m
(, n) =
m,n
(
,
+
,
)
+ 2
n
m
1
2

cZ{0}
H
c
(, m, , n)J
1
4
[c[
mn
. (1.12)
Here H
c
H
c
(, m, , n) =
e
i sgn(c)/2
[c[
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
,
(1.13)
and J
the Bessel function of the rst kind ([AbSt] Chap. 9). In particular
P
,m
S
,L
.
Recall the abbreviations
1
= SL
2
(Z) and
= (
1 n
0 1
) ; n Z. The
sum in (1.13) runs over all primitive residues d modulo c and

a b
c d
is a
`
1
/
with lower row (c d
) and
d
a b
c d
ma+nd
c
does not
depend on the choice of the coset representative.
The coecients
a b
c d
are universally bounded, because

L
factors
through a nite group. Hence there exists a constant C > 0 such that
H
c
(, m, , n) < C for all L
/L, n Z q(), and c Z 0.

This, together with the asymptotic behavior J
(t) = O(t
) (t 0) of the
J-Bessel function, implies that the series (1.12) converges absolutely. If we
also use the asymptotic property J
(t) = O(t
1/2
) as t , we nd that
p
,m
(, n) = O(n
1
) for n . (This is actually a very poor bound. For
even l Delignes theorem, previously the Ramanujan-Petersson conjecture,
states that p
,m
(, n) = O(n
(1)/2+
).)
Proof of Theorem 1.4. Let L
/L and n Z q(). According to (1.10)

the coecient p
,m
(, n) can be computed as
p
,m
(, n) =
1
2
1

(M,)
\ Mp
2
(Z)
e
(m) [
(M, ), e
(n )
dx.
Since e
(m) is invariant under the action of Z

2
, this can be written in the
form
1
0
'e
(m), e
(n )` dx +
1
0
'e
(m) [
Z, e
(n )` dx
+
1
a b
c d
\
1
c=0
e
(m) [
a b
c d
, e
(n )
dx
=
m,n
(
,
+
,
)
+
c=0
a b
c d
\
1
/
(c +d)
a b
c d
1
e
(m
a b
c d
), e
(n )
dx.
Since
L
is unitary we have
a b
c d
1
e
(m
a b
c d
), e
(n )
a b
c d
a b
c d
e(n),
and thereby
p
,m
(, n) =
m,n
(
,
+
,
)
+
c=0
a b
c d
\
1
/
a b
c d
(c +d)
a b
c d
dx.
(1.14)
We now use the elementary identities

c +d = sgn(c)
+d/c and
a b
c d
=
a
c

1
c
2
( +d/c)
. (1.15)
We nd that the latter sum in (1.14) equals
c=0
a b
c d
\
1
/
[c[
sgn(c)
a b
c d
( +d/c)
ma
c

m
c
2
( +d/c)
n
dx
=
c=0
a b
c d
\
1
/
[c[
sgn(c)
a b
c d
ma
c
m
c
2
n( d/c)
dx
=
c=0
[c[
sgn(c)
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
m
c
2
dx.
We substitute = iw in the integral and obtain for the above expression
2i
c=0
[c[
sgn(c)
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
1
2i
C+i
Ci
w
exp
2m
c
2
w
+ 2nw
dw,
where C is a positive real constant. We now use i
sgn(c)
= e
i sgn(c)/2
and the denition of the generalized Kloosterman sum (1.13) and nd
p
,m
(, n) =
m,n
(
,
+
,
) + 2
c=0
[c[
1
H
c
(, m, , n)
1
2i
C+i
Ci
w
exp
2m
c
2
w
+ 2nw
dw. (1.16)
If n 0, we consider the limit C and deduce that p
,m
(, n) = 0. If
n > 0, then the latter integral is an inverse Laplace transform that can be
found in any standard reference on integral transforms. According to [E2]
p. 245 (40) one has
1
2i
C+i
Ci
w
exp
2m
c
2
w
+ 2nw
dw =
c
2
n
m
1
2
J
1
4
[c[
mn
.
If we insert this into (1.16) we obtain the assertion. .
1.2.2 The Petersson scalar product
Let f, g be modular forms in M
,L
. Then it can be easily checked that
'f(), g()`y
is Mp
2
(Z)-invariant. (As before, ', ` denotes the scalar product on C[L
/L].)
If f or g is a cusp form, then 'f(), g()`y
is bounded on the standard

fundamental domain
T = = x +iy H; [x[ 1/2, [[ 1
for the action of
1
on H. (Hence the invariance property implies that
'f(), g()`y
is bounded on the whole upper half plane H.) We dene the

Petersson scalar product of f and g by
(f, g) =
F
'f(), g()`y
dxdy
y
2
. (1.17)
It is easily seen that the integral does not depend on the choice of the fun-
damental domain. The space S
,L
endowed with the scalar product (, ) is a
nite dimensional Hilbert space.
The Petersson coecient formula for S
,L
has the following form.
/L and m Z q() with m > 0. Let f be a

cusp form in S
,L
and denote its Fourier coecients by c(, n) (as in (1.9)).
Then the scalar product of f with the Poincare series P
,m
is given by
(f, P
,m
) = 2
( 1)
(4m)
1
c(, m).
Proof. Let
1
=
1
/1. We use the usual unfolding argument and nd
(f, P
,m
) =
1
2
1
\H
f(),
(M,)
\ Mp
2
(Z)
e
(m) [
(M, )
dxdy
y
2
=
1
\H
\
1
'f(M), e
(mM)`(M)
dxdy
y
2
= 2
\H
'f(), e
(m)`y
2
dxdy.
We insert the Fourier expansion of f and infer:
(f, P
,m
) = 2
0
1
nZq()
n>0
c(, n)e(n)e(mx +miy)y
2
dxdy
= 2
y=0
c(, m)e
4my
y
2
dy
= 2
( 1)
(4m)
1
c(, m).
.
As a corollary we obtain that the Poincare series P
,m
generate the space
S
,L
.
1.2.3 Eisenstein series
Let L
/L with q() Z. Then the vector e
= e
(0) C[L
/L], consid-
ered as a constant function H C[L
/L], is invariant under the [
-action of
T, Z
2
Mp
2
(Z). The Eisenstein series
E
L
() =
1
2
(M,)
\ Mp
2
(Z)
e
(M, ) (1.18)
converges normally on H and therefore denes a Mp
2
(Z)-invariant holomor-
phic function on H.
Theorem 1.6. The Eisenstein series E

E
() =
/L
nZq()
n0
q
(, n)e
(n)
with
q
(, n) =
,
+
,
, if n = 0,
(2)
n
1
()
cZ{0}
[c[
1
H
c
(, 0, , n), if n > 0.
(1.19)
Here, H
c
(, 0, , n) denotes the generalized Kloosterman sum dened in
(1.13). In particular E
is an element of M
,L
.
Proof. We proceed in the same way as in the proof of Theorem 1.4. Let
L
/L and n Z q(). By (1.10) we have

q
(, n) =
1
2
1

(M,)
\ Mp
2
(Z)
e
(M, ), e
(n )
dx.
Because e
is invariant under the action of Z

2
, this equals
0,n
(
,
+
,
) +
c=0
a b
c d
\
1
/
(c +d)
a b
c d
1
e
, e
(n )
dx
=
0,n
(
,
+
,
) +
c=0
a b
c d
\
1
/
a b
c d
(c +d)
e(n) dx.
We now evaluate the integral. Since

c +d = sgn(c)
+d/c, we have
(c +d)
e(n) dx = [c[
sgn(c)
( +d/c)
e(n) dx
= [c[
sgn(c)
nd
c
e(n) dx.
We substitute = iw and nd
(c +d)
e(n) dx
= 2i
[c[
sgn(c)
nd
c
1
2i
C+i
Ci
w
e
2nw
dw,
where C is a positive real constant. If n 0, one can consider the limit C
to infer that the integral is 0. It follows that q
(, n) =
0,n
(
,
+
,
)
in this case. Now let n > 0. Then the integral is an inverse Laplace transform.
According to [E2] p. 238 (1) we have
1
2i
C+i
Ci
w
e
2nw
dw =
(2n)
1
()
.
(In fact, one can deform the path of integration and essentially obtains the
Hankel integral for 1/().) Hence we get
q
(, n) =
(2)
n
1
()
c=0
a b
c d
\
1
/
[c[
sgn(c)
a b
c d
nd
c
.
If we use i
sgn(c)
= e
i sgn(c)/2
and substitute in the denition of
H
c
(, 0, , n), we obtain the assertion. .
For computational purposes the formula (1.19) for the coecients q
(, n)
is not very useful. An explicit formula, involving special values of Dirichlet
L-series and nitely many representation numbers modulo prime powers at-
tached to the lattice L, is derived in [BK].
To give an example we briey consider the case that L is unimodular. Then
L
/L is trivial and M
,L
is just the space of elliptic modular forms of weight
1 + l/2. We have one Eisenstein series E with Fourier coecients q(n) and
q(0) = 2. For n > 0 the above expression for the coecients can be simplied
as follows. First note that l 2 (mod 8). This implies e
i sgn(c)/2
= 1
and
H
c
(, 0, , n) =
1
[c[
d(c)
nd
c
=
1
[c[
d|(c,n)
d>0
([c[/d)d.
Here we have used the evaluation of the Ramanujan sum

d(c)
e(
nd
c
) by
means of the Moebius function . Thus we nd
q(n) = 2
(2)
n
1
()
c=1
c
d|(c,n)
d>0
(c/d)d
= 2
(2)
n
1
()()
d|n
d
1
= 2
(2)
()()
1
(n),
where
1
(n) denotes the sum of the (1)-th powers of the positive divisors
of n and (s) the Riemann zeta function. Using () =
(2i)
2!
B
with the
-th Bernoulli number B
, and the fact that 2 (mod 4), we get

q(n) =
4
B
1
(n).
Thus in this case the Eisenstein series E(z) is the classical Eisenstein series
for
1
, normalized such that its constant term equals 2.
Let us now turn back to the general case and determine the number of
linearly independent Eisenstein series E
. The [
-invariance of E
under Z
and the identity e
Z = e
imply
E
= E
Z =
1
2
(M,)
\ Mp
2
(Z)
e
Z [
(M, ) = E
.
Thus the space M
Eis
,L
spanned by all Eisenstein series E
equals the space

spanned by the E
where runs through a set of representatives of

L
/L; q() Z modulo the action of 1. Comparing constant terms one

nds that the set of Eisenstein series E
with as above is already linearly

independent. We obtain
dim(M
Eis
,L
) = # L
/L; q() Z, 2 = 0 +L
+
1
2
# L
/L; q() Z, 2 = 0 +L. (1.20)

Now let f M
,L
and write the Fourier expansion of f in the form
f() =
/L
nZq()
c(, n)e
(n).
The invariance of f under Z and (1.4) imply that c(, n) = c(, n). Hence
f
1
2
/L
q()Z
c(, 0)E

lies in S
,L
. This shows that M
,L
= M
Eis
,L
S
,L
and (1.20) is a formula for
the codimension of S
,L
in M
,L
.
In later applications we will mainly be interested in the Eisenstein series
E
0
() which we simply denote by E(). In the same way we write q(, n) for
the Fourier coecients q
0
(, n) of E().
Let us nally cite the following result of [BK]:
Proposition 1.7. The coecients q(, n) of E() are rational numbers.
1.3 Non-holomorphic Poincare series of negative weight
As in the previous section we assume that L is an even lattice of signature
(b
+
, b
) = (2, l), (1, l 1), or (0, l 2) with l 3. Put k = 1 l/2 and

= 1 + l/2. We now construct certain vector valued Maass-Poincare series
for Mp
2
(Z) of weight k. Series of a similar type are well known and appear
in many places in the literature (see for instance [He, Ni, Fa]).
Let M
,
(z) and W
,
(z) be the usual Whittaker functions as dened
in [AbSt] Chap. 13 p. 190 or [E1] Vol. I Chap. 6 p. 264. They are linearly
independent solutions of the Whittaker dierential equation
d
2
w
dz
2
+
1
4
+

z

2
1/4
z
2
w = 0. (1.21)
The functions M
,
(z) and M
,
(z) are related by the identity
W
,
(z) =
(2)
(
1
2
)
M
,
(z) +
(2)
(
1
2
+ )
M
,
(z) (1.22)
([AbSt] p. 190 (13.1.34)). This implies in particular W
,
(z) = W
,
(z). As
z 0 one has the asymptotic behavior
M
,
(z) z
+1/2
( /
1
2
N), (1.23)
W
,
(z)
(2)
( + 1/2)
z
+1/2
( 1/2). (1.24)
If y R and y one has
M
,
(y) =
(1 + 2)
( + 1/2)
e
y/2
y
(1 +O(y
1
)), (1.25)
W
,
(y) = e
y/2
y
(1 +O(y
1
)). (1.26)
For convenience we put for s C and y R
>0
:
s
(y) = y
k/2
M
k/2, s1/2
(y). (1.27)
In the same way we dene for s C and y R 0:
)
s
(y) = [y[
k/2
W
k/2 sgn(y), s1/2
([y[). (1.28)
If y < 0 then equation (1.22) implies that
s
([y[) =
(1 +k/2 s)
(1 2s)
)
s
(y)
(1 +k/2 s)(2s 1)
(1 2s)(s +k/2)

1s
([y[).
(1.29)
The functions
s
(y) and )
s
(y) are holomorphic in s. Later we will be
interested in certain special s-values. For y > 0 we have
k/2
(y) = y
k/2
M
k/2, k/21/2
(y) = e
y/2
, (1.30)
)
1k/2
(y) = y
k/2
W
k/2, 1/2k/2
(y) = e
y/2
. (1.31)
(1/2 +)W
,
(z) = e
z/2
z
+1/2
0
e
tz
t
1/2+
(1 +t)
1/2++
dt
(1() > 1/2, 1(z) > 0) of the W-Whittaker function ([E1] Vol. I p. 274
(18)), we nd for y < 0:
)
1k/2
(y) = [y[
k/2
W
k/2, 1/2k/2
([y[)
= e
|y|/2
[y[
1k
0
e
t|y|
(1 +t)
k
dt
= e
|y|/2
[y[
1k
1
e
t|y|
t
k
dt.
If we insert the denition of the incomplete Gamma function (cf. [AbSt] p. 81)
(a, x) =
x
e
t
t
a1
dt, (1.32)
we obtain for y < 0 the identity
)
1k/2
(y) = e
y/2
(1 k, [y[). (1.33)
The usual Laplace operator of weight k (cf. [Ma1])
k
= y
2

2
x
2
+

2
y
2
+iky

x
+i

y
(1.34)
acts on smooth functions f : H C[L
/L] component-wise. Let L
/L
and m Z+q() with m < 0. Then it can be easily checked that the function
s
(4[m[y)e
(mx)
is invariant under the [
k
-operation of T Mp
2
(Z) (as dened in (1.7)) and
an eigenfunction of
k
with eigenvalue
s(1 s) + (k
2
2k)/4.
Denition 1.8. With the above notation we dene the Poincare series F
L
,m
of index (, m) by
F
L
,m
(, s) =
1
2(2s)
(M,)
\ Mp
2
(Z)
[
s
(4[m[y)e
(mx)] [
k
(M, ),
(1.35)
where = x + iy H and s = + it C with > 1. If it is clear from the
context, to which lattice L the series (1.35) refers, we simply write F
,m
(, s).
By (1.23) and (1.27) the above series (1.35) has the local majorant
(M,)
\ Mp
2
(Z)
(M)
.
Hence (1.35) converges normally for H, s C and > 1 and thereby
denes a Mp
2
(Z)-invariant function on H. As in [Ma1] it can be seen that
k
commutes with the action of Mp
2
(Z), i.e. that
k
(f [
k
(M, )) = (
k
f) [
k
(M, )
for any smooth function f : H C[L
/L] and any (M, ) Mp

2
(Z). We may
infer that F
,m
is an eigenfunction of
k
:
k
F
,m
=
s(1 s) + (k
2
2k)/4
F
,m
.
(Thus F
,m
is automatically real analytic as a solution of an elliptic dier-
ential equation.) The invariance of F
,m
under the action of Z Mp
2
(Z)
implies that F
,m
= F
,m
.
Let f : H C[L
/L] be a continuous function with the property f [

k
T =
f. By virtue of the same argument as in section 1.1 one nds that f has a
Fourier expansion of the form
f() =
/L
nZ+q()
c(, n, y)e
(nx), (1.36)
where the coecients are given by
c(, n, y) =
1
0
'f(), e
(nx)` dx. (1.37)

Theorem 1.9. The Poincare series F
,m
F
,m
(, s) =
(1 +k/2 s)
(2 2s)(s +k/2)
1s
(4[m[y)(e
(mx) + e
(mx))
+
/L
q()Z
b(, 0, s)y
1sk/2
e
/L
nZ+q()
n>0
b(, n, s))
s
(4ny)e
(nx) +

F
,m
(, s),
where

F
,m
(, s) is the T-invariant function
F
,m
(, s) =
/L
nZ+q()
n<0
b(, n, s))
s
(4ny)e
(nx),
and the Fourier coecients b(, n, s) are given by
2[n/m[
k1
2
(s +k/2)
cZ{0}
H
c
(, m, , n)I
2s1
4
[c[
[mn[
, if n > 0,
4
1k/2
1+sk/2
[m[
sk/2
(2s 1)(s +k/2)(s k/2)
cZ{0}
[c[
12s
H
c
(, m, , 0), if n = 0,
(1 +k/2 s)
(2s)(1 2s)
m,n
(
,
+
,
)
+
2[n/m[
k1
2
(s k/2)
cZ{0}
H
c
(, m, , n)J
2s1
4
[c[
[mn[
, if n < 0.
Here H
c
H
c
(, m, , n) =
e
i sgn(c)k/2
[c[
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
(1.38)
and J
(z), I
(z) the usual Bessel functions as dened in [AbSt] Chap. 9.

The sum in (1.38) runs over all primitive residues d modulo c and

a b
c d
is a
`
1
/
with lower row (c d
) and
d

1
a b
c d
ma+nd
c
does not
depend on the choice of the coset representative. The fact that H
c
(, m, , n)
is universally bounded implies that the series for the coecients b(, n, s)
converge normally in s for > 1.
Proof of Theorem 1.9. We split the sum in equation (1.35) into the sum over
1, Z, Z
2
, Z
3
` Mp
2
(Z) and the sum over (M, )

` Mp
2
(Z), where
M =

a b
c d
with c = 0. The latter sum will be denoted by H(, s). Since

e
(mx) [
k
Z = e
(mx), we nd that the rst term equals

1
(2s)
s
(4[m[y)(e
(mx) + e
(mx)).
According to (1.29) this can be rewritten as
(1 +k/2 s)
(2 2s)(s +k/2)
1s
(4[m[y)(e
(mx) + e
(mx))
+
(1 +k/2 s)
(2s)(1 2s)
)
s
(4my)(e
(mx) + e
(mx)). (1.39)
We now calculate the Fourier expansion of H(, s). Since e
(mx) is in-
variant under the action of Z
2
, we may write H(, s) in the form
1
(2s)
a b
c d
\
1
c=0
[
s
(4[m[y)e
(mx)] [
k
a b
c d
.
Let L
/L and n Z+q() and denote the (, n)-th Fourier coecient of

H(, s) by c(, n, y) (cp. (1.36)). Then we have
c(, n, y) =
1
0
'H(, s), e
(nx)` dx
=
c=0
a b
c d
\
1
/
(c +d)
k
(2s)

s
4[m[
y
|c+d|
2
1
L
a b
c d
m1
a b
c d
, e
(nx)
dx.
By (1.15) we nd that
1
L
a b
c d
m1
a b
c d
, e
(nx)
=
1
a b
c d
ma
c
1
m
c
2
( +d/c)
e(nx),
and therefore
c(, n, y) =
1
(2s)
c=0
a b
c d
\
1
/
a b
c d
ma
c
(c +d)
k
4[m[
y
[c +d[
2
m
c
2
( +d/c)
nx
dx.
If we use

c +d = sgn(c)
+d/c and substitute x by x d/c in the

integral, we obtain
c(, n, y) =
1
(2s)
c=0
a b
c d
\
1
/
a b
c d
ma +nd
c
[c[
k
sgn(c)
k
4[m[
y
c
2
[[
2
mx
c
2
[[
2
nx
dx
=
1
(2s)
c=0
[c[
1k
i
k
H
c
(, m, , n)
4[m[
y
c
2
[[
2
mx
c
2
[[
2
nx
dx.
By the denition of
s
this equals
(4[m[y)
k/2
(2s)
c=0
[c[H
c
(, m, , n)
k/2
M
k/2, s1/2
4[m[
y
c
2
[[
2
mx
c
2
[[
2
nx
dx
=
(4[m[y)
k/2
(2s)
c=0
[c[H
c
(, m, , n)
y ix
y +ix
k/2
M
k/2, s1/2
4[m[y
c
2
(x
2
+y
2
)
mx
c
2
(x
2
+y
2
)
nx
dx. (1.40)
We abbreviate the latter integral by I and substitute x = yu. Then I is equal
to
y
1 iu
1 +iu
k/2
M
k/2, s1/2
4[m[
c
2
y(u
2
+ 1)
[m[u
c
2
y(u
2
+ 1)
nyu
du;
and if we set A = ny and B =
|m|
c
2
y
, we get
I = y
1 iu
1 +iu
k/2
M
k/2, s1/2
4B
u
2
+ 1
exp
2iBu
u
2
+ 1
+ 2iAu
du.
The integral I/(y(2s)) was evaluated by Hejhal in [He] p. 357. (Notice that
M
k/2, s1/2
(y)/(2s) is the same as Hejhals function G(y) with = k/2.
Moreover, W
,
(y) equals e
y/2
y
+1/2
( + 1/2 , 1 + 2, y) in Hejhals
notation.) We nd
I
y(2s)
=
[B/A[
(s k/2)
W
k/2, s1/2
(4[A[)J
2s1
[AB[
, if A > 0,
4
1+s
(2s 1)(s +k/2)(s k/2)
B
s
, if A = 0,
2
[B/A[
(s +k/2)
W
k/2, s1/2
(4[A[)I
2s1
[AB[
, if A < 0.
We resubstitute for A and B and obtain
[c[I
(2s)
=
[m/n[
(s k/2)
(4[n[y)
k/2
J
2s1
4
[c[
[mn[
)
s
(4ny), if n < 0,
4
1+s
[m[
s
[c[
12s
y
1s
(2s 1)(s +k/2)(s k/2)
, if n = 0,
2
[m/n[
(s +k/2)
(4[n[y)
k/2
I
2s1
4
[c[
[mn[
)
s
(4ny), if n > 0.
If we insert this into (1.40) we get c(, n, y) is given by
2[n/m[
k1
2
(s k/2)
c=0
H
c
(, m, , n)J
2s1
4
[c[
[mn[
)
s
(4ny), n < 0,
4
1k/2
1+sk/2
[m[
sk/2
(2s 1)(s +k/2)(s k/2)
c=0
[c[
12s
H
c
(, m, , 0)y
1sk/2
, n = 0,
2[n/m[
k1
2
(s +k/2)
c=0
H
c
(, m, , n)I
2s1
4
[c[
[mn[
)
s
(4ny), n > 0.
Combining this with (1.39) we obtain the assertion. .
Since
k
F
,m
(, 1 k/2) = 0, the Poincare series F
,m
(, s) are in par-
ticular interesting in the special case s = 1 k/2.
Proposition 1.10. For s = 1k/2 the Fourier expansion of F
,m
(, s) given
in Theorem 1.9 can be simplied as follows:
F
,m
(, 1 k/2) = e
(m) + e
(m) +
/L
nZ+q()
n0
b(, n, 1 k/2)e
(n)
+

F
,m
(, 1 k/2).
The function

F
,m
(, 1 k/2) has the expansion
F
,m
(, 1 k/2) =
/L
nZ+q()
n<0
b(, n, 1 k/2))
1k/2
(4ny)e
(nx),
and the Fourier coecients b(, n, 1 k/2) are given by
n
m
k1
2

cZ{0}
H
c
(, m, , n)I
1k
4
[c[
[mn[
, if n > 0,
(2)
2k
[m[
1k
(2 k)
cZ{0}
[c[
k1
H
c
(, m, , 0), if n = 0,
1
(1 k)
m,n
(
,
+
,
)
+
2
(1 k)
n
m
k1
2

cZ{0}
H
c
(, m, , n)J
1k
4
[c[
[mn[
, if n < 0.
Proof. By virtue of (1.30) and (1.31) this immediately follows from Theorem
1.9. .
Denition 1.11. A function f : H C[L
/L] is called a nearly holomorphic

modular form of weight k (with respect to
L
and Mp
2
(Z)), if
i) f [
k
2
(Z),
iii) f has a pole in , i.e. f has a Fourier expansion of the form
f() =
/L
nZ+q()
n
c(, n)e
(n).
The space of these nearly holomorphic modular forms is denoted by M
!
k,L
.
The Fourier polynomial
/L
nZ+q()
n<0
c(, n)e
(n)
is called the principal part of f.
The following proposition is an analogue of the theorem in [He] on p. 660
(see also [Ni] Theorem 6).
Proposition 1.12. Let f() be a nearly holomorphic modular form of weight
k and denote its Fourier coecients by c(, n) ( L
/L, n Z+q()). Then

f() =
1
2
/L
nZ+q()
n<0
c(, n)F
,n
(, 1 k/2).
Proof. Consider the function
g() = f()
1
2
/L
nZ+q()
n<0
c(, n)F
,n
(, 1 k/2).
The invariance of f under Z Mp
2
(Z) implies that c(, n) = c(, n).
According to Proposition 1.10 we have F
,n
(, 1k/2) = e
(n) +e
(n) +
O(1) for all , n. Hence g() is bounded as y . But g also satises
g [
k
(M, ) = g ((M, ) Mp
2
(Z)),
k
g = 0.
Thus g has to vanish identically (see [He] Chap. 9 Prop. 5.14c). .

Then
H
c
(, m, , n) = H
c
(, m, , n).
Proof. From Proposition 1.1 it can be easily deduced that
a b
c d
a b
c d
and
1
a b
c d
=
1
a b
c d
.
Now the assertion immediately follows from the denition of H
c
(, m, , n).
.
Remark 1.14. This implies that the Fourier coecients b(, n, 1k/2) of the
function F
,m
(, 1 k/2) are real numbers.
We now compare the Fourier coecients of F
,m
(, 1 k/2) with the
coecients of the holomorphic Poincare series P
,n
S
,L
constructed in the
previous section.

Then we have
H
c
(, m, , n) = H
c
(, n, , m).
Proof. We use the fact that (

M)
1
=

M
1
for any M
1
, and that =
2 k. We nd
H
c
(, m, , n) =
e
i sgn(c)k/2
[c[
d(c)
a b
c d
\
1
/
a b
c d
ma +nd
c
=
e
i sgn(c)(2k)/2
[c[
d(c)
a b
c d
\
1
/
d b
c a
md +na
c
=
e
i sgn(c)/2
[c[
d(c)
a b
c d
\
1
/
a b
c d
na md
c
= H
c
(, n, , m).
.
/L, n Z+q(), m Z+q(), and m < 0.

Denote the (n, )-th Fourier coecient b(, n, 1 k/2) of F
,m
(, 1 k/2)
(see Proposition 1.10) by b
,m
(, n). If n < 0, then
b
,m
(, n) =
1
(1 k)
p
,n
(, m),
where p
,n
(, m) is the (, m)-th coecient of the Poincare series
P
,n
() S
,L
(see 1.12). If n = 0, then
b
,m
(, 0) = q
(, m),
where q
(, m) is the (, m)-th coecient of the Eisenstein series E
()
M
,L
(see 1.19).
Proof. We compare the formulas for the Fourier coecients given in Propo-
sition 1.10 and Theorem 1.4 (resp. Proposition 1.10 and Theorem 1.6) and
apply Lemma 1.15. Note that k 1 = 1 . .
For L
/L and n Zq() with n > 0 let a

,n
: S
,L
C denote the
functional in the dual space S
,L
of S
,L
which maps a cusp form f to its
(, n)-th Fourier coecient a
,n
(f). According to Proposition 1.5, a
,n
can
be described by means of the Petersson scalar product as
a
,n
=
(4n)
1
2( 1)
( , P
,n
). (1.41)
The following theorem can also be proved using Serre duality (see [Bo3]
Thm. 3.1). We deduce it from Proposition 1.16.
Theorem 1.17. There exists a nearly holomorphic modular form f M
!
k,L
with prescribed principal part
/L
mZ+q()
m<0
c(, m)e
(m)
(c(, m) C with c(, m) = c(, m)), if and only if the functional
/L
mZ+q()
m<0
c(, m)a
,m
,
equals 0 in S
,L
.
Proof. First, suppose that there exists such a nearly holomorphic modular
form f. Then by Proposition 1.12
f() =
1
2
/L
mZ+q()
m<0
c(, m)F
,m
(, 1 k/2),
and in particular
/L
mZ+q()
m<0
c(, m)
F
,m
(, 1 k/2) = 0.
Since

)
1k/2
(4ny)e
(nx) = 0,
this implies that
/L
mZ+q()
m<0
c(, m)b
,m
(, n) = 0
for all L
/L and n Z + q() with n < 0. (Here we have used the same

notation as in Proposition 1.16.) Applying Proposition 1.16 we nd that
/L
mZ+q()
m<0
c(, m)p
,n
(, m) =
/L
mZ+q()
m<0
c(, m)a
,m
(P
,n
)
= 0
for all L
/L and n Z+q() with n < 0. Since the Poincare series P

,n
generate the space S
,L
we obtain
/L
mZ+q()
m<0
c(, m)a
,m
= 0. (1.42)
Now assume that (1.42) holds. Then we may reverse the above argument
to infer that
/L
mZ+q()
m<0
c(, m)
F
,m
(, 1 k/2) = 0.
According to Proposition 1.10 this implies that
1
2
/L
mZ+q()
m<0
c(, m)F
,m
(, 1 k/2)
is a nearly holomorphic modular form with the right principal part. .
2 The regularized theta lift
In this chapter we consider a certain regularized theta lift of the Maass-
Poincare series studied in the previous chapter.
2.1 Siegel theta functions
We use the same notation as in section 1.1. In particular L denotes an even
lattice of signature (b
+
, b
). The bilinear form on L induces a symmetric

bilinear form on V := LR. We write Gr(L) for the Grassmannian of L, this
is the real analytic manifold of b
+
-dimensional positive denite subspaces of
V . (For b
+
= 0 or b
= 0 the Grassmannian is simply a point.) If v Gr(L),

we write v
for the orthogonal complement of v in V such that V = v v
.
For any vector x V let x
v
resp. x
v
be the orthogonal projection of x to v
resp. v
. Then q(x) =
1
2
(x, x) = q(x
v
) +q(x
v
).
The Siegel theta function attached to L is dened by
L
(, v) =
L
e (q(
v
) + q(
v
)) ( H, v Gr(L)).
Following Borcherds we introduce a more general theta function. Let r, t V ,
L
/L, H and v Gr(L). Dene
(, v; r, t) =
+L
e (q(( +t)
v
) + q(( +t)
v
) ( +t/2, r)) (2.1)
and
L
(, v; r, t) =
/L
e
(, v; r, t). (2.2)
If r and t are both 0 we will omit them and simply write
(, v) resp.
L
(, v).
Let be the lattice Z
2
equipped with the standard symplectic form and
W = R (such that Sp() = SL
2
(Z)). Then W V is a symplectic vec-
tor space. The usual theta function attached to the lattice L W V
restricted to the symmetric subspace H Gr(L) Sp(W V )/K (where
K denotes a maximal compact subgroup) equals the Siegel theta function
L
(, v).
40 2 The regularized theta lift
It is well known that
L
(, v; r, t) is a real analytic function in (, v)
HGr(L).
Let
O(V ) = g SL(V ); (gx, gy) = (x, y) for all x, y V
be the (special) orthogonal group of V and
O(L) = g O(V ); gL = L .
be the orthogonal group of L. We denote by O
d
(L) the discriminant kernel of
O(L). This is the subgroup of nite index of O(L) consisting of all elements
which act trivially on the discriminant group L
/L. Observe that O

d
(L) is
functorial in the following sense: If

L L is a sublattice, then O
d
(
L) O
d
(L).
As a function in v Gr(L) the theta function
L
(, v) is obviously in-
variant under the action of O
d
(L). As a function in H it transforms
like an elliptic modular form under the action of Mp
2
(Z). Using the Poisson
summation formula the following transformation law can be established (see
[Bo2] Theorem 4.1).
Theorem 2.1. If (M, ) Mp
2
(Z) and M =

a b
c d
, then
L
(, v; r, t) sat-
ises
L
(M, v; ar +bt, cr +dt) = ()
b
+
()
b
L
(M, )
L
(, v; r, t).
Borcherds showed that
L
can be written as a Poincare series involving
similar theta series
K
which come from a smaller lattice K of signature
(b
+
1, b
1) (cf. [Bo2] section 5). This is very useful for the evaluation of
theta integrals

F
'F(),
L
(, v)` y
b
+
/2
dxdy
y
2
by the usual unfolding trick (Rankin-Selberg method).
In the rest of this section we review Borcherds construction in a slightly
more explicit way. First we need some basic facts about lattices.
Let z L be a primitive norm 0 vector (i.e. Qz L = Zz and q(z) = 0).
Then there exists a z
with (z, z
) = 1. Let N be the unique positive

integer with (z, L) = NZ. Then we have z/N L
. Denote by K the lattice

K = L z
. (2.3)
Over Q we have
L Q = (K Q) (Qz Qz
). (2.4)
2.1 Siegel theta functions 41
Hence K has signature (b
+
1, b
1). If n V = L R then we write n

K
for the orthogonal projection of n to KR. It can be easily checked that n
K
can be computed as
n
K
= n (n, z)z
+ (n, z)(z
, z
)z (n, z
)z.
If n L
then n
K
lies in the dual lattice K
of K. (Warning: Notice that K
is
not necessarily contained in L
. Thus if n L, then n
K
does not necessarily
lie in K.)
Let L be a lattice vector with (, z) = N. It can be uniquely repre-
sented as
=
K
+Nz
+Bz (2.5)
with
K
K
and B Q.
Proposition 2.2. The lattice L can be written as L = K Z Zz.
Proof. Let n L. Then the vector
n = n (n, z/N) (n, z
)z + (n, z/N)(, z
)z
lies in L and an easy calculation shows that n z and n z
. Hence n K
and therefore n K +Z +Zz. The sum is obviously direct. .
If one uses the fact that [L
/L[ is given by the absolute value of the

Gram determinant of L, then the above proposition in particular shows that
[L
/L[ = N
2
[K
/K[.
Moreover, we may infer that (, )z/N denes an isometric
embedding K
. The kernel of the induced map K
/L equals
K; (, ) NZ.
Consider the sub-lattice
L
0
= L
; (, z) 0 (mod N) (2.6)
of L
. Obviously L is contained in L
0
. Beside the orthogonal projection from
L
0
to K
there is another natural projection L
0
K
: If we combine the
linear map L
0
L
,
(,z)
N
with the orthogonal projection from
L
to K
, we get a projection
p : L
0
K
, p() =
K

(, z)
N

K
. (2.7)
This map has the property that p(L) = K. In fact, if = k + a + bz L
with k K and a, b Z, then it is easily seen that p() = k. Thus p induces
a surjective map L
0
/L K
/K which will also be denoted by p. Note that

L
0
/L = L
/L; (, z) 0 (mod N).

We introduce some more notation. If x V then we sometimes sim-
ply write x
2
instead of (x, x) or 2q(x). Furthermore, we abbreviate [x[ :=
[(x, x)[
1/2
. Let v Gr(L). We denote by w the orthogonal complement of z
v
in v and by w
the orthogonal complement of z

v
in v
. Hence one has the

orthogonal decomposition
V = v v
= w Rz
v
w
Rz
v
.
If x V , we write x
w
resp. x
w
for the orthogonal projection of x to w
resp. w
. It can be easily veried that w and w
are contained in (KR)Rz.

This implies that the orthogonal projection V KR induces an isometric
isomorphism ww
w
K
w
K
= KR. In particular we may identify w
with w
K
and consider w as an element of the Grassmannian Gr(K).
We denote by the vector
= z
+
z
v
2z
2
v
+
z
v
2z
2
v
(2.8)
in V z
= (K R) Rz.
Lemma 2.3. (cp. [Bo2] Lemma 5.1.) Let L
/L. We have
(, v) =
1
2yz
2
v
+KZ
dZ
e (q(
w
) + q(
w
))
e
[(, z) +d[
2
4iyz
2
v
d(, z
v
z
v
)
2z
2
v
.
Proof. For +K Z and d R we dene a function
g(, v; d) = e (q(( +dz)
v
) + q(( +dz)
v
)) .
Then
(, v) =
+KZ
dZ
g(, v; d).
We may apply the Poisson summation formula to rewrite the inner sum. We
nd
(, v) =
+KZ
dZ
g(, v; d),
where g(, v; d) denotes the partial Fourier transform with respect to the
variable d. We now compute g(, v; d). Using q(z
v
) +q(z
v
) = 0 we get
g(, v; d) = e
d
2
( )q(z
v
) +d((, z
v
) + (, z
v
)) +q(
v
) + q(
v
)
= e(Ad
2
+Bd +C),
with A = ( )q(z
v
), B = ((, z
v
) + (, z
v
)), and C = q(
v
) + q(
v
).
The Fourier transform of e(Ax
2
+Bx+C) equals (
2A
i
)
1/2
e
(x+B)
2
4A
+C
(see for instance [Bo2] Cor. 3.3). Hence we nd that g(, v; d) is equal to
1
2yz
2
v
e
d
2
2d((, z
v
) + (, z
v
)) ((, z
v
) + (, z
v
))
2
2( )z
2
v
e (q(
v
) + q(
v
)) .
Using q(
v
) = q(
w
) + (, z
v
)
2
/2z
2
v
and q(
v
) = q(
w
) + (, z
v
)
2
/2z
2
v
we obtain
g(, v; d) =
1
2yz
2
v
e
(, z
v
)
2
(, z
v
)
2
2z
2
v
+q(
w
) + q(
w
)
d
2
2d((, z
v
) + (, z
v
)) ((, z
v
) + (, z
v
))
2
2( )z
2
v
.
(2.9)
Since
2d((, z
v
) + (, z
v
)) = d [(, z
v
z
v
)( ) + (, z)( + )] ,
the right hand side of (2.9) can be written as
1
2yz
2
v
e
q(
w
) + q(
w
)
d(, z
v
z
v
)
2z
2
v
d
2
+d(, z) +d(, z) + (, z)
2

2( )z
2
v
,
which is equal to
1
2yz
2
v
e
q(
w
) + q(
w
)
d(, z
v
z
v
)
2z
2
v
[d + (, z)[
2
4iyz
2
v
.
This implies the assertion. .
Theorem 2.4. (cp. [Bo2] Theorem 5.2.) Let L
. We have
L+
(, v) =
1
2yz
2
v
c,dZ
c(,z) (N)
e
[c +d[
2
4iyz
2
v
d(, z
) +cdq(z
K+p(cz
)
(, w; d, c),
where p denotes the projection L
0
K
dened in (2.7).
Remark 2.5. Regarding the denition (2.1) of
K
we should write more cor-
rectly
K
instead of (note that
K
= (, z
)z). Since
w
= (
K
)
w
= z
w
and (, z) = (
K
, z), we permit ourselves this abuse of notation.
Proof. We want to apply Lemma 2.3. Write =
K
+a
+b
z with
K
K
,
a
= (, z) Z and b
= (, z
) (, z)z
2
Q and observe that every
+K Z can be uniquely written as
=
K
+cz
+f(c, )z
with c Z, c a
(mod N), and

K
K+
K
+
K
(ca
)/N. The function

f(c, ) is dened by f(c, ) = b
+(ca
)B/N, and
K
, B are given by (2.5).
It can be easily checked that
K
+
K
(c a
)/N = p( cz
).
We nd that
L+
(, v) is given by
1
2yz
2
v
c,dZ
c(,z) (N)
K
K+p(cz
)
e (q((
K
+cz
)
w
) + q((
K
+cz
)
w
))
e
[(
K
+cz
, z) +d[
2
4iyz
2
v
d(
K
+cz
+f(c, )z, z
v
z
v
)
2z
2
v
,
where we have used z
w
= z
w
= 0 and q(z) = 0. Regarding the denition
(2.1) of
K
(, w; d, c), we see that we have to prove
d(
K
+cz
+f(c, )z, z
v
z
v
)
2z
2
v
(
K
c/2, d) d(, z
) +cdq(z
)
modulo 1. Since (z, z
v
z
v
)/2z
2
v
= 1, we nd
d
(f(c, )z, z
v
z
v
)
2z
2
v
= df(c, )
= db
+
(, z) c
N
dB
= d(, z
) +d(, z)z
2
+
(, z) c
N
dB.
From (, z
) Z it follows that B Nz
2
(mod 1), and this implies that
d
(f(c, )z, z
v
z
v
)
2z
2
v
d(, z
) +cdz
2
(mod 1).
Thus it suces to show
d
(
K
+cz
, z
v
z
v
)
2z
2
v
= (
K
c/2, d) cdq(z
). (2.10)
The left hand side of (2.10) equals
d
K
+cz
, z
v
/2z
2
v
+z
v
/2z
2
v
= d
K
+c(z
z
v
/2z
2
v
z
v
/2z
2
v
)/2, z
v
/2z
2
v
+z
v
/2z
2
v
dc
2
, z
v
/2z
2
v
+z
v
/2z
2
v
dc
2
z
v
/2z
2
v
+z
v
/2z
2
v
, z
v
/2z
2
v
+z
v
/2z
2
v
. (2.11)
Since q(z) = 0, the third term on the right hand side of (2.11) is 0. The
second term can be written as
d
K
+c(z
z
v
/2z
2
v
z
v
/2z
2
v
)/2, z
dc
2
(z
, z
).
Thus the left hand side of (2.10) equals
d
K
+c(z
z
v
/2z
2
v
+z
v
/2z
2
v
)/2, z
+z
v
/2z
2
v
+z
v
/2z
2
v
dcq(z
)
= d(
K
c/2, ) cdq(z
).
.
Let k
1
2
Z and F
L
() =

L
/L
e
() be a function H C[L
/L]
with F
L
[
k
(M, ) = F
L
for all (M, ) Mp
2
(Z). Let r, t be given integers.
We dene a function
F
K
(; r, t) =
/K
e
(; r, t), (2.12)
by putting
f
(; r, t) =
0
/L
p()=
e (r(, z
) rtq(z
)) f
L++tz
()
for K
/K. (This is just a dierent notation for the denition given in

[Bo2] on page 512.) Let K
. Then as a set of representatives for L
0
/L
with p() = +K we may take (, )z/N+bz/N where b runs modulo N.
A set of representatives for L
0
/L is given by = (, )z/N+bz/N, where
runs through a set of representatives for K
/K and b runs modulo N. (Since

the index of L
0
in L
equals N, we nd again that [L
/L[ = N
2
[K
/K[.)
Theorem 2.6. We use the same notation as above. Let (M, ) Mp
2
(Z),
M =

a b
c d
. Then F
K
(; r, t) satises
F
K
(M; ar +bt, cr +dt) = ()
2k
K
(M, )F
K
(; r, t).
Proof. See [Bo2] Theorem 5.3. .
For the Poincare series F
L
,m
(, s) (see Denition 1.8) we denote the
C[K
/K]-valued function constructed in (2.12) by F

K
,m
(, s; r, t).
/L and m Z +q() with m < 0.

i). If (, z) 0 (mod N), then F
K
,m
(, s; 0, 0) = 0.
ii). If (, z) 0 (mod N), then the function F
K
,m
(, s; 0, 0) equals the
Poincare series F
K
p(),m
(, s) of index (p(), m) attached to the lattice K (see
Denition 1.35).
Proof. i). Assume that 1 k/2. From the denition of F
K
,m
(, s; 0, 0),
the Fourier expansion of F
L
,m
(, s) (see Theorem 1.9), and the asymptotic
property (1.26) of the Whittaker function W
,
(y) it follows that
F
K
,m
(, s; 0, 0) = O(1)
for y . Since F
K
,m
(, s; 0, 0) also satises
F
K
,m
(, s; 0, 0) [
k
(M, ) = F
K
,m
(, s; 0, 0), (M, ) Mp
2
(Z),
k
F
K
,m
(, s; 0, 0) =

s(1 s) + (k
2
2k)/4
F
K
,m
(, s; 0, 0),
we may infer that it vanishes identically (cf. [He] Chap. 9 Prop. 5.14c). Be-
cause F
K
,m
(, s; 0, 0) is holomorphic in s, we nd that it vanishes for all s.
ii) In the same way as in (i) it follows that
F
K
,m
(, s; 0, 0)
=
(1 +k/2 s)
(2 2s)(s +k/2)
1s
(4[m[y)(e
p()
(mx) + e
p()
(mx)) +O(1)
for y , and thereby
F
K
,m
(, s; 0, 0) F
K
p(),m
(, s) = O(1) (y ).
The assertion can be deduced as in (i). .
2.2 The theta integral
As in section 1.3 let L be an even lattice of signature (b
+
, b
) = (2, l), (1, l1),

or (0, l 2) with l 3. Put k = 1 l/2 and
0
= max1, b
+
/2 k/2.
Let L
/L and m Z +q() with m < 0. We consider the theta integral
L
,m
(v, s) =
F
'F
L
,m
(, s),
L
(, v)`y
b
+
/2
dxdy
y
2
, (2.13)
2.2 The theta integral 47
where F
,m
(, s) denotes the Maass-Poincare series of index (, m) dened in
(1.35) and
L
(, v) the Siegel theta function (2.2). If it is clear from the con-
text, to which lattice L the function (2.13) refers, we simply write
,m
(v, s).
According to Theorem 2.1, the integrand
'F
,m
(, s),
L
(, v)`y
b
+
/2
is Mp
2
(Z)-invariant. Because
L
(, v) is invariant under the action of O
d
(L),
the function
,m
(v, s) is formally invariant under O
d
(L), too.
Unfortunately, since [F
,m
(, s)[ increases exponentially as y , the
integral (2.13) diverges. However, we shall show that it can be regularized as
follows. Let T
u
be the truncated fundamental domain
T
u
= = x +iy T; y u (u R
>0
). (2.14)
Then for s = +it C with >
0
we may dene the regularized theta lift
of F
,m
(, s) by
L
,m
(v, s) = lim
u
F
u
'F
L
,m
(, s),
L
(, v)`y
b
+
/2
dxdy
y
2
. (2.15)
Roughly speaking this means that we rst carry out the integration over
x and afterwards the integration over y. We will show that
,m
(v, s) as
dened in (2.15) is holomorphic in s for >
0
. It can be continued to a
holomorphic function on s C; > 1, s = b
+
/2 k/2 with a simple
pole at s = b
+
/2 k/2. Hence for > 1 and s = b
+
/2 k/2 the function
,m
(v, s) is given by holomorphic continuation. For s = b
+
/2 k/2 we may
dene the regularized theta integral to be the constant term of the Laurent
expansion in s of
,m
(v, s) at the point s = b
+
/2 k/2.
This construction is very similar to the denition of the regularized theta
lift due to Harvey, Moore, and Borcherds [HM, Bo2]. Following Borcherds, one
could also dene
,m
(v, s) to be the constant term of the Laurent expansion
in s
of
lim
u
F
u
'F
,m
(, s),
L
(, v)`y
b
+
/2s
dxdy
y
2
(2.16)
at s
= 0. Here s
is an additional complex variable. The above expression

makes sense if 1(s
) 0. Note that the integrand in (2.16) is only invariant

under Mp
2
(Z) if s
= 0. We will prove in Proposition 2.11 that (2.15) and

(2.16) essentially yield to the same
,m
(v, s). However, for our purposes the
denition of the regularized theta lift given in (2.15) is more natural.
As in [Bo2] we will show that
,m
(v, s) is a real analytic function on
Gr(L) with singularities along smaller sub-Grassmannians. To make this more
precise, we dene a subset H(, m) of the Grassmannian Gr(L) by
H(, m) =
+L
q()=m
. (2.17)
Here
means the orthogonal complement of in Gr(L), i.e. the set of all

positive denite b
+
-dimensional subspaces v V with v .
1
It is known
that for any subset U with compact closure U Gr(L) the set
o(, m, U) = +L; q() = m, v U with v (2.18)
is nite. Hence H(, m) is locally a nite union of codimension b
+
sub-
Grassmannians. It will turn out that
,m
(v, s) is a real analytic function
on Gr(L) H(, m).
Proposition 2.8. Let v Gr(L)H(, m) and s C with >
0
. Then the
regularized theta integral (2.15) converges and denes a holomorphic function
in s. It can be continued holomorphically to
s C; > 1, s = b
+
/2 k/2,
and has a pole of rst order with residue b(0, 0, b
+
/2k/2) at s = b
+
/2k/2
(supposed that b
+
/2 k/2 > 1).
Proof. Recall that F
,m
(, s) is holomorphic in s for > 1. The integral
F
1
'F
,m
(, s),
L
(, v)`y
b
+
/2
dxdy
y
2
over the compact set T
1
is clearly holomorphic in s for > 1. Thus it suces
to consider the function
(v, s) =
y=1
1
x=0
'F
,m
(, s),
L
(, v)`y
b
+
/22
dxdy. (2.19)
We insert the Fourier expansions
F
,m
(, s) =
/L
nZ+q()
c(, n; y, s)e
(nx)
L
(, v) =
e(iyq(
v
) iyq(
v
))e
(q()x)
into (2.19), carry out the integral over x, and obtain
1
If b
+
= 0 we put H(, m) = .
(v, s) =
y=1
c(, q(); y, s) exp(2yq(

v
) + 2yq(
v
))y
b
+
/22
dy
=
y=1
c(0, 0; y, s)y
b
+
/22
dy
+
y=1
0
c(, q(); y, s) exp(4yq(
v
) + 2yq())y
b
+
/22
dy.
According to Theorem 1.9 the Fourier expansion of F
,m
(, s) can be written
more explicitly in the form
F
,m
(, s) =
1
(2s)
s
(4[m[y)(e
(mx) + e
(mx)) +H(, s),

where
H(, s) =
/L
q()Z
b(, 0, s)y
1sk/2
+
/L
nZ+q()
n=0
b(, n, s))
s
(4ny)e
(nx),
b(, n, s) = b(, n, s)
(1 +k/2 s)
(2s)(1 2s)
m,n
(
,
+
,
).
Observe that all Fourier coecients

b(, n, s) are holomorphic in s for >
1. (This is not true for b(, m, s). That is why we have introduced the
b(, n, s).) We nd that (v, s) is given by

b(0, 0, s)
1
y
b
+
/2k/2s1
dy
+
0
q()=0
b(, 0, s) exp(4yq(
v
))y
b
+
/2k/2s1
dy
+
q()=0
b(, q(), s))

s
(4q()y) exp(2yq(
v
) 2yq(
v
))y
b
+
/22
dy
+
2
(2s)
+L
q()=m
s
(4[m[y) exp(2ym4yq(
v
))y
b
+
/22
dy. (2.20)
The rst integral obviously converges for > b
+
/2 k/2 and equals
1
s b
+
/2 +k/2
.
Thus the rst summand in (2.20) has a holomorphic continuation to s
C; > 1, s = b
+
/2 k/2. If b
+
/2 k/2 > 1 it has a simple pole at
s = b
+
/2 k/2 with residue b(0, 0, b
+
/2 k/2).
Therefore it suces to show that the remaining terms in (2.20) converge
locally uniformly absolutely for > 1.
For the convergence of the second term, notice that q() = 0 implies
q(
v
) =
1
2
q(
v
)
1
2
q(
v
). The latter quadratatic form is a multiple of the
(positive denite) majorant associated with v. Thus the sum over is domi-
nated by a subseries of a theta series attached to a denite lattice.
To prove the convergence of the third term we also use the majorant
associated with v, and in addition the asymptotic property (1.26) of the
Whittaker function )
,
(y) and the estimates
b(, n, s) = O
exp(4
[mn[ )
, n +, (2.21)
b(, n, s) = O
[n[
+k/21
, n (2.22)
(locally uniformly in s for > 1). These bounds can be easily obtained from
the asymptotic behavior of the I
- resp. J
-Bessel function and the fact that

H
c
(, m, , n) is universally bounded.
If b
+
= 0, then L is a denite lattice, and
v
= 0 for all . The convergence
of the last term in (2.20) follows from the asymptotic property (1.25) of the
M-Whittaker function.
To prove the convergence of the last term for b
+
= 1, 2 we again use the
asymptotic property (1.25) and the following fact: For any C 0 and any
compact subset B Gr(L) the set
L
; q() = m, v B with q(
v
) C
is nite. By assumption v / H(, m). Hence there exists an > 0 such that
q(
v
) > for all L
with q() = m. It follows that
+L
q()=m
s
(4[m[y) exp(4yq(
v
) + 2ym)y
b
+
/22
<
+L
q()=m
exp(4yq(
v
))
<e
2y
q()=m
e
(m+q(
v
)q(
v
))
uniformly for y [1, ) and locally uniformly in s. This implies the desired
convergence statement. .
Since F
,m
= F
,m
, we have
,m
(v, s) =
,m
(v, s).
Denition 2.9. Let D C be an open subset, a D, and f a meromorphic
function on D. We denote the constant term of the Laurent expansion of f
at s = a by (
s=a
[f(s)].
Denition 2.10. For v Gr(L) H(, m) we dene
,m
(v) = (
s=1k/2
[
,m
(v, s)] .
If b
+
= 0 or b
+
= 1, then
,m
(v, s) is holomorphic at s = 1 k/2 and
we may also write
,m
(v) =
,m
(v, 1 k/2).
Proposition 2.11. Borcherds regularization of the theta integral as de-
scribed in (2.16) equals
,m
(v, s), if s = b
+
/2 k/2,
(
s=b
+
/2k/2
[
,m
(v, s)] b
(0, 0, b
+
/2 k/2), if s = b
+
/2 k/2.
Here b
(0, 0, b
+
/2 k/2) means the derivative of the coecient b(0, 0, s) at
s = b
+
/2 k/2.
Proof. The same argument as in Proposition 2.8 shows that Borcherds reg-
ularization of the theta integral can be written as
F
1
'F
,m
(, s),
L
(, v)`y
b
+
/2
dxdy
y
2
+(
s
=0
b(0, 0, s)
s
+s b
+
/2 +k/2
0
c(, q(); y, s) exp(4yq(
v
) + 2yq())y
b
+
/22
dy.
If we compare this with the expression for (v, s) given in Proposition 2.8, we
immediately obtain the assertion in the case s = b
+
/2 k/2. For s = b
+
/2
k/2 we nd that the dierence of our regularization (
s=b
+
/2k/2
[
,m
(v, s)]
and Borcherds regularization is given by
(
s=b
+
/2k/2
b(0, 0, s)
s b
+
/2 +k/2
b(0, 0, b
+
/2 k/2)(
s
=0
= b
(0, 0, b
+
/2 k/2).
.
For the rest of this section we assume that b
+
= 1, 2. In Proposition 2.8
we considered
,m
(v, s) for a xed v and varying s. Now we show that for
xed s the function
,m
(v, s) is real analytic in v on Gr(L) H(, m).
Let U Gr(L) be an open subset and f, g functions on a dense open
subset of U. Then we write
f g,
if f g can be continued to a real analytic function on U. In this case we
will say that f has a singularity of type g.
Theorem 2.12. i) For xed s s C; > 1, s = b
+
/2 k/2 the
function
,m
(v, s) is real analytic in v on Gr(L) H(, m).
ii) Let U Gr(L) be an open subset with compact closure U Gr(L).
The function
,m
(v) is real analytic on Gr(L) H(, m). On U it has a
singularity of type
2
S(,m,U)
log q(
v
),
if b
+
= 2, and of type
4
S(,m,U)
[
v
[,
if b
+
= 1.
Proof. We use the same argument as in [Bo2] Theorem 6.2. We only give a
proof for (ii), because (i) can be treated similarly.
Since
L
(, v) is real analytic and F
,m
(, s) is holomorphic in s, the
integral

F
1
'F
,m
(, s),
L
(, v)`y
b
+
/2
dxdy
y
2
is holomorphic in s and real analytic in v. As in the proof of Proposition 2.8
it suces to consider the function (v, s) dened by (2.19). We insert the
Fourier expansions of
L
(, v) and F
,m
(, s) (as given in Theorem 1.9) and
infer as in the proof of Proposition 2.8 that (v, s) is equal to
b(0, 0, s)
b
+
/2 k/2 s
+
0
q()=0
b(, 0, s) exp(4yq(
v
))y
b
+
/2k/2s1
dy
+
q()=0
b(, q(), s))
s
(4q()y) exp(2yq(
v
) + 2yq(
v
))y
b
+
/22
dy
+
2(1 +k/2 s)
(2 2s)(s +k/2)
+L
q()=m
1s
(4[m[y)
exp(4yq(
v
) + 2ym)y
b
+
/22
dy. (2.23)
The rst quantity does not depend on v. The second and the third term are
holomorphic in s near s = 1k/2 and real analytic in v Gr(L). Hence they
do not contribute to the singularity. The last term is holomorphic in s near
s = 1 k/2 and real analytic for v Gr(L) H(, m). We nd that
,m
(v) (
s=1k/2
[(v, s)]
2
+L
q()=m
k/2
(4[m[y) exp(4yq(
v
) + 2ym)y
b
+
/22
dy
2
+L
q()=m
1
exp(4yq(
v
))y
b
+
/22
dy.
In the last line we have used
k/2
(y) = e
y/2
. The integrals in the last line
only become singular if q(
v
) = 0. Thus for v U we get
,m
(v) 2
S(,m,U)
1
exp(4yq(
v
))y
b
+
/22
dy
2
S(,m,U)
(4q(
v
))
1b
+
/2
(b
+
/2 1, 4q(
v
)),
where (a, x) denotes the incomplete Gamma function dened in (1.32)
(cf. [AbSt] p. 81). Now the singularity can be read o from [Bo2] Lemma
6.1. .
Note that the functions (0, x) and (1/2, x) that occur at the end
of the proof of Theorem 2.12 can be described more explicitly as follows:
According to [E1] Vol. II p. 143 (5) one has
(0, x) = log(x)
n=1
(x)
n
n! n
, (2.24)
where denotes the Euler-Mascheroni constant. Using the recurrence relation
(a + 1, x) = a(a, x) +x
a
e
x
([E1] Vol. II p. 134 (3)) and the identity
(1/2, x
2
) =
(1 erf(x))
([AbSt] p. 82 (6.5.17)), one nds
(1/2, x
2
) = 2[x[
1
e
x
2
2
(1 erf(x)). (2.25)
Here erf(x) =
2
x
0
e
t
2
dt denotes the error function (cf. [AbSt] chapter 7).
It has the expansion
erf(x) =
2
n=0
(1)
n
x
2n+1
n!(2n + 1)
.
In particular, from (2.24) and (2.25) we can also read o the singularities of
(0, x) and (1/2, x).
,m
As in section 1.3 let L be an even lattice of signature (b
+
, b
) = (2, l), (1, l1),

or (0, l 2) with l 3, and put k = 1 l/2. In this section we nd a nice
invariant expression for
,m
(v, s) by unfolding the theta integral
,m
(v, s)
against the Poincare series F
,m
(, s). In [Br1] we started with such an invari-
ant expression for
,m
(v, s), calculated its Fourier expansion, and thereby
obtained a connection to certain elliptic modular forms.
Lemma 2.13. The Siegel theta function satises
(, v) = O(y
b
+
/2b
/2
) (y 0)
uniformly in x.
Proof. This can be proved in the same way as the analogous statement for
holomorphic elliptic modular forms. .
Let F(a, b, c; z) denote the Gauss hypergeometric function
F(a, b, c; z) =
n=0
(a)
n
(b)
n
(c)
n
z
n
n!
, (2.26)
where (a)
n
= (a + n)/(a) (cf. [AbSt] Chap. 15 or [E1] Vol. I Chap. 2).
The circle of convergence of the series (2.26) is the unit circle [z[ = 1. Using
the estimate ([E1] Vol. I p. 57 (5))
(a)
n
(b)
n
(c)
n
n!
=
(c)
(a)(b)
n
a+bc1
1 +O(n
1
)
, n ,
one nds that (2.26) converges absolutely for [z[ = 1, if 1(c a b) > 0.
Then the value for z = 1 is given by ([E1] Vol. I p. 104 (46))
F(a, b, c; 1) =
(c)(c a b)
(c a)(c b)
(c = 0, 1, 2, . . . ). (2.27)
,m
55
Theorem 2.14. Let L
/L and m Z + q() with m < 0. We have the

identity
,m
(v, s) = 2
(b
/4 +b
+
/4 +s 1)
(2s)(4[m[)
k/2s
+L
q()=m
(4[q(
v
)[)
1b
/4b
+
/4s
F(b
/4 +b
+
/4 +s 1, s b
/4 +b
+
/4, 2s; m/q(
v
)).
The series converges normally for v Gr(L)H(, m) and > b
/4+b
+
/4.
In particular, if (b
+
, b
) = (0, l 2), then
,m
(v, s) =
8[m[
(s +l/4 3/2)(s + 3/2 l/4)
# +L; q() = m.
Proof. Let v Gr(L) H(, m) and s C with > b
/4 + b
+
/4. By
denition the function
,m
(v, s) is equal to
1
(2s)
\
1
[
s
(4[m[y)e
(mx)] [
k

M,
L
(, v)
y
b
+
/2
dxdy
y
2
,
where the integral is to be understood in the regularized sense (cp. (2.15)).
According to the theta transformation formula (Theorem 2.1) this can be
written as
1
(2s)
\
1
s
(4[m[(M))(M)
b
+
/2
'e
(m1(M)),
L
(M, v)`
dxdy
y
2
=
2
(2s)
s
(4[m[y)y
b
+
/2
e(mx)
(, v)
dxdy
y
2
+
1
(2s)
M=
a b
c d
\
1
c=0
s
(4[m[(M))(M)
b
+
/2
e(m1(M))
(M, v)
dxdy
y
2
. (2.28)
The second integral on the right hand side of (2.28) can be evaluated by the
usual unfolding trick. It equals
2
(2s)
s
(4[m[y)y
b
+
/2
e(mx)
(, v)
dxdy
y
2
, (2.29)
where
( = H; [x[ 1/2, [[ 1 (2.30)
is a fundamental domain for the action of
on H
MT. By virtue
of (1.23) and Lemma 2.13, the integral in (2.29) converges absolutely if >
1 + b
/4 + b
+
/4, and therefore the unfolding is justied. Combining (2.29)
with the rst integral on the right hand side of (2.28), we obtain
,m
(v, s) =
2
(2s)
y=0
1
x=0
s
(4[m[y)y
b
+
/22
e(mx)
(, v) dxdy.
We now insert the Fourier expansion
(, v) =
+L
exp(2yq(
v
) + 2yq(
v
))e(q()x)
of
(, v) and carry out the integration over x. We nd
,m
(v, s) =
2
(2s)
y=0
+L
q()=m
s
(4[m[y)y
b
+
/22
e
4yq(
v
)+2ym
dy
=
2(4[m[)
k/2
(2s)
+L
q()=m
0
M
k/2, s1/2
(4[m[y)y
b
/4+b
+
/42
e
4yq(
v
)+2ym
dy. (2.31)
The integral in (2.31) is a Laplace transform. If
v
= 0, it equals (cp. [E2]
p. 215 (11))
(4[m[)
s
(b
/4 +b
+
/4 +s 1)
(4[q(
v
)[)
b
/4+b
+
/4+s1
F
/4 +b
+
/4 +s 1, s b
/4 +b
+
/4, 2s; m/q(
v
)
. (2.32)
In the case (b
+
, b
) = (0, l 2) the projection

v
is always 0, and the
integral in (2.31) is
0
M
k/2, s1/2
(4[m[y)y
l/45/2
e
2ym
dy. (2.33)
By the asymptotic behavior of the M-Whittaker function, this integral con-
verges absolutely for > 3/2 l/4. It can be easily checked that (2.32) still
holds:
L
57
(2.33) = lim
0+
0
M
k/2, s1/2
(4[m[y)y
l/45/2
e
2ymy
dy
= (4[m[)
3
2
l
4
(s +l/4 3/2)F(s +l/4 3/2, s + 1/2 l/4, 2s; 1).
Using (2.27) we nd
(2.33) = (4[m[)
3/2l/4
(s +l/4 3/2)(2s)
(s l/4 + 3/2)(s +l/4 1/2)
. (2.34)
If we substitute (2.32) resp. (2.34) into (2.31), we obtain the assertion. .
L
Throughout the following section, let L be an indenite even lattice of sig-
nature (b
+
, b
) = (2, l) or (1, l 1) with l 3, and put k = 1 l/2. Recall

the notation introduced in section 2.1.
Assume that L contains a primitive norm 0 vector z. Then we may use
the method of Borcherds to evaluate the theta integral
,m
(v, s) (cf. [Bo2]
Chap. 7). Thereby we obtain the Fourier expansion of
,m
(v, s) in terms of
the Fourier coecients of F
,m
(, s).
We denote the components of F
,m
(, s) by f
(, s), so that
F
,m
(, s) =
/L
e
(, s).
Here we write the Fourier expansion of f
(, s) in the form
f
(, s) =
nZ+q()
c(, n; y, s)e
(nx).
The coecients c(, n; y, s) were calculated in section 1.3. For instance, for
L
/L with q() Z we found:

c(, 0; y, s) = b(, 0, s)y
1sk/2
,
b(, 0, s) =
4
1k/2
1+sk/2
[m[
sk/2
(2s 1)(s +k/2)(s k/2)
cZ{0}
[c[
12s
H
c
(, m, , 0).
Theorem 2.15. (cp. [Bo2] Theorem 7.1.) Let L
/L and m Z +q(),
m < 0. If v Gr(L) H(, m) and z
2
v
<
1
4|m|
, then
L
,m
(v, s) is equal to
1
2[z
v
[
K
,m
(w, s) +
2
2z
2
v
sb
+
/4b
/4
(s + 1/2 b
+
/4 b
/4)
(N)
b(z/N, 0, s)
n1
e(n/N)n
b
+
/2+b
/212s
+
2
[z
v
[
n1
e(n(, ))
0
/L
p()=+K
e(n(, z
))
0
c(, q(); y, s)y
b
+
/25/2
exp
n
2
2z
2
v
y
2yq(
w
) + 2yq(
w
)
dy.
(2.35)
Here
K
,m
(w, s) is dened by
K
,m
(w, s) =
0, if (, z) 0 (mod N),
K
p(),m
(w, s), if (, z) 0 (mod N).
The third summand in the above expression for
L
,m
(v, s) converges normally
in s for > 1.
Proof. We use Theorem 2.4 to rewrite
L
,m
(v, s) as
1
2z
2
v
/L
f
(, s)
c,dZ
c(,z) (N)
e
[c +d[
2
4iyz
2
v
+d(, z
) cdq(z
K+p(cz
)
(, w; d, c)y
(b
+
1)/2
dxdy
y
2
=
1
2z
2
v
/L
(,z)0 (N)
f
(, s)
K+p()
(, w)y
(b
+
1)/2
dxdy
y
2
+
1
2z
2
v
c,dZ
/L
(,z)c (N)
e (d(, z
) cdq(z
)) f
(, s)
e
[c +d[
2
4iyz
2
v
K+p(cz
)
(, w; d, c)y
(b
+
1)/2
dxdy
y
2
. (2.36)
Here
c,dZ
means that the sum runs over all pairs (c, d) Z
2
with (c, d) =
(0, 0). If we substitute in the denition (2.12) of F
K
,m
(, s; 0, 0), we obtain for
the rst term on the right hand side of (2.36):
1
2z
2
v
F
K
,m
(, s; 0, 0),
K
(, w)
y
(b
+
1)/2
dxdy
y
2
.
L
59
By Proposition 2.7 this equals
1
2z
2
v
K
,m
(w, s).
In the second term on the right hand side of (2.36) we change to +cz
and substitute in the denition (2.12) of F

K
,m
(, s; d, c). We nd that it is
equal to
1
2z
2
v
c,dZ
e
[c +d[
2
4iyz
2
v
F
K
,m
(, s; d, c),
K
(, w; d, c)
y
(b
+
1)/2
dxdy
y
2
=
1
2z
2
v
(c,d)=1
n1
e
n
2
[c +d[
2
4iyz
2
v
F
K
,m
(, s; nd, nc),
K
(, w; nd, nc)
y
(b
+
1)/2
dxdy
y
2
. (2.37)
By Theorem 2.1 and Theorem 2.6 we get
1
2z
2
v
\
1
n1
exp
n
2
2z
2
v
(M)
F
K
,m
(M, s; n, 0),
K
(M, w; n, 0)
(M)
(b
+
1)/2
dxdy
y
2
.
As in the proof of Theorem 2.14 we split this up into
2
[z
v
[
n1
exp
n
2
2z
2
v
y
F
K
,m
(, s; n, 0),
K
(, w; n, 0)
y
(b
+
1)/2
dxdy
y
2
+
2
[z
v
[
M=
a b
c d
\
1
c=0
n1
exp
n
2
2z
2
v
(M)
F
K
,m
(M, s; n, 0),
K
(M, w; n, 0)
(M)
(b
+
1)/2
dxdy
y
2
, (2.38)
where
1
= /1. The rst integral on the right hand side of (2.38) is to
be understood in the regularized sense. As in Lemma 2.13 it can be seen that
F
K
,m
(, s; n, 0) = O(e
2|m|/y
) (2.39)
for y 0. This implies that the second integral converges absolutely if z
2
v
<
1
4|m|
. By the usual unfolding argument it can be written in the form
2
[z
v
[
n1
exp
n
2
2z
2
v
y
F
K
,m
(, s; n, 0),
K
(, w; n, 0)
y
(b
+
1)/2
dxdy
y
2
,
(2.40)
with ( as in (2.30). By (2.39) the latter integral also converges absolutely for
z
2
v
<
1
4|m|
. We nd that (2.38) equals
2
[z
v
[
0
1
n1
exp
n
2
2z
2
v
y
F
K
,m
(, s; n, 0),
K
(, w; n, 0)
y
b
+
2

5
2
dxdy.
We insert the Fourier expansions
K
(, w; n, 0) =
e(iyq(
w
) iyq(
w
) (, n))e
(q()x)
F
K
,m
(, s; n, 0) =
/K
0
/L
p()=
rZ+q()
e(n(, z
))c(, r; y, s)e
(rx)
and carry out the integration over x to get
2
[z
v
[
y=0
n1
exp
n
2
2z
2
v
y
0
/L
p()=+K
e(n(, z
))c(, q(); y, s)
e (iyq(
w
) iyq(
w
) + (, n)) y
b
+
/25/2
dy. (2.41)
In the above sum we consider the = 0 term
2
[z
v
[
y=0
n1
exp
n
2
2z
2
v
y
0
/L
p()=0+K
e(n(, z
))c(, 0; y, s)y
b
+
/25/2
dy
separately. A set of representatives for L
0
/L with p() = 0 + K is
given by z/N where runs modulo N. We substitute in c(, 0; y, s) =
b(, 0, s)y
1sk/2
and obtain
2
[z
v
[
(N)
b(z/N, 0, s)
n1
e(n/N)
y=0
exp
n
2
2z
2
v
y
y
b
+
/4+b
/43/2s
dy.
Since the latter integral equals
2z
2
v
n
2
s+1/2b
+
/4b
/4
(s + 1/2 b
+
/4 b
/4),
we nd for the = 0 term:
2
2z
2
v
sb
+
/4b
/4
(s + 1/2 b
+
/4 b
/4)
(N)
b(z/N, 0, s)
n1
e(n/N)n
b
+
/2+b
/212s
.
L
61
Regarding the remaining sum over K
0 in (2.41) we observe the

following: Using the explicit formulas for the coecients c(, r; y, s) (Theo-
rem 1.9) and the asymptotic properties of the Whittaker functions, it can
be seen that the sum of the absolute values of the terms with = 0 is in-
tegrable for > 1. Hence we may interchange integration and summation
and obtain the third term on the right hand side of (2.35). The resulting
sum con verges normally in s. (If (, z) 0 (mod N) and if there is an
p() + K with q() = m and
w
= 0, then this gives a singular con-
tribution in (2.41). However, this singularity cancels out with the singularity
of
1
2|z
v
|
K
,m
(w, s).) .
3 The Fourier expansion of
the theta lift
We now determine the Fourier expansion of the regularized theta integral
more explicitly. We nd that it consists of two dierent contributions. In
the case of signature (2, l) the rst part gives rise to a certain generalized
Borcherds product, whereas the second part carries some cohomological in-
formation.
3.1 Lorentzian lattices
Throughout this section we assume that L is Lorentzian, i.e. has signature
(1, l 1). Then
,m
(v, s) is holomorphic in s at 1 k/2. Therefore the
function
,m
(v) (see Denition 2.10) is simply given by
,m
(v, 1 k/2).
Moreover, Gr(L) is real hyperbolic space of dimension l 1, and H(, m)
is a union of hyperplanes of codimension 1. The set of points Gr(L)H(, m)
where
,m
(v) is real analytic is not connected. We call its components the
Weyl chambers of Gr(L) of index (, m).
Observe that the smaller lattice K is negative denite. Hence the Grass-
mannian of K is a point, and the projection
w
is 0 for any K
.
We now compute the Fourier expansion of
L
,m
(v) more explicitly. We
show that it can be written as the sum of two functions
L
,m
(v) and
L
,m
(v),
where
L
,m
(v) is real analytic on the whole Gr(L) and
L
,m
(v) is the re-
striction of a continuous piecewise linear function on V . Both,
L
,m
(v) and
L
,m
(v) are eigen functions of the hyperbolic Laplacian on Gr(L).
We need some basic properties of Bernoulli polynomials (cf. [E1] Vol. I
Chap. 1.13). For r N
0
the r-th Bernoulli polynomial B
r
(x) is dened by
the generating series
ze
xz
e
z
1
=
r=0
B
r
(x)
r!
z
r
([z[ < 2).
The rst Bernoulli polynomials are
B
0
(x) = 1, B
1
(x) = x 1/2, B
2
(x) = x
2
x + 1/6.
The identity
64 3 The Fourier expansion of the theta lift
B
r
(x) = rB
r1
(x)
implies that B
r
(x) is a polynomial of degree r. Let B
r
(x) be the 1-periodic
function with B
r
(x) = B
r
(x) for 0 x < 1. The Fourier expansion of B
r
(x)
can be easily determined (cf. [E1] Vol. I p. 37). For r 2 we have
B
r
(x) = r!
nZ{0}
e(nx)
(2in)
r
. (3.1)
If f(, ) is a function depending on K
and L
/L, then we use

the abbreviation
p()+K
f(, ) =
p()+K
f(, ) +
p()+K
f(, ), if L
0
/L,
0, if / L
0
/L.
(3.2)
Recall that the projection p is only dened on L
0
/L.
Proposition 3.1. Let v Gr(L) H(, m) with z
2
v
<
1
4|m|
. Then
L
,m
(v) =
1
2[z
v
[
K
,m
+ 4
2[z
v
[
(N)
b(z/N, 0)B
2
(/N)
+ 4
2[z
v
[
p()+K
q()=m
B
2
((, ) + (, z
))
+ 4
2(/[z
v
[)
k
0
/L
p()=+K
b(, q())[[
1k
n1
n
k1
e(n(, ) +n(, z
))K
1k
(2n[[/[z
v
[), (3.3)
where
K
,m
denotes the constant
K
,m
(w, 1 k/2), and b(, n) = b(, n, 1
k/2) denote the Fourier coecients of the Poincare series F
L
,m
(, 1 k/2).
The third term on the right hand side is to be interpreted as 0 if (, z)
0 (mod N). As usual K
is the modied Bessel function of the third kind

(cf. [AbSt] Chap. 9).
Proof. We apply Theorem 2.15. Since L is Lorentzian, the rst and the second
term on the right hand side of (2.35) are holomorphic in s near 1 k/2. We
nd
3.1 Lorentzian lattices 65
L
,m
(v) =
1
2[z
v
[
K
,m
+
2
2[z
v
[
(N)
b(z/N, 0)
n1
e(n/N)n
2
+
2
[z
v
[
n1
e(n(, ))
0
/L
p()=+K
e(n(, z
))
0
c(, q(); y, 1 k/2)y
2
exp
n
2
2z
2
v
y
+ 2yq()
dy. (3.4)
According to Proposition 1.10 we have
c(, q(); y, 1 k/2)
= (
,
+
,
)
m,q()
e
2my
+b(, q()))
1k/2
(4q()y)
for K
0. Moreover, we know that )

1k/2
(y) = e
y/2
(1 k, [y[) for
y < 0 by (1.33). Hence the last term in (3.4) can be rewritten as
2
[z
v
[
p()+K
q()=m
n1
e(n(, ) +n(, z
))
0
exp
n
2
2z
2
v
y
y
2
dy
+
2
[z
v
[
n1
e(n(, ))
0
/L
p()=+K
e(n(, z
))b(, q())
0
(1 k, 4[q()[y) exp
n
2
2z
2
v
y
y
2
dy. (3.5)
The rst integral in (3.5) equals
2z
2
v
n
2
. If we replace the sum over n 1 by a
sum over n = 0 and substitute in identity (3.1) for B
2
(x), we nd
L
,m
(v) =
1
2[z
v
[
K
,m
+ 4
2[z
v
[
(N)
b(z/N, 0)B
2
(/N)
+ 4
2[z
v
[
p()+K
q()=m
B
2
((, ) + (, z
))
+
2
[z
v
[
n1
e(n(, ))
0
/L
p()=+K
e(n(, z
))
b(, q())
0
(1 k, 4[q()[y) exp
n
2
2z
2
v
y
y
2
dy. (3.6)
We now compute the latter integral, which we abbreviate by
I(A, B) =
0
(1 k, Ay)e
B/y
y
2
dy (A, B > 0).
The integral representation (1.32) of the incomplete Gamma function obvi-
ously implies
(a, x)
= x
a1
e
x
.
If we integrate by parts, we nd
I(A, B) = A
1k
B
1
0
e
B/yAy
y
k
dy.
0
e
AyB/y
y
1
dy = 2(B/A)
/2
K
(2
AB) (3.7)
for the K-Bessel function ([E2] Vol. I p. 313 (6.3.17)), we get
I(A, B) = 2A
(1k)/2
B
(1+k)2
K
1k
(2
AB).
If we put this into (3.6), we obtain the assertion. .
The set of norm 1 vectors in V = L R has two components, one being
given by
V
1
= v
1
V ; v
2
1
= 1, (z, v
1
) > 0.
We may identify Gr(L) with V
1
via
V
1
Gr(L), v
1
Rv
1
.
This is the hyperboloid model of hyperbolic space. In this identication we
have
z
v
= (z, v
1
)v
1
,
[z
v
[ = (z, v
1
).
The vector dened in (2.8) equals
= z
+
(z, v
1
)v
1
2(z, v
1
)
2

z (z, v
1
)v
1
2(z, v
1
)
2
=
v
1
(z, v
1
)
z
z
2(z, v
1
)
2
.
After a short calculation one nds that v
1
can be expressed in terms of (z, v
1
)
and the projection
K
= (, z
)z K R by
v
1
= (z, v
1
)
K
+ (z, v
1
)z
+bz. (3.8)
Here b R is uniquely determined by the condition v
2
1
= 1. Conversely every
x
0
> 0, KR denes a unique v
1
V
1
as in (3.8). To be precise we may
identify
R
>0
(K R) V
1
, (x
0
, ) /x
0
+z
/x
0
+bz = v
1
, (3.9)
where b is again determined by v
2
1
= 1. Thereby we have realized Gr(L) as
the subset
1 = (x
0
, . . . , x
l2
) R
l1
; x
0
> 0
of R
l1
= R (K R). This is known as the upper half space model of
hyperbolic space.
Let V with =
K
+
1
z
+
2
z,
K
K R, and
1
,
2
R. The
scalar product of (x
0
, ) 1 with is given by
((x
0
, ), ) =
1
2x
0
2(,
K
) +
1
z
2
+ 2
2
+
1
x
0
/2. (3.10)
Let us now consider the Weyl chambers of index (, m), i.e. the connected
components of Gr(L)H(, m), in the hyperboloid model. For a Weyl cham-
ber W V
1
and L
we write (, W) > 0, if has positive inner product

with all elements in the interior of W.
Lemma 3.2. Let W V
1
be a Weyl chamber of index (, m) and assume
that L
with q() 0 or q() = m, +L = . Suppose that (, v

1
) > 0
for a v
1
W. Then (, W) > 0.
Proof. Obviously = 0. Assume that there is a v
2
W with (, v
2
) 0.
Then, since W is connected, there exists a v
0
W satisfying (, v
0
) = 0,
i.e. v
0
. Because v
0
V is a negative denite subspace, it follows that
q() < 0. Without any restriction we may conclude that q() = m and
+L = . But then (, v
0
) = 0 implies v
0
H(, m), a contradiction. .
In the coordinates of V
1
the Fourier expansion of
L
,m
is given by
L
,m
(v
1
) =
1
2(z, v
1
)
K
,m
+ 4
2(z, v
1
)
(N)
b(z/N, 0)B
2
(/N)
+ 4
2(z, v
1
)
p()+K
q()=m
B
2
(, v
1
)
(z, v
1
)
+ (, z
+ 4
2(/(z, v
1
))
k
0
/L
p()=+K
b(, q())[[
1k
n1
n
k1
e
n
(, v
1
)
(z, v
1
)
+n(, z
K
1k
(2n[[/(z, v
1
)).
(3.11)
Denition 3.3. Using the above notation we dene two functions
L
,m
,
L
,m
on Gr(L) by
L
,m
(v
1
) =

K
,m
1
(z, v
1
)
2(z
, v
1
)
+
4
2
(z, v
1
)
p()+K
q()=m
(, v
1
)
2
+ 4
2(/(z, v
1
))
k
0
/L
p()=+K
b(, q())[[
1k
n1
n
k1
e
n
(, v
1
)
(z, v
1
)
+n(, z
K
1k
(2n[[/(z, v
1
)) (3.12)
and
L
,m
(v
1
) =
L
,m
(v
1
)
L
,m
(v
1
).
Note that this denition depends on the choice of the vectors z and z
. It
is easily checked that
L
,m
=
L
,m
.
By Theorem 2.14,
L
,m
is a real valued function and
K
,m
a real constant.
According to Lemma 1.13 the coecients b(, q()) are real numbers. Hence
L
,m
and
L
,m
are real-valued functions.
Using the boundedness of the coecients b(, n) with n < 0 (cf. (2.22))
and the asymptotic property
K
(y)
2y
e
y
(y ) (3.13)
of the K-Bessel function, one can show that
L
,m
(v
1
) is real analytic on the
whole Gr(L).
Theorem 3.4. The function
L
,m
is the restriction of a continuous piece-
wise linear function on V . Its only singularities lie on H(, m). For v
1
V
1
with (z, v
1
)
2
<
1
4|m|
it equals
L
,m
(v
1
) =
2(z
, v
1
)
K
,m
+ 4
2(z, v
1
)
(N)
b(z/N, 0)B
2
(/N)
+ 4
2(z, v
1
)
p()+K
q()=m
B
2
(, v
1
)
(z, v
1
)
+ (, z
(, v
1
)
2
(z, v
1
)
2
.
(3.14)
Proof. Equality (3.14) immediately follows from (3.11) and the denition of
L
,m
(v
1
). Using the fact that
B
2
(x) = B
2
(x [x]) = x
2
(2[x] + 1)x + [x]
2
+ [x] + 1/6,
one nds that the quadratic terms in the sum on the right hand side of (3.14)
cancel out. Hence
L
,m
(v
1
) is piecewise linear on v
1
V
1
; (z, v
1
)
2
<
1
4|m|
.
Since
L
,m
(v
1
) is real analytic on Gr(L), the only singularities of
L
,m
(v
1
)
lie on H(, m) by Theorem 2.12. Moreover,
L
,m
(v
1
) is continuous on Gr(L).
Let W
1
and W
2
be two dierent Weyl chambers of index (, m) such that
the interior of W
1
W
2
is connected. Then there is a +L with q() = m
and W
1
W
2

. If
1
resp.
2
is a real analytic function on W
1
W
2
with
1
[
W
1
=
L
,m
[
W
1
resp.
2
[
W
2
=
L
,m
[
W
2
then Theorem 2.12 implies that
1
(v
1
)
2
(v
1
) = const (, v
1
).
The functions
1
=
1

L
,m
and
2
=
2

L
,m
are real analytic on
W
1
W
2
, dier by a constant multiple of (, v
1
), and satisfy
1
[
W
1
=
L
,m
[
W
1
resp.
2
[
W
2
=
L
,m
[
W
2
. We may conclude: If
L
,m
[
W
1
is linear then
L
,m
[
W
2
is linear, too.
This fact, together with the rst observation, implies that
L
,m
(v
1
) is
piecewise linear on the whole Grassmannian V
1
. .
Denition 3.5. Let W be a Weyl chamber of index (, m). Then we write
,m
(W) for the unique vector in V with the property
L
,m
(v
1
) = 8
2(v
1
,
,m
(W))
on W. We call
,m
(W) the Weyl vector of W.
Later we will need the following theorem.
Theorem 3.6. (cp. [Bo2] Theorem 10.3.) Let
L
(v
1
) =
,m
c
,m
L
,m
(v
1
) (c
,m
C)
be a nite linear combination of the
L
,m
. If
L
(v
1
) is a rational function on
V
1
, then it vanishes identically.
Remark 3.7. The condition that
L
is rational is obviously equivalent to say-
ing that
,m
c
,m
b(, q()) = 0
for all K
0 and L
0
/L with p() = +K.
Proof of Theorem 3.6. We use the following notation: Let f and g be functions
on V
1
. Then we write
f
.
= g,
if f g is the restriction of a piecewise linear function on V .
We rst prove that
L
(v
1
)
.
= 0. Let
L
(v
1
) =
,m
c
,m
L
,m
(v
1
).
By Theorem 3.4 we have
L
(v
1
)
.
=
L
(v
1
). Hence it suces to show that
L
(v
1
)
.
= 0.
The assumption that
L
is rational implies that
L
(v
1
)
.
=
1
,m
c
,m

K
,m
(z, v
1
)
2(z
, v
1
)
K
,m
+
8
(z, v
1
)
p()+K
q()=m
(, v
1
)
2
.
(3.15)
This formula for
L
(v
1
) depends on the choice of a primitive norm 0 vector
z L. We compute
L
(v
1
) using a dierent norm 0 vector and compare the
result with (3.15).
The vector z

z
2
2
z L Q has norm 0. Let a be the unique positive
integer such that
z = a
z
2
2
z
is a primitive element of L. Dene z
,

K = L z
, and p analogous
to z
, K, and p (see section 2.1). Since

L
(v
1
) is the sum of a rational function
and a piecewise linear function, we nd by Proposition 3.1 and Theorem 3.4
that
L
(v
1
)
.
=
1
2( z, v
1
)
,m
c
,m
K
,m
+ 8
p()+

K
q()=m
(, v
1
)
2
. (3.16)
It is easily seen that (LQ) z
= (KQ)Q z. Thus

KQ (KQ)Q z.
Hence every

K
in the sum in (3.16) can be written as =

K
+b z with
K
K Q and b Q. We have
(, v
1
)
2
( z, v
1
)
.
=
(
K
, v
1
)
2
( z, v
1
)
,
and (3.16) can be rewritten in the form
L
(v
1
)
.
=
1
2( z, v
1
)
,m
c
,m
K
,m
+ 8
(
K
, v
1
)
2
. (3.17)
We now compare (3.15) and (3.17) in the upper half space model 1. If v
1
is
represented by (x
0
, ) as in (3.9), we have
( z, v
1
) =
a(x
2
0
2
)
2x
0
,
2
x
0
=
1
(z, v
1
)
2(z
, v
1
). (3.18)
We nd
1
x
0
,m
c
,m
K
,m
+ 8
(, )
2
.
=
2x
0
a(x
2
0
2
)
,m
c
,m
K
,m
+
8
x
2
0
(
K
, )
2
.
We multiply this equation by (x
2
0
2
) and consider x
0
as a complex variable.
Putting x
0
= i[[ we obtain
8
,m
c
,m
(
K
, )
2
=
2
,m
c
,m
K
,m
.
Inserting this into (3.17) we nd
L
(v
1
)
.
=
2
ax
0
,m
c
,m
K
,m
=
2(z, v
1
)
a
,m
c
,m
K
,m
.
= 0
and thereby
L
(v
1
)
.
= 0.
In the upper half space model we have
L
((x
0
, )) =
1
2x
0
,m
c
,m
K
,m
+ 8
(, )
2
. (3.19)
Comparing coecients in (3.19) and (3.10), we may infer that
L
((x
0
, ))
.
= 0
already implies
L
((x
0
, )) = 0. .
3.1.1 The hyperbolic Laplacian
We now consider the functions
L
,m
and
L
,m
in the upper half space model
1 and determine the action of the of the hyperbolic Laplacian. For (x
0
, )
1 = R
>0
(K R) we have
L
,m
((x
0
, )) =

2
2x
0
K
,m
+
4
2
x
0
p()+K
q()=m
(, )
2
+ 4
2(x
0
)
l/21

K
0
/L
p()=+K
b(, q())[[
l/2
n1
n
l/22
e (n(, ) +n(, z
)) K
l/2
(2n[[x
0
). (3.20)
In the coordinates of 1 the hyperbolic Laplace operator has the form
(cp. [Ma2])
L
= x
l1
0
l2
j=0
x
j
x
3l
0
x
j
= x
2
0
l2
j=0
2
x
2
j
+ (l 3)x
0
x
0
.
It is invariant under the action of SO(1, l 1) on 1.
Theorem 3.8. The functions
L
,m
((x
0
, )) and
L
,m
((x
0
, )) are eigenfunc-
tions of the hyperbolic Laplacian with eigenvalue 1 l.
Proof. First we consider
L
,m
. Since it is piecewise linear, it suces to show
that
L
((x
0
, ), ) = (1 l)((x
0
, ), )
for any V . Using (3.10) this can be veried by a straightforward compu-
tation.
To prove the statement for
L
,m
we compute the action of
L
on the
dierent terms of the Fourier expansion (3.20). It is easily seen that
2
x
0
= (1 l)
2
x
0
+ 2(l 2)x
0
. (3.21)
Regarding the second term we nd
L
4
2
x
0
p()+K
q()=m
(, )
2
= (1 l)
4
2
x
0
p()+K
q()=m
(, )
2
16
2[m[x
0
# p() +K; q() = m. (3.22)
3.2 Lattices of signature (2, l) 73
Moreover, a lengthy but trivial calculation shows that
L
x
l/21
0
K
l/2
(2n[[x
0
)e(n(, )) = (1 l)x
l/21
0
K
l/2
(2n[[x
0
)e(n(, ))
(3.23)
for K
0. Here one has to use that K
is a solution of the dierential

equation
z
2
d
2
f
dz
2
+z
df
dz
(z
2
+
2
)f = 0.
If we put (3.21), (3.22), and (3.23) together, we see that
L
,m
((x
0
, )) = (1 l)
L
,m
((x
0
, )) + (l 2)
2x
0
K
,m
16
2[m[x
0
# p() +K; q() = m. (3.24)
But Theorem 2.14 tells us that
K
,m
=
8[m[
l/2 1
# p() +K; q() = m.
Thus the last two terms in (3.24) cancel and
L
,m
((x
0
, )) = (1 l)
L
,m
((x
0
, )).
This proves the assertion. .
3.2 Lattices of signature (2, l)
For the rest of this book let L denote a lattice of signature (2, l) with l 3.
Then the Grassmannian Gr(L) is a Hermitean symmetric space. Let z L
be a primitive norm 0 vector and choose a z
with (z, z
) = 1. Then
K = L z
is a Lorentzian lattice. We assume that K also contains a

primitive norm 0 vector. We set k = 1 l/2 and = 1 +l/2.
If v Gr(L), then we write w for the corresponding element in the Grass-
mannian Gr(K) as in section 2.1 and w
1
for its realization in the hyperboloid
model.
We shall determine the Fourier expansion of the function
,m
(v) (see
Denition 2.10) more explicitly. In the same way as in the previous section, we
show that it can be written as the sum of two real valued functions functions
L
,m
(v) and
L
,m
(v). Here
L
,m
(v) is real analytic on the whole Gr(L), and
L
,m
(v) is essentially the logarithm of the absolute value of a holomorphic
function on Gr(L) whose only zeros lie on H(, m). These functions
L
,m
,
L
,m
generalize the functions
m
and
m
considered in [Br1, Br2].
In the Fourier expansion of
L
,m
(v) the following special function will
occur. For A, B R and R with > 1 we dene
1
(A, B) =
0
( 1, A
2
y)e
B
2
y1/y
y
3/2
dy. (3.25)
Let > 0. Then it can be easily checked that
1
(A, B) <
e
2(1)
A
2
+B
2
. (3.26)
For N the function 1
(A, B) can be expressed as a nite sum of K-

Bessel functions: If n N
0
then by repeated integration by parts one nds
that (1+n, x) = n!e
x
e
n
(x), where e
n
(x) denotes the truncated exponential
series
e
n
(x) =
n
r=0
x
r
r!
(also see [E1] Vol. II p. 136 (18)). Hence we have for integral :
1
(A, B) = ( 2)!
0
e
2
(A
2
y)e
(A
2
+B
2
)y1/y
y
3/2
dy
= ( 2)!
2
r=0
A
2r
r!
0
e
(A
2
+B
2
)y1/y
y
r3/2
dy.
By virtue of (3.7) we obtain
1
(A, B) = 2( 2)!
2
r=0
A
2r
r!
(A
2
+B
2
)
1/4r/2
K
r1/2
(2
A
2
+B
2
).
Theorem 3.9. Let v Gr(L) H(, m) with z
2
v
<
1
2|m|
. Then
L
,m
(v) is
equal to
1
2[z
v
[
K
,m
(w, 1 k/2) +C
,m
+b(0, 0) log(z
2
v
)
2
p()+K
q()=m
Log (1 e((, z
) + (, ) +i[
w
[/[z
v
[))
2
0
q()0
0
/L
p()=+K
b(, q()) Log (1 e((, z
) + (, ) +i[
w
[/[z
v
[))
+
2
q()<0
0
/L
p()=+K
b(, q())
n1
1
n
e(n(, z
) +n(, ))
1
2k
(n[[/[z
v
[, n[
w
[/[z
v
[) ,
with
C
,m
= b(0, 0) (log(2) +
(1)) +b
(0, 0)
2
(N)
0 (N)
b(z/N, 0) log [1 e(/N)[.
Here b(, n) = b(, n, 1 k/2) denote the Fourier coecients of the Poincare
series F
L
,m
(, 1 k/2) as in Proposition 1.10, and b
(0, 0) means the deriva-

tive of b(0, 0, s) at s = 1 k/2. The sum over in the second line is to be
interpreted as in (3.2).
Proof. We rst assume z
2
v
<
1
4|m|
and apply Theorem 2.15. The only term
on the right hand side of (2.35) that is not holomorphic in s near 1 k/2 is
the second one
h(s) =
2
2z
2
v
/
s1/2l/4
(s l/4)
(N)
b(z/N, 0, s)
n1
e(n/N)n
l/22s
.
Here the summand with 0 (mod N) gives the singular contribution
2
2z
2
v
/
s1/2l/4
(s l/4)b(0, 0, s)(2s l/2), (3.27)
where (s) denotes the Riemann zeta function. The remaining terms with
0 (mod N) converge for > l/4 and are therefore holomorphic. Their
value at s = 1 k/2 = 1/2 +l/4 is
2
(N)
0 (N)
b(z/N, 0) log [1 e(/N)[. (3.28)
To compute the constant term in the Laurent expansion of (3.27) at s =
1 k/2 we note that
(2s l/2) =
1
2
(s 1/2 l/4)
1
(1) +. . . ,
2z
2
v
/
s1/2l/4
= 1 + (s 1/2 l/4) log
2z
2
v
/
+. . . ,
(s l/4) =
(1/2)(s 1/2 l/4) +. . . ,

b(0, 0, s) = b(0, 0) +b
(0, 0)(s 1/2 l/4) +. . . .

(For the rst expansion for instance see [E1] p. 34 (17); note that
(1) =
= Euler-Mascheroni constant.) We may infer that it is equal to
b(0, 0)
log
2z
2
v
(1/2)
(1)
+b
(0, 0).
From the duplication formula ()( + 1/2) = 2
12
(2) for the

gamma function it follows that
(1/2)/
(1)2 log 2. Thus the above

expression for the constant term in the Laurent expansion of (3.27) can be
simplied to
b(0, 0)
log(z
2
v
) log(2)
(1)
+b
(0, 0). (3.29)

Putting (3.28) and (3.29) together, we get
(
s=1k/2
[h(s)] = b(0, 0) log(z
2
v
) +C
,m
.
Hence the function
L
,m
(v) = (
s=1k/2
[
L
,m
(v, s)] is given by
1
2[z
v
[
K
,m
(w, 1 k/2) +b(0, 0) log(z
2
v
) +C
,m
+
2
[z
v
[
n1
e(n(, ))
0
/L
p()=+K
e(n(, z
))
0
c(, q(); y, 1 k/2) exp
n
2
2z
2
v
y
4yq(
w
) + 2yq()
y
3/2
dy.
(3.30)
We now insert the explicit formulas for the Fourier coecients c(, q(); y, 1
k/2) given in Proposition 1.10. We saw that c(, q(); y, 1 k/2) equals
b(, q())e
2q()y
, if q() 0,
b(, q()))
1k/2
(4q()y), if q() < 0, q() = m,
(
,
+
,
)e
2my
+b(, q()))
1
k
2
(4q()y), if q() = m.
Furthermore, according to (1.33) we have
)
1k/2
(y) = e
y/2
(1 k, [y[)
for y < 0. Hence the last summand in (3.30) can be rewritten in the form
2
[z
v
[
0
q()0
n1
e(n(, ))
0
/L
p()=+K
e(n(, z
))b(, q())
0
exp
n
2
2z
2
v
y
4yq(
w
)
dy
y
3/2
+
2
[z
v
[
+K=
q()=m
n1
e(n(, ) +n(, z
))
0
exp
n
2
2z
2
v
y
4yq(
w
)
dy
y
3/2
+
2
[z
v
[
q()<0
n1
e(n(, ))
0
/L
p()=+K
e(n(, z
))b(, q())
0
(1 k, 4[q()[y) exp
n
2
2z
2
v
y
4yq(
w
)
dy
y
3/2
. (3.31)
Here the sum over in the second term is to be understood as in (3.2).
We now compute the integrals in the above expression. Using (3.7) and
the identity
2z
K
1/2
(z) = e
z
(3.32)
(cf. [AbSt] p. 160 (10.2.17)), we obtain for A, B > 0:

0
e
AyB/y
y
3/2
dy =
B
1/2
e
2
AB
.
For instance by considering the limit A 0 it can be shown that this formula
still holds for A = 0. We may conclude that
0
exp
n
2
2z
2
v
y
4yq(
w
)
y
3/2
dy =
2[z
v
[
n
e
2n|
w
|/|z
v
|
.
In the second integral in (3.31)
0
(1 k, 4[q()[y) exp
n
2
2z
2
v
y
4yq(
w
)
y
3/2
dy
we substitute u =
2z
2
v
n
2
y and nd that it is equal to
2[z
v
[
n
1
2k
(n[[/[z
v
[, n[
w
[/[z
v
[) .
We may rewrite (3.31) in the form
2
0
q()0
0
/L
p()=+K
b(, q())
n1
1
n
e(n(, ) +n(, z
))e
2n|
w
|/|z
v
|
+ 2
p()+K
q()=m
n1
1
n
e(n(, ) +n(, z
))e
2n|
w
|/|z
v
|
+
2
q()<0
0
/L
p()=+K
b(, q())
n1
1
n
e(n(, ) +n(, z
))
1
2k
(n[[/[z
v
[, n[
w
[/[z
v
[) . (3.33)
If we insert the power series expansion of the principal branch of the logarithm
and put everything together, we obtain the stated identity for z
2
v
<
1
4|m|
.
Using the estimates (2.21) and (2.22) it can be seen that (3.33) even con-
verges and denes a real analytic function on z
2
v
<
1
2|m|
. Hence the assertion
follows by real analytic continuation. .
We now introduce a complex structure on the Grassmannian Gr(L). Let
V (C) = V C be the complexication of V = LR, and extend the bilinear
form (, ) on V to a C-bilinear form on V (C). Let P(V (C)) be the associated
projective space and denote the canonical projection by
V (C) P(V (C)), Z
L
[Z
L
].
In the zero-quadric
A = [Z
L
] P(V (C)); (Z
L
, Z
L
) = 0
we consider the open subset
/ =
[Z
L
] A; (Z
L
,

Z
L
) > 0
.
It is easily seen that / is a complex manifold that consists of 2 connected
components. The action of the orthogonal group O(V ) on V induces an action
on /. The connected component O
+
(V ) of the identity (i.e. the subgroup of
elements with positive spinor norm) preserves the components of /, whereas
O(V )` O
+
(V ) interchanges them. We choose one xed component of / and
denote it by /
+
.
If we write Z
L
= X
L
+iY
L
with X
L
, Y
L
V , then [Z
L
] / is equivalent
to
X
L
Y
L
, and X
2
L
= Y
2
L
> 0. (3.34)
But this means that X
L
and Y
L
span a two dimensional positive denite
subspace of V and thereby dene an element of Gr(L). Conversely for a
given v Gr(L) we may choose an orthogonal base X
L
, Y
L
as in (3.34) and
obtain a unique [Z
L
] = [X
L
+ iY
L
] in /
+
. (Then [
Z
L
] / corresponds to
the same v Gr(L).) We get a bijection between Gr(L) and /
+
and hereby
a complex structure on Gr(L).
We may realize /
+
as a tube domain in the following way. Suppose that
z L is a primitive norm 0 vector and z
with (z, z
) = 1. As in section
2.1, dene a sub-lattice K by K = L z
. Then K is Lorentzian and

V = (K R) Rz
Rz.
If Z
L
LC and Z
L
= Z +az
+bz with Z K C and a, b C, then we

briey write Z
L
= (Z, a, b).
We have a map from the set
Z = X +iY K C; Y
2
> 0 (3.35)
of vectors in K C with positive imaginary part to / given by
Z [Z
L
] = [(Z, 1, q(Z) q(z
))]. (3.36)
Conversely assume that [Z
L
] /, Z
L
= X
L
+ iY
L
. From the fact that X
L
,
Y
L
span a two dimensional positive denite subspace of V it follows that
(Z
L
, z) = 0. Thus [Z
L
] has a unique representative of the form (Z, 1, b).
The condition q(Z
L
) = 0 implies b = q(Z) q(z
), and thereby [Z
L
] =
[(Z, 1, q(Z) q(z
))]. Moreover, from (Z

L
,

Z
L
) > 0 one easily deduces Y
2
>
0. We may infer that the map (3.36) is biholomorphic.
The set (3.35) has two components. Let H
l
be the component which is
mapped to /
+
under (3.36). Then H
l
is a realization of /
+
as a tube domain
and can be viewed as a generalized upper half plane.
The cone iC = H
l
i(K R) is given by one of the two components of
the set Y K R; Y
2
> 0 of positive norm vectors of K R and H
l
=
K R + iC. Without any restriction we may assume that Y/[Y [; Y C
coincides with the hyperboloid model of Gr(K) that we used in section 3.1.
Let Z = X +iY be a point of H
l
and v = RX
L
+RY
L
the corresponding
element of the Grassmannian. We have
Z
L
= X
L
+iY
L
= (Z, 1, q(Z) q(z
)),
X
L
= (X, 1, q(Y ) q(X) q(z
)),
Y
L
= (Y, 0, (X, Y )).
Using X
2
L
= Y
2
L
= Y
2
we nd
z
v
= (z, X
L
)X
L
/X
2
L
+ (z, Y
L
)Y
L
/Y
2
L
= X
L
/Y
2
,
z
2
v
= 1/Y
2
.
The vector
= z
+
z
v
2z
2
v
+
z
v
2z
2
v
= z
+
z
v
z
2
v
z
2z
2
v
is given by
= (X, 0, q(X) q(z
))
and its orthogonal projection
K
by
K
= (, z
)z = X. The subspace
w V , i.e. the orthogonal complement of z
v
in v, is equal to RY
L
. Recall
from section 2.1 that we have identied w with its orthogonal projection
w
K
K R. Obviously w
K
= RY , w
1
= Y/[Y [, and thereby
w
= (, Y )Y/Y
2
,
[
w
[ = [(, Y )[[z
v
[.
The intersection
(L) = O
+
(V ) O
d
(L) (3.37)
of the discriminant kernel O
d
(L) with O
+
(V ) is a discrete subgroup of O
+
(V ).
We will be interested in its action on /
+
and H
l
.
Let L
be a vector of negative norm. If we write =

K
+ az
+ bz
with
K
K
, a Z, and b Q, then
= Z H
l
; aq(Z) (Z,
K
) aq(z
) b = 0
in the coordinates of H
l
. This set denes a prime divisor on H
l
. The sum
+L
q()=m
is a (L)-invariant divisor on H
l
with support H(, m). Following Borcherds
we call it the Heegner divisor of discriminant (, m). The proof of the next
lemma will be left to the reader.
Lemma 3.10. Let L
with q() = m < 0 and (, z) = a = 0. Assume

that Z
H
l
. Then q(Y ) < [m[/a
2
.
Assume that (, z) 0 (mod N). Let W be a Weyl chamber of Gr(K) of
index (p(), m). Then we also call the subset
Z = X +iY H
l
; Y/[Y [ W H
l
a Weyl chamber of index (, m) and denote it by W. For the corresponding
Weyl vector in K R we write
,m
(W).
If (, z) 0 (mod N), then by a Weyl chamber of index (, m) we simply
mean H
l
and put
,m
(W) = 0. (In this case there is no contribution to
L
,m
from the smaller lattice K.)
We may consider
L
,m
as a function on H
l
. By construction it is invariant
under the action of (L).
Let W H
l
be a Weyl chamber of index (, m). Suppose that Z W
with q(Y ) > [m[. Then Lemma 3.10 implies that Z / H(, m). According to
Theorem 3.9 the Fourier expansion of
L
,m
(Z) is given by
[Y [
K
,m
(Y/[Y [, 1 k/2) +C
,m
b(0, 0) log(Y
2
)
4
p()+K
(,W)>0
q()=m
log [1 e((, z
) + (, Z))[
4
(,W)>0
q()0
0
/L
p()=+K
b(, q()) log [1 e((, z
) + (, Z))[
+
2
q()<0
0
/L
p()=+K
b(, q())
n1
1
n
e(n(, z
) +n(, X))
1
2k
(n[[[Y [, n(, Y )) .
Here we have used b(, n) = b(, n) and the fact that [
w
[ = (, Y )[z
v
[ for
any K
with (, W) > 0.
Denition 3.11. We dene a function
L
,m
: H
l
R by
L
,m
(Z) =
[Y [
2

K
,m
(Y/[Y [)b(0, 0) log(Y
2
)+
2
q()<0
0
/L
p()=+K
b(, q())
n1
1
n
e(n(, z
) +n(, X))1
2k
(n[[[Y [, n(, Y ))
and a function
L
,m
: H
l
H(, m) R by
L
,m
(Z) =
L
,m
(Z)
L
,m
(Z).
Here
K
,m
is given by
K
,m
=
0, if (, z) 0 (mod N),
K
p(),m
, if (, z) 0 (mod N) (see Def. 3.3).
Note that this denition depends on the choice of z and z
. It is easily
checked that
L
,m
=
L
,m
and
L
,m
=
L
,m
Using the boundedness of the coecients b(, n) with n < 0 (cf. (2.22))
and the asymptotic property (3.26) of the function 1
2k
(A, B), one can show
that
L
,m
is real analytic on the whole H
l
.
According to Lemma 1.13 the coecients b(, n) are real numbers. We
may infer that
L
,m
and
L
,m
are real-valued functions.
Denition 3.12. Let D be a domain in C
l
and f : D R a twice continu-
ously dierentiable function. Then f is called pluriharmonic, if
2
z
i
z
j
f = 0
for all 1 i, j l.
Lemma 3.13. Let D C
l
be a simply connected domain and f : D R
a twice continuously dierentiable function. Then f is pluriharmonic if and
only if there is a holomorphic function h : D C with f = 1(h).
Proof. See [GR] chapter IX section C. .
Now let W H
l
be a xed Weyl chamber of index (, m) and
,m
(W)
the corresponding Weyl vector. For Z W with q(Y ) > [m[ we may write
the Fourier expansion of
L
,m
(Z) in the form
C
,m
+ 8(
,m
(W), Y ) 4
p()+K
(,W)>0
q()=m
log [1 e((, z
) + (, Z))[
4
(,W)>0
q()0
0
/L
p()=+K
b(, q()) log [1 e((, z
) + (, Z))[ . (3.38)
Denition 3.14. For Z H
l
with q(Y ) > [m[ we dene
,m
(Z) = e
(
,m
(W), Z)

p()+K
(,W)>0
q()=m
1 e((, z
) + (, Z))
(,W)>0
q()0
0
/L
p()=+K
1 e((, z
) + (, Z))
b(,q())
. (3.39)
Lemma 3.15. The innite product (3.39) converges normally for Z H
l
with q(Y ) > [m[.
Proof. It suces to show that the two series
p()+K
(,W)>0
q()=m
e
2(,Y )
and
(,Y )>0
q()0
0
/L
p()=+K
b(, q())e
2(,Y )
converge normally. The convergence of the second series for q(Y ) > [m[ fol-
lows from the asymptotic behavior (2.21) of the coecients b(, n).
It is easily seen that the rst series converges normally for Z W. Now
suppose that U is an arbitrary open subset of H
l
with compact closure U
H
l
. Then there are only nitely many K
with q() = m and + K =

p() such that (, W) > 0 and (, Y ) 0 for some Z = X + iY U.
Using this fact, one immediately obtains the convergence of the rst series
on U. .
By Lemma 3.15 the function
,m
(Z) is holomorphic on Z H
l
; q(Y ) >
[m[ and satises
log [
,m
(Z)[ =
1
4
(
L
,m
(Z) C
,m
) (3.40)
on the complement of H(, m).
,m
(Z) has a holomorphic continuation to
H
l
, and (3.40) holds on H
l
H(, m). Let U H
l
be an open subset with
compact closure U H
l
and denote by o(, m, U) the nite set
o(, m, U) = +L; q() = m, Z U with (Z
L
, ) = 0 .
Then
,m
(Z)
S(,m,U)
(, Z
L
)
1
(3.41)
is a holomorphic function without any zeros on U.
Proof. It suces to show that
,m
(Z) can be continued holomorphically to
any simply connected open set U with compact closure U H
l
and nonempty
intersection U Z H
l
; q(Y ) > [m[, and that (3.40), (3.41) hold on U.
According to Theorem 2.12 we may consider
L
,m
(Z) + 4
S(,m,U)
log [(, Z
L
)[ (3.42)
as a real analytic function on U. Moreover, the Fourier expansion (3.38) of
L
,m
on Z W; q(Y ) > [m[ implies that (3.42) is even pluriharmonic (by
real analytic continuation on U). By Lemma 3.13 there exists a holomorphic
function f : U C with
L
,m
(Z) + 4
S(,m,U)
log [(, Z
L
)[ = 1(f(Z)).
On the non-empty intersection of the open sets U and Z H
l
; q(Y ) > [m[
we therefore have
1
Log
,m
(Z)
S(,m,U)
(, Z
L
)
1
C
,m
/4
=
1
4
1(f(Z)).
But now
1
4
f can only dier by an additive constant from the expression in
brackets on the left-hand side. We may assume that this constant equals zero
and nd
,m
(Z)
S(,m,U)
(, Z
L
)
1
= e
C
,m
/4
e
f(Z)/4
.
Since e
f(Z)/4
is a holomorphic function without any zeros on U, we obtain
the assertion. .
If 2 = 0 in L
/L, then all zeros of

,m
(Z) have order 2, because
o(, m, U) implies o(, m, U). If 2 = 0 in L
/L, then the zeros of
,m
have order 1, and
,m
=
,m
.
According to (3.40) the function
,m
is independent of the choice of W
up to multiplication by a constant of absolute value 1.
3.3 Modular forms on orthogonal groups
In this section we review a few basic facts on modular forms for orthogonal
groups. The theory can be developed similarly as in [Fr1], [Fr2].
Recall the realizations /
+
and H
l
of the Hermitean symmetric space
Gr(L) introduced in the previous section. Let
/
+
= W V (C) 0; [W] /
+
V (C) (3.43)
be the cone over /
+
P(V (C)). The map
Z Z
L
= (Z, 1, q(Z) q(z
)) (3.44)
denes an embedding of H
l
into

/
+
, and the assignment Z [Z
L
]
P(V (C)) induces an isomorphism H
l
/
+
. The linear action of the or-
thogonal group O
+
(V ) on /
+
induces an action on H
l
such that the diagram
/
+
[W][W]
//
/
+
H
l
Z[Z
L
]
OO
ZZ
//
H
l
OO
(3.45)
3.3 Modular forms on orthogonal groups 85
commutes for all O
+
(V ). It is easily veried that Z is the unique
element of H
l
with the property
((Z
L
), z)(Z)
L
= (Z
L
). (3.46)
The function
j(, Z) = ((Z
L
), z)
on O
+
(V )H
l
is an automorphy factor for O
+
(V ), i.e. it satises the cocycle
relation
j(
1
2
, Z) = j(
1
,
2
Z)j(
2
, Z)
and does not vanish on H
l
.
Let r be a rational number. If O
+
(V ) and Z H
l
, we dene
j(, Z)
r
:= e
r Log j(,Z)
,
where Log j(, Z) denotes a xed holomorphic logarithm of j(, Z). There
exists a map w
r
from O
+
(V ) O
+
(V ) to the set of roots of unity (of order
bounded by the denominator of r) such that
j(
1
2
, Z)
r
= w
r
(
1
,
2
)j(
1
,
2
Z)
r
j(
2
, Z)
r
.
This map obviously only depends on r modulo Z.
Denition 3.17. Let O
+
(V ) be a subgroup and r Q as above. By a
multiplier system of weight r for we mean a map
: S
1
= t C; [t[ = 1
satisfying
(
1
2
) = w
r
(
1
,
2
)(
1
)(
2
).
If r Z, then is actually a character
1
of . If is a multiplier system of
weight r for , then ()j(, Z)
r
is a cocycle of .
We now dene modular forms for congruence subgroups of O
+
(V ).
Denition 3.18. Let (L) be a subgroup of nite index and a mul-
tiplier system for of weight r Q. A meromorphic function F on H
l
is
called a meromorphic modular form of weight r and multiplier system with
respect to , if
F(Z) = ()j(, Z)
r
F(Z) (3.47)
for all . If F is even holomorphic on H
l
then it is called a holomorphic
modular form. (Since l 3, then the Koecher principle ensures that F is also
holomorphic on the Satake boundary.)
1
Here multiplier systems and characters are always unitary.
Remark 3.19. Since l 3, the Lie group O
+
(V ) has no almost simple factor
of real rank 1. This implies that the factor group of as in Denition 3.18
modulo its commutator subgroup is nite (see [Mar] Proposition 6.19 on
p. 333). Thus any multiplier system of has nite order.
Sometimes it is convenient to consider modular forms as homogeneous
functions on

/
+
. Here we temporarily have to assume that r Z. (Otherwise
one would have to work with covers of /
+
.)
If F is a meromorphic modular form on H
l
of weight r, then we may
dene a meromorphic function

F on

/
+
by
F(tZ
L
) = t
r
F(Z) (t C 0).
It is easily seen that:
i)

F is homogeneous of degree r, i.e.

F(tZ
L
) = t
r

F(Z
L
) for any t C0;
ii)

F is invariant under , i.e.

F(Z
L
) = ()

F(Z
L
) for any .
Conversely, if G is any meromorphic function on

/
+
satisfying (i) and (ii),
then the function F(Z) = G(Z
L
) is a meromorphic modular form of weight r
on H
l
in the sense of Denition 3.18. Thus modular forms can also be viewed
as meromorphic functions on

/
+
satisfying (i) and (ii).
Let u V = L R be an isotropic vector and v V be orthogonal to u.
The Eichler transformation E(u, v), dened by
E(u, v)(a) = a (a, u)v + (a, v)u q(v)(a, u)u (3.48)
for a V , is an element of O
+
(V ). If u, v belong to the lattice L, then E(u, v)
lies in the discriminant kernel (L).
For any k K the special Eichler transformation E(z, k) (L) acts on
H
l
as the translation Z Z k and the corresponding automorphy factor
equals 1. This implies that any holomorphic modular form F of weight r Q
for the group (L) is periodic with period lattice K in the following sense:
There exists a vector K Q (which is unique modulo K
) such that
F(Z +k) = e((, k))F(Z)
for all k K. Thus F has a Fourier expansion of the form
F(Z) =
+K
a()e((, Z)).
If F has integral weight and trivial multiplier system, then = 0. The
Koecher principle tells us that a() = 0, only if lies in the closure of
the positive cone C K R (see page 79), in particular q() 0. A precise
version will be stated later in Proposition 4.15.
The following lemma generalizes the elementary formula (M) =
y
|c+d|
2
for H and M =
a b
c d
SL
2
(R).
3.4 Borcherds products 87
Lemma 3.20. If O
+
(V ) and Z = X +iY H
l
, then
q((Z)) =
q(Y )
[j(, Z)[
2
.
Proof. We use the formula q(Y ) =
1
4
(Z
L
, Z
L
) and (3.46). One has
q((Z)) =
1
4
(Z
L
)
(Z
L
, z)
,
(Z
L
)
(Z
L
, z)
=
1
4[(Z
L
, z)[
2
(Z
L
), (Z
L
)
=
q(Y )
[j(, Z)[
2
.
.
Lemma 3.21. (cp. [Bo2] Lemma 13.1.) Let r Q and (L) be sub-
group of nite index. Suppose that is a meromorphic function on H
l
for
which [(Z)[q(Y )
r/2
is invariant under . Then there exists a multiplier sys-
tem of weight r for such that is a meromorphic modular form of weight
r and multiplier system with respect to .
Proof. Let . By assumption we have
[(Z)[q((Z))
r/2
= [(Z)[q(Y )
r/2
.
By Lemma 3.20 this means that the function
|(Z)|
|(Z)|
[j(, Z)[
r
is constant
with value 1. Hence, according to the maximum modulus principle, there
exists a constant () of absolute value 1 such that
(Z)
(Z)
j(, Z)
r
= ().
The function : C is a multiplier system of weight r. .
3.4 Borcherds products
Theorem 3.16 states that
,m
(Z) is a holomorphic function on H
l
with
divisor H(, m). Observe that
,m
(Z) is not necessarily automorphic; by
construction we only know that
[
,m
(Z)[e
L
,m
(Z)/4
is invariant under (L). However, taking suitable nite products of the
,m
one can attain that the main parts of the
L
,m
cancel out. Thereby one nds
a new explanation of [Bo2] Theorem 13.3 and [Bo3] from a cohomological
point of view.
Let f : H C[L
/L] be a nearly holomorphic modular form for Mp

2
(Z)
of weight k = 1 l/2 with principal part
/L
nZ+q()
n<0
c(, n)e
(n).
Then we call the components of
Gr(K)
0
/L
nZ+q()
n<0
c(,n)=0
H(p(), n)
the Weyl chambers of Gr(K) with respect to f. Let W be such a Weyl cham-
ber. For every L
0
/L, n Z +q() with n < 0 and c(, n) = 0 there is a
Weyl chamber W
,n
of Gr(K) of index (p(), n) (as in section 3.1) such that
W W
,n
. Then
W =
0
/L
nZ+q()
n<0
c(,n)=0
W
,n
.
We dene the Weyl vector
f
(W) attached to W and f by
f
(W) =
1
2
0
/L
nZ+q()
n<0
c(, n)
p(),n
(W
,n
).
As before, for K
we write (, W) > 0, if has positive inner product

with all elements in the interior or W. We will also call the subset Z =
X +iY H
l
; Y/[Y [ W H
l
a Weyl chamber with respect to f.
Theorem 3.22 (Borcherds). Let L be an even lattice of signature (2, l)
with l 3, and z L a primitive isotropic vector. Let z
, K = Lz
, and the projection p be dened as in section 2.1. Moreover, assume that

K also contains an isotropic vector.
Let f be a nearly holomorphic modular form of weight k = 1 l/2 whose
Fourier coecients c(, n) are integral for n < 0. Then
(Z) =
/L
mZ+q()
m<0
,m
(Z)
c(,m)/2
is a meromorphic function on H
l
with the following properties:
i) It is a meromorphic modular form of (rational) weight c(0, 0)/2 for the
orthogonal group (L) with some multiplier system of nite order. If
c(0, 0) 2Z, then is a character.
ii) The divisor of (Z) on H
l
is given by
() =
1
2
/L
mZ+q()
m<0
c(, m)H(, m).
(The multiplicities of H(, m) are 2, if 2 = 0 in L
/L, and 1, if 2 = 0
in L
/L. Note that c(, m) = c(, m) and H(, m) = H(, m).)

iii) Let W H
l
be a Weyl chamber with respect to f and m
0
= minn
Q; c(, n) = 0. On the set of Z H
l
, which satisfy q(Y ) > [m
0
[, and
which belong to the complement of the set of poles of (Z), the function
(Z) has the normally convergent Borcherds product expansion
(Z) = Ce
(
f
(W), Z)
(,W)>0
0
/L
p()=+K
1 e((, z
) +(, Z))
c(,q())
.
Here C is a constant of absolute value 1, and
f
(W) K R denotes
the Weyl vector attached to W and f. (Usually
f
(W) can be computed
explicitly using Theorem 3.4.)
Proof. Throughout the proof we write b
,m
(, n) instead of b(, n, 1 k/2)
for the (, n)-th Fourier coecient of the Poincare series F
,m
(, 1 k/2) to
emphasize the dependence on (, m). Note that
f() =
1
2
/L
mZ+q()
m<0
c(, m)F
,m
(, 1 k/2)
by Proposition 1.12. Thereby the Fourier coecients c(, n) of f with n 0
are given by
c(, n) =
1
2
/L
mZ+q()
m<0
c(, m)b
,m
(, n). (3.49)
In particular, by Lemma 1.13 the coecients c(, n) are real numbers. Dene
two functions
(Z) =
1
2
/L
mZ+q()
m<0
c(, m)
L
,m
(Z),
(Z) =
1
2
/L
mZ+q()
m<0
c(, m)
L
,m
(Z).
i) First, we show that the existence of the nearly holomorphic modular
form f implies that the main terms in (Z) cancel. By virtue of Proposition
1.16 we may replace the coecients b(, q()) in the Fourier expansion of
L
,m
(see Denition 3.11) by
1
(1 k)
p
,q()
(, m),
where p
,n
(, m) denotes the (, m)-th coecient of the Poincare series
P
,n
S
,L
. Hence the Fourier expansion of (Z) can be written in the form
(Z) =
1
2
/L
mZ+q()
m<0
c(, m)
[Y [
2

K
,m
(Y/[Y [) b
,m
(0, 0) log(Y
2
)
(1 k)
q()<0
0
/L
p()=+K
/L
mZ+q()
m<0
c(, m)p
,q()
(, m)
n1
1
n
e(n(, z
) +n(, X))1
2k
(n[[[Y [, n(, Y )) .
But Theorem 1.17 implies that the sums
,m
c(, m)p
,q()
(, m)
vanish. Thus (Z) can be simplied:
(Z) =
1
2
log(Y
2
)
,m
c(, m)b
,m
(0, 0) +
[Y [
2
,m
c(, m)
K
,m
(Y/[Y [).
If we apply the same argument to the Fourier expansion of the
K
,m
, we nd
that the function
,m
c(, m)
K
,m
is rational. Hence, according to Theorem
3.6, it vanishes identically. Using (3.49) we nally obtain
(Z) = c(0, 0) log(Y
2
)
and hereby
(Z) = c(0, 0) log(Y
2
) +
1
2
/L
mZ+q()
m<0
c(, m)
L
,m
(Z). (3.50)
By Theorem 3.16 there is a real constant C such that
[(Z)[ = C exp
1
8
/L
mZ+q()
m<0
c(, m)
L
,m
(Z)
.
Inserting (3.50) we get
[(Z)[q(Y )
c(0,0)/4
= C
e
(Z)/4
.
Since (Z) is invariant under (L), the function [(Z)[q(Y )
c(0,0)/4
is invari-
ant, too. Using Lemma 3.21 we may infer that (Z) is a modular form of
weight
c(0,0)
2
with some multiplier system for (L).
ii) This immediately follows from Theorem 3.16.
iii) It is a consequence of Theorem 3.16 and equation (3.49) that
log [(Z)[ = 2(
f
(W), Y )
+
(,W)>0
0
/L
p()=+K
c(, q()) log [1 e((, z
) + (, Z))[.
This implies the last assertion. .
Remark 3.23. According to Proposition 1.16 the coecient c(0, 0) is equal to
the linear combination
1
2
/L
mZ+q()
m<0
c(, m)q(, m)
of the coecients q(, n) of the Eisenstein series E M
,L
. In particular, the
assumption that the c(, n) are integral for n < 0 implies that c(0, 0) Q by
Proposition 1.7.
The fact that has nite order implies that the Weyl vector
f
(W) lies
in K Q. This can also be proved directly for the individual Weyl vectors
,n
(W) (see Denition 3.5) using Theorem 3.4 and Theorem 2.14.
3.4.1 Examples
Let H be the lattice Z
2
with the quadratic form q((a, b)) = ab. Any lattice
isomorphic to H is called a hyperbolic plane.
1. Let L = H H be the even unimodular lattice of signature (2, 2).
Strictly speaking this case is not covered by our approach to Borcherds prod-
ucts because l = 2. However, if we dene all Poincare and Eisenstein series in
section 1.2 using Heckes trick, then our argument still works. There are two
minor diculties: To get all nearly holomorphic modular forms of weight 0
(Proposition 1.12) we also have to consider Eisenstein series of weight 0.
Moreover, the Eisenstein seri es E of weight 2 (Theorem 1.6) is in gen-
eral no longer holomorphic. Some steps of the proof of Theorem 3.22 can
be simplied. For instance one could use Theorem 2.14 and the fact that
F(1, 1, 2; z) = z
1
log(1 z) to compute the Weyl vector
f
(W) directly.
The results of section 3.1 are not needed in this case.
The space M
!
0,L
of nearly holomorphic modular forms of weight 0 is equal
to the polynomial ring C[j()], where
j() = q
1
+ 744 + 196884q +. . .
is the usual j-function, and q = e(). In particular any Fourier polynomial
n<0
c(n)q
n
is the principal part of a nearly holomorphic modular form.
This is equivalent to the fact that there are no elliptic cusp forms of weight
2, i.e. S
2,L
= 0. We consider the nearly homomorphic modular form
J() = j() 744 =
n1
c(n)q
n
in M
!
0,L
. Modular forms for the orthogonal group (L) can be identied with
modular forms on HH for the group SL
2
(Z)SL
2
(Z). The Heegner divisors
H(m) =
1
2
H(0, m) are just the Hecke correspondences T([m[) on H H. In
particular H(1) is given by the translates of the diagonal (, ); H.
In this case Theorem 3.22 (when extended to l = 2) tells us that there exists
a (on HH) holomorphic modular form for SL
2
(Z) SL
2
(Z) of weight 0
with divisor T(1). It is easily checked that equals the function j(
1
)j(
2
).
We obtain the product expansion
j(
1
) j(
2
) = p
1

m>0
nZ
(1 p
m
q
n
)
c(mn)
,
where p = e(
1
) and q = e(
2
). The product converges for (
1
)(
2
) > 1.
This is the famous denominator formula for the monster Lie algebra [Bo4].
It was proved independently in the 80s by Koike, Norton, and Zagier.
2. This example is due to Gritsenko and Nikulin [GN]. Let D be the
negative denite lattice Z with the quadratic form q(a) = a
2
and L the
lattice DHH of signature (2, 3). In this case the space M
!
1/2,L
of nearly
holomorphic modular forms of weight 1/2 can be identied with the space
J
0,1
of weak Jacobi forms of weight 0 and index 1 in the sense of [EZ]. The
obstruction space S
5/2,L
can be identied in the same way with the space of
skew holomorphic Jacobi forms of weight 3 and index 1. This space equals
0. Therefore every Fourier polynomial is the principal part of a weak Jacobi
form in

J
0,1
. We consider the special weak Jacobi form
0,1
(, z) =

12,1
(, z)
()
=
n0
rZ
c(n, r)q
n
r
= ( + 10 +
1
) +q(10
2
64
1
+ 108 64 + 10
2
) +O(q
2
)
in

J
0,1
. Here we have used the notation of [EZ] '9, in particular = e(z).
Modular forms for the orthogonal group (L) can be identied with Siegel
modular forms for the symplectic group Sp(4, Z) of genus 2. The Heegner
divisor H(, m) only depends on m and is equal to the Humbert surface of
discriminant [4m[ in the Siegel upper half plane H
S
of genus 2. For Z =
(
z
z
) H
S
we write q = e(), r = e(z), and s = e(
). According to
Theorem 3.22 there exists a holomorphic modular form (Z) of weight 5 for
the group Sp(4, Z), whose divisor is the Humbert surface of discriminant 1.
It is known that the product
(2)
0
=
a,b
(Z)
of the ten even theta constants is a Siegel modular form of weight 5 with the
same divisor (see [Fr1] chapter 3.1). Thus has to be a constant multiple of
(2)
0
. It easily checked that the right constant factor is 1/64. We obtain the
following product expansion for
(2)
0
:
1
64
(2)
0
= (qrs)
1/2

i,j,kZ
ij/2+k>0
(1 q
i
r
j
s
k
)
c(ik,j)
.
The product converges on det(Y ) > 1/4, where Y = (Z). Maass showed in
[Ma3] that
(2)
0
can also be constructed as a Saito-Kurokawa lifting.
3. Let K be the even unimodular lattice of signature (1, 33) and L =
K H. Then k = 16, = 18, and S
18,L
= Cg(), where g() is the unique
normalized elliptic cusp form of weight 18, i.e.
g() = ()E
6
() = q 528q
2
4284q
3
+ 147712q
4
+. . .
=
n=1
a(n)q
n
.
Here () = q
n1
(1 q
n
)
24
and E
r
() denotes the Eisenstein series of
weight r normalized such that its constant term equals 1. The nearly holo-
morphic modular form
f() = E
8
()/()
2
= q
2
+ 528q
1
+ 86184 + 4631872q +. . .
=
n=2
c(n)q
n
belongs to M
!
16,L
. Write H(m) =
1
2
H(0, m). Theorem 3.22 implies that
there is a holomorphic modular form of weight 43092 for the orthogonal
group (L) with divisor H(2) + 528H(1) and product expansion
(Z) = e((
f
(W), Z))
K
(,W)>0
(1 e((, Z)))
c(q())
for a suitable choice of W and
f
(W) K Q. The product converges for
q(Y ) > 2. Since S
18,L
= 0, in this case not any Fourier polynomial is the
principal part of a nearly holomorphic modular form. The obstructions are
given by the coecients a(n) of g.
A conjecture of Lehmer says that a(n) = 0 for all positive integers n.
(This conjecture was originally stated for the coecients of () but was
generalized later.) Freitag pointed out that Lehmers conjecture can be re-
stated in terms of the geometry of the modular variety A
L
= H
l
/(L). The
vanishing of a(n) implies by Theorem 3.22 that the Heegner divisor H(n)
vanishes in the modied divisor class group

Cl(A
L
) Q of A
L
(see chapter
5). On the other hand it is a consequence of our converse-theorem 5.12 that
the vanishing of H(n) in
Cl(A
L
) Q implies that a(n) = 0.
4 Some Riemann geometry on O(2, l)
The aim of this chapter is to prove Theorem 4.23. To this end we have to
exploit the Riemann structure of the quotient H
l
/(L).
4.1 The invariant Laplacian
In this section we consider the O
+
(V )-invariant Laplacian which acts on
functions on the generalized upper half plane H
l
. General references are the
books [GH, Hel, Wel].
As in section 3.2 let z L be a primitive norm 0 vector and z
with (z, z
) = 1. Then z = z
q(z
)z is an isotropic vector in L Q. Put

K = L z
and recall the denition of H

l
and /
+
.
First, we introduce suitable coordinates on H
l
. Let d K be a primitive
norm 0 vector and d
with (d, d
) = 1. Then D = K d
is
a negative denite lattice and K Q = (D Q) Qd
Qd. The vector
d = d
q(d
)d K Q has norm 0 and satises (
d, d) = 1. The lattice
D Z
d Zd is commensurable with K.
Let d
3
, . . . , d
l
be an orthogonal basis of DR normalized such that d
2
3
=
= d
2
l
= 2. Then

d, d, d
3
, . . . , d
l
is a basis of K R. If Z = z
1

d +
z
2
d + z
3
d
3
+ + z
l
d
l
K C, we briey write Z = (z
1
, z
2
, z
3
, . . . , z
l
) and
denote its real part by X = (x
1
, x
2
, x
3
, . . . , x
l
), resp. its imaginary part by
Y = (y
1
, y
2
, y
3
, . . . , y
l
). We obviously have
q(Y ) = y
1
y
2
y
2
3
y
2
4
y
2
l
and
H
l
= Z = (z
1
, . . . , z
l
) K C; y
1
> 0, y
1
y
2
y
2
3
y
2
4
y
2
l
> 0.
Moreover, sometimes we just write Z = (z
1
, z
2
, Z
D
) for Z = (z
1
, . . . , z
l
) H
l
,
where Z
D
= z
3
d
3
+ + z
l
d
l
D C. The real part of Z
D
is denoted by
X
D
, the imaginary part by Y
D
.
The generalized upper half plane H
l
can be viewed as a realization of the
irreducible Hermitean symmetric space O
+
(V )/H, where H

= SO(2) SO(l)
denotes a maximal compact subgroup. The Killing form on the Lie algebra of
96 4 Some Riemann geometry on O(2, l)
O
+
(V ) induces an O
+
(V )-invariant Riemann metric g on H
l
. According to
[Hel] Chap. VIII '5 any O
+
(V )-invariant Riemann metric on H
l
is a constant
multiple of g. In this sense the geometry on H
l
is determined uniquely. By
[Hel] Chap. VIII '4 the metric g is Kahlerian.
We will now describe g more explicitly. Lemma 3.20 implies that the
(1, 1)-form
=
i
2
log(q(Y ))
is invariant under O
+
(V ). Moreover, is positive, that is
=
i
2
,
h
(Z)dz
d z
,
where h(Z) = (h
(Z)) is a positive denite Hermitean matrix for any Z

H
l
. In fact, since is O
+
(V )-invariant it suces to check this at the special
point I = i(1, 1, 0, . . . , 0) H
l
. A somewhat lengthy but trivial computation
shows that
1
h(I) =
1
4
1
1
2
.
.
.
2
.
Furthermore, is obviously closed. Hence it denes an O
+
(V )-invariant
K ahler metric on H
l
. Its underlying Riemann metric has to be a constant
multiple of g. It is easily seen that the attached invariant volume element
equals
dX dY
q(Y )
l
,
where dX = dx
1
dx
l
and dY = dy
1
dy
l
. The invariant Laplace operator
is given by
(d d + d d).
Here denotes the Hodge star operator for the Riemann metric g. Nakajima
[Nak] showed that, up to a constant multiple, the invariant Laplace operator
is equal to
=
l
,=1
y
q(Y )
2
+

2

1
2
l
=3
, (4.1)
where
1
Zero matrix entries are often omitted.
4.1 The invariant Laplacian 97
=

z
=
1
2
i

y
=

z
=
1
2
+i

y
.
(Note that there is a misprint in [Nak] in the denition of
1
.) The proof
relies on the fact that there is up to a constant multiple only one O
+
(V )-
invariant second order dierential operator on H
l
. Therefore one only has to
verify the invariance of .
We will now show that for > 1 k/2 and Z / H(, m) the function
L
,m
(Z, s) is an eigenfunction of . First, we prove some lemmas.
Lemma 4.1. If Z = X +iY H
l
and s C, then
q(Y )
s
=
s
2
2

sl
4
q(Y )
s
, (4.2)
log(q(Y )) = l/4. (4.3)
Proof. This can be veried by a straightforward computation. .
Recall the denition (1.34) of the Laplace operator
k
of weight k. We de-
ne the usual Maass dierential operators on smooth functions H C[L
/L]
by
R
k
= 2i

+ky
1
, (4.4)
L
k
= 2iy
2

. (4.5)
For any smooth function f : H C[L
/L] and (M, ) Mp

2
(Z) we have
(R
k
f) [
k+2
(M, ) = R
k
(f [
k
(M, )) ,
(L
k
f) [
k2
(M, ) = L
k
(f [
k
(M, )) .
Hence R
k
raises the weight, whereas L
k
lowers it. The operator
k
can be
expressed in terms of R
k
and L
k
by
k
= L
k+2
R
k
k = R
k2
L
k
.
Lemma 4.2. Let f, g : H C[L
/L] be smooth functions satisfying f [

k
(M, ) = f and g [
k+2
(M, ) = g for all (M, ) Mp
2
(Z). Then
F
u
'f, (L
k+2
g)`y
k2
dxdy
F
u
'(R
k
f), g`y
k
dxdy =
1/2
1/2
'f, g`y
k
y=u
dx.
Here T
u
denotes the truncated fundamental domain (2.14).
Proof. The assumptions imply that = y
k
'f(), g()`d is a SL
2
(Z)-
invariant 1-form on H. By Stokes theorem we have
F
u
y
k
'f(), g()`d =
F
u
d
y
k
'f(), g()`d
F
u

y
y
k
'f(), g()` i

x
y
k
'f(), g()
dx dy
=
F
u
y
k2
'f, (L
k+2
g)` y
k
'(R
k
f), g`
dx dy.
In the integral over T
u
on the left hand side the contributions from SL
2
(Z)-
equivalent boundary pieces cancel. Thus
F
u
y
k
'f(), g()`d =
1/2
1/2
y
k
'f(), g()`
y=u
dx.
Lemma 4.3. Let f, g : H C[L
/L] be smooth functions satisfying f [

k
(M, ) = f and g [
k
(M, ) = g for all (M, ) Mp
2
(Z). Then
F
u
'(
k
f), g`y
k2
dxdy
F
u
'f, (
k
g)`y
k2
dxdy
=
1/2
1/2
'f, (L
k
g)`y
k2
y=u
dx
1/2
1/2
'(L
k
f), g`y
k2
y=u
dx.
Proof. We write
k
= R
k2
L
k
and apply the previous lemma twice. .
Lemma 4.4. For v Gr(L) H(, m) and s = +it C with > 1 k/2
we have the identity
lim
u
F
u
'F
,m
(, s),
k
(
L
(, v)y
l/2
)`y
k
dxdy
y
2
= (l/4 + 1/2 s)(l/4 1/2 +s)
,m
(v, s).
Proof. Since
k
F
,m
(, s) = (l/4 + 1/2 s)(l/4 1/2 +s)F
,m
(, s),
Lemma 4.3 implies that
F
u
'F
,m
(, s),
k
(
L
(, v)y
l/2
)`y
k
dxdy
y
2
(s 1/2 l/4)(s 1/2 +l/4)
F
u
'F
,m
(, s),
L
(, v)y
l/2
`y
k
dxdy
y
2
=
1/2
1/2
'(L
k
F
,m
(, s)),
L
(, v)y
l/2
`y
k2
y=u
dx
1/2
1/2
'F
,m
(, s), (L
k
L
(, v)y
l/2
)`y
k2
y=u
dx. (4.6)
Thus it suces to show that the integral on the right hand side tends to 0
as u . We have
1/2
1/2
'(L
k
F
,m
(, s)),
L
(, v)y
l/2
`y
k2
dx
= i
1/2
1/2
y

x
+i

y
F
,m
(, s),
L
(, v)
dx. (4.7)
If we insert the Fourier expansions
F
,m
(, s) =
/L
nZ+q()
c(, n; y, s)e
(nx),
L
(, v) =
exp(4yq(
v
) + 2yq())e
(q()x),
and carry out the integration, we nd that (4.7) equals
ye
4yq(
v
)+2yq()
2q()c(, q(); y, s) +

y
c(, q(); y, s)
.
Using the explicit formulas for the coecients c(, n; y, s) given in Theorem
1.9 and the asymptotic behavior of the Whittaker functions we may infer
that
lim
y
(4.7) = lim
y
y

y
c(0, 0; y, s) = b(0, 0, s) lim
y
y

y
y
1sk/2
.
Since > 1 k/2 this is equal to 0. In the same way it can be seen that
lim
y
1/2
1/2
'F
,m
(, s), (L
k
L
(, v)y
l/2
)`y
k2
dx = 0.
Taking the limit u in (4.6), we obtain the assertion. .
The following proposition is crucial for our argument.
Proposition 4.5. The Siegel theta function
L
(, Z), considered as a func-
tion on HH
l
, satises the dierential equation
L
(, Z)y
l/2
= 2
L
(, Z)y
l/2
.
Proof. For H and v Gr(L) we dene a function on V = L R by
F
,v
: V C, F
,v
() = e(q(
v
) + q(
v
)).
If Z H
l
corresponds to v via the isomorphism H
l

= Gr(L), then we also
write F
,Z
instead of F
,v
. For any xed V we may consider F
,Z
() as
a function in and Z. To prove the proposition it suces to show that
k
F
,Z
()y
l/2
= 2F
,Z
()y
l/2
.
This essentially follows from [Sn] Lemma 1.5 as we will now indicate. As far
as possible we use the same notation as in [Sn].
We choose a basis for L and identify V with R
l+2
. Denote by Q the Gram
matrix of L with respect to this basis, so that (x, y) = x
t
Qy. Let g r
0
(g, Q)
be the Weil representation of Mp
2
(R) O(V ) in L
2
(V ) as dened in [Sn].
For xed H and Z H
l
the function F
,Z
is rapidly decreasing.
According to [Sn] Lemma 1.2 we have
(r
0
(, Q)F
,Z
)() = ()
2
()
l
F
M,Z
(), = (M, ) Mp
2
(R),
(r
0
(g, Q)F
,Z
)() = F
,Z
(g
1
) = F
,gZ
(), g O
+
(V ).
It is easily seen that F
i,Z
satises the condition (1.19) in [Sn] with m = 2l.
If we put
y
1/2
xy
1/2
0 y
1/2
, 1
Mp
2
(R)
for = x +iy H, then r
0
(
, Q)F
i,Z
= y
(2+l)/4
F
,Z
.
The map (1.1) gives a locally isomorphic embedding of SL
2
(R) into
Mp
2
(R). Thus for any element u of the universal enveloping algebra U of
the Lie algebra of SL
2
(R) we may consider r
0
(u, Q) as a dierential operator
on functions V C. Let C be the Casimir element in U and C
Q
= r
0
(C, Q).
Then by [Sn] Lemma 1.4 we may infer
4
k
y
l/2
F
,Z
() = (1 l
2
/4)y
l/2
F
,Z
() +y
k/2
(r
0
(
, Q)C
Q
F
i,Z
) ().
(4.8)
Let denote the representation of O
+
(V ) on C
(H
l
) given by
((g)f)(Z) = f(g
1
Z), g O
+
(V ).
For V we obviously have
(g)(F
,Z
()) =
r
0
(g
1
, Q)F
,Z
(). (4.9)
Let D be the Casimir element of the universal enveloping algebra of the Lie
algebra of O
+
(V ). Then D
Q
= r
0
(D, Q) is a dierential operator on functions
V C, and (D) a dierential operator on functions H
l
C. According
to [Hel] p. 451 ex. 5 the operator (D) equals (up to the sign) the Laplace
operator on H
l
. By (4.9) we have (D)(F
,Z
()) = (D
Q
F
,Z
)(). Hence
there exists a non-zero constant c
l
, only depending on l, such that
c
l
(F
,Z
()) = (D
Q
F
,Z
)() (4.10)
for all H and V . According to [Sn] Lemma 1.5 for any smooth rapidly
decreasing function F on V the identity
C
Q
F = (D
Q
+l
2
/4 1)F (4.11)
holds. Combining (4.8), (4.10), and (4.11), we nd
4
k
y
l/2
F
,Z
() = c
l
y
l/2
F
,Z
() (4.12)
for any V .
To determine the constant c
l
we consider the above identity (4.12) for
= d K. In this case we have
y
l/2
F
,Z
(d) = y
l/2
exp
y
x
2
1
+y
2
1
q(Y )
,
and also (4.12) becomes much easier. (Here we have to be careful with our
notation: y = () and Y = (y
1
, . . . , y
l
) = (Z).) A straightforward compu-
tation shows that c
l
= 8. This completes the proof. .
Theorem 4.6. For Z H
l
H(, m) and > 1 k/2 the function
,m
(Z, s) is an eigenfunction of the invariant Laplacian on H
l
. More
precisely we have
,m
(Z, s) =
1
2
(s 1/2 l/4)(s 1/2 +l/4)
,m
(Z, s).
Proof. Arguing as in section 2.2 we nd that all iterated partial derivatives
with respect to Z of
F
u
'F
,m
(, s),
L
(, Z)y
l/2
`y
k
dxdy
y
2
converge locally uniformly in Z as u . Therefore we have
,m
(Z, s) = lim
u
F
u
'F
,m
(, s),
L
(, Z)y
l/2
`y
k
dxdy
y
2
.
By Proposition 4.5 this equals
1
2
lim
u
F
u
'F
,m
(, s),
k
(
L
(, Z)y
l/2
)`y
k
dxdy
y
2
.
Using Lemma 4.4, we obtain the assertion. .
Theorem 4.7. For Z H
l
H(, m) the regularized theta integral
,m
(Z)
satises
,m
(Z) =
l
4
b
,m
(0, 0).
Here b
,m
(0, 0) = b
,m
(0, 0, 1 k/2) denotes the (0, 0)-th Fourier coecient
of the Poincare series F
,m
(, 1 k/2).
Proof. Dene a function
R(Y, s) =
2
(q(Y ))
1/2+l/4s
(s l/4)b
,m
(0, 0, s)(2s l/2).
In the proof of Theorem 3.9 we saw that
,m
(Z, s) :=
,m
(Z, s) R(Y, s)
is holomorphic in s near 1k/2. Using Theorem 2.15, it can be shown in the
same way that

,m
(Z, s) is twice continuously dierentiable as a function
in (Z, s) for Z H
l
H(, m) and s varying in a neighborhood of 1 k/2.
This implies that
,m
(Z) = (
s=1k/2
[
,m
(Z, s)]
= (
s=1k/2
,m
(Z, s)
+ (
s=1k/2
[R(Y, s)] . (4.13)
By Theorem 4.6 and Lemma 4.1 we have
,m
(Z, s) =
,m
(Z, s) R(Y, s)
=
1
2
(s 1/2 l/4)(s 1/2 +l/4)
,m
(Z, s)
and therefore
(
s=1k/2
,m
(Z, s)
= 0. (4.14)
In the proof of Theorem 3.9 we found that
p
-estimates 103
(
s=1k/2
[R(Y, s)] = b
,m
(0, 0) log(q(Y )) + const.
Thus, using Lemma 4.1, we get
(
s=1k/2
[R(Y, s)] =
l
4
b
,m
(0, 0). (4.15)
If we insert (4.14) and (4.15) into (4.13), we obtain the assertion. .
Remark 4.8. Theorem 4.6 and Theorem 4.7, together with a regularity result
for elliptic dierential operators on analytic Riemann manifolds, also imply
that the functions
,m
(Z, s),
,m
(Z), and
,m
(Z) are real analytic (see
[Rh] '34).
p
-estimates
Recall that the invariant volume element on the upper half plane H
l
is
given by
dX dY
q(Y )
l
. Let (L) be a subgroup of nite index. As usual we
write L
p
(H
l
/) for the Banach space of equivalence classes of measurable
-invariant functions f : H
l
C with
H
l
/
[f(Z)[
p
dX dY
q(Y )
l
< . (4.16)
To decide whether a function satises this integrability condition we need a
fundamental domain for the action of on H
l
.
We x a coordinate system on H
l
as in section 4.1 page 95. (This depends
on the choice of the vectors z, z
, d, d
and the orthogonal basis d

3
, . . . , d
l
of
D R.) We use the notation of section 4.1.
Let
O
+
Q
(L) = g O
+
(V ); g(L Q) = L Q (4.17)
be the rational orthogonal group of the lattice L.
Denition 4.9. Let t > 0. We dene the Siegel domain o
t
in H
l
to be the
subset of Z = X +iY H
l
satisfying
x
2
1
+x
2
2
+[q(X
D
)[ < t
2
, (4.18a)
1/t < y
1
, (4.18b)
y
2
1
< t
2
q(Y ), (4.18c)
[q(Y
D
)[ < t
2
y
2
1
. (4.18d)
Proposition 4.10. Let (L) be a subgroup of nite index.
i) For any t > 0 and any g O
+
Q
(L) the set
; go
t
o
t
=
is nite.
ii) There exist a t > 0 and nitely many rational transformations
g
1
, . . . , g
a
O
+
Q
(L) such that
o = g
1
o
t
g
a
o
t
is a fundamental domain for the action of on H
l
, i.e. o = H
l
.
Proof. This can be deduced from the general reduction theory for algebraic
groups (cf. [Bl1], [Bl2]). We briey indicate the argument. See [Me] for a more
detailed treatment.
For any algebraic group A we denote by A
+
the component of the identity.
There is a basis e
3
, . . . , e
l
of DQ such that the restriction of q to DQ
has the diagonal Gram matrix C = diag(c
3
, . . . , c
l
) with positive rational
numbers c
j
. Then, with respect to the basis z, d, e
3
, . . . , e
l
,

d, z of L Q, the
quadratic form q has the Gram matrix
F =
0 0 J
0 C 0
J 0 0
,
where J = (
0 1
1 0
). The real orthogonal group
O(V ) = M SL
l+2
(R); M
t
FM = F
of V is dened over Q. The group G = O
Q
(L) of rational points is simply the
subgroup of matrices with rational entries, and is an arithmetic subgroup
of G.
Let I
L
denote the special point z + id + i
d + z V (C) and [I
L
] the
corresponding Element of /
+
P(V (C)). The upper half plane H
l
is iden-
tied with /
+
via Z [q(Z)z + Z + z] as in section 3.2. Here the point
I = (i, i, 0) H
l
is mapped to [I
L
]. The group O
+
(V ) = G
+
R
acts transitively
on /
+
and the stabilizer K of [I
L
] is a maximal compact subgroup
2
. The
assignment Kg g
1
[I
L
] identies K` O
+
(V ) with /
+
.
A minimal parabolic Q-subgroup of G is given by
P =
H
0
H
1
H
2
0 B H
3
0 0 H
4
,
2
Only in this proof K denotes a maximal compact subgroup. Usually it denotes
a sublattice of L.
p
-estimates 105
where H
0
, H
4
are upper triangular and B is an orthogonal transformation in
O
Q
(D), with additional relations that ensure P G. The unipotent radical
U of P is the set of matrices H in P, where H
0
, H
4
are unipotent, B = E,
and
H
4
= H
1
0
, (4.19)
CH
3
+H
t
1
JH
4
= 0, (4.20)
H
t
4
JH
2
+H
t
2
JH
4
+H
t
3
CH
3
= 0. (4.21)
The inverse of a matrix H U equals
H
1
=
H
4
JH
t
3
C JH
t
2
J
0 E C
1
H
t
1
J
0 0 H
0
.
If we write
H
0
=
1 h
12
0 1
, H
1
=
h
13
. . . h
1l
h
23
. . . h
2l
, H
2
=
h
1,l+1
h
1,l+2
h
2,l+1
h
2,l+2
,
then H is completely determined by h
12
, . . . , h
1,l+1
, h
23
, . . . , h
2l
, and these
may be freely chosen. Thus U
R
is homeomorphic to R
2l2
. Equality (4.21)
implies in particular that
h
2,l+1
=
1
2
(c
1
3
h
2
23
+ +c
1
l
h
2
2l
).
For any t > 0 the set
t
=
H U; h
2
12
+h
2
1,l+1
+
1
2
l
j=3
c
1
j
h
2
1j
< t
2
and h
2,l+1
< t
2
is a relatively compact open neighborhood of the identity in U

R
. This can be
checked using the above relations. Moreover, we have U
R
=
t>0
t
.
A maximal Q-split torus S in P is given by the set of diagonal matrices
2
E
1
2
1
1
.
The set of simple roots of G with respect to S and P is equal to =
1
,
2
,
where
1
(T) =
1
/
2
,
2
(T) =
2
for T S as above (cf. [Bl1] p. 80, [Bl2] p. 16). Let t > 0. We denote by A
t
the subset
A
t
= T S
+
R
;
1
(T) < t,
2
(T) < t
of S
+
R
. Then the open subset
S
t
= K A
t

t
of G
R
is a Siegel domain (with respect to K, P, S) in the sense of denition
12.3 in [Bl1]. (Note the remark at the end of '12.3 in [Bl1] and the fact that
S is also a maximal split torus over R.) Now, by a straightforward calculation
it can be veried that S
t
precisely corresponds to o
t
under the identication
K` O
+
(V ) H
l
, Kg g
1
I. (Caution: o
t
is dened with respect to
the basis z, d, d
3
, . . . , d
l
,

d, z of L R, whereas A
t
and
t
are dened with
respect to the basis z, d, e
3
, . . . , e
l
,

d, z.) Therefore assertion (i) follows from
[Bl1] Theorem 15.4 and (ii) from [Bl1] Theorem 13.1. .
Let f : H
l
C be a measurable -invariant function. Proposition 4.10
immediately implies that f L
p
(H
l
/) if and only if

S
[f(Z)[
p dX dY
q(Y )
l
< .
Let t > 0. In the next few lemmas we derive some useful estimates that
hold on o
t
.
Lemma 4.11. Suppose that p < l/2. Then
S
t
q(Y )
p
dX dY
q(Y )
l
< .
Proof. Since the set of X dened by (4.18a) is compact, it suces to prove
that

R
t
q(Y )
pl
dY < ,
where {
t
denotes the set of Y dened by (4.18b)(4.18d). It is easily seen
that
y
1
y
2
1 +t
4
< q(Y ) < y
1
y
2
(4.22)
for Y {
t
. Thus it suces to show that
I =
y
1
>1/t
t
2
q(Y )>y
2
1
|q(Y
D
)|<t
2
y
2
1
(y
1
y
2
)
pl
dy
1
dy
2
dY
D
is nite. We have
p
-estimates 107
I <
t
y
1
>1/t
y
2
/y
1
>1/t
2
|q(Y
D
)|<t
2
y
2
1
(y
1
y
2
)
pl
dy
1
dy
2
dY
D
<
t
y
1
>1/t
y
2
/y
1
>1/t
2
(y
1
y
2
)
pl
y
l2
1
dy
1
dy
2
<
t
y
1
>1/t
u>1/t
2
y
2pl1
1
u
pl
dy
1
du.
Since l 2, the latter integral is nite for p < l/2. .
In particular any Siegel domain has nite volume.
Lemma 4.12. Let p < l 1. Then
S
t
[(, Y )[
p
dX dY
q(Y )
l
< .
for any K R.
Proof. Using Minkowskis inequality and Lemma 4.11, we nd that it suces
to prove that
S
t
y
p
2
q(Y )
l
dX dY < . This can be done as in Lemma 4.11.
.
Lemma 4.13. There exists an > 0 such that
2(, Y )
2
Y
2
2
>
y
2
2
2
1
+y
2
1
2
2
+q(Y )[
2
D
[
for any = (
1
,
2
,
D
) K
and any Y {
t
. Here {
t
denotes the set of Y
dened by (4.18b)(4.18d).
Proof. Let h(, Y ) denote the function
h(, Y ) = 2(, Y )
2
Y
2
2
( K R and iY H
l
). For any real orthogonal transformation g
O
+
(K R) we have h(g, gY ) = h(, Y ). Moreover, if Y is of the form
(y
1
, y
2
, 0), then
h(, Y ) = 2y
2
2
2
1
+ 2y
2
1
2
2
+Y
2
[
2
D
[. (4.23)
Recall that for any isotropic vector u K R and any v K R with
(u, v) = 0 one has the Eichler transformation E(u, v) O
+
(K R), dened
by
E(u, v)(a) = a (a, u)v + (a, v)u q(v)(a, u)u.
For a given Y = (y
1
, y
2
, Y
D
) {
t
we consider the special Eichler transfor-
mations
E = E(d, Y
D
/y
1
),
E = E(
d, Y
D
/y
2
).
It is easily checked that
E(Y ) = (y
1
, y
2
+q(Y
D
)/y
1
, 0) ,
E() =
1
, ,
D

1
y
1
Y
D
E(Y ) = (y
1
+q(Y
D
)/y
2
, y
2
, 0) ,
E() =
,
2
,
D

2
y
2
Y
D
.
Using 4.23 we obtain the following estimate:
h(, Y ) =
1
2
h(E, EY ) +
1
2
h(

E,

EY )
(y
2
+q(Y
D
)/y
1
)
2
2
1
+ (y
1
+q(Y
D
)/y
2
)
2
2
2
+q(Y )
(
D

1
y
1
Y
D
)
2
+q(Y )
(
D

2
y
2
Y
D
)
2
.
Let (0, 1). Then
q(Y )
(
D

1
y
1
Y
D
)
2
+q(Y )
(
D

2
y
2
Y
D
)
2
q(Y )
(
D

1
y
1
Y
D
)
2
+ (
D

2
y
2
Y
D
)
2
= q(Y )
2
D
+
D

1
y
2
+
2
y
1
y
1
y
2
Y
D
2
2
2
y
1
y
2
Y
2
D
q(Y )[
2
D
[ 2[
1
2
[[Y
2
D
[.
Hence we nd that
h(, Y ) (y
2
+q(Y
D
)/y
1
)
2
2
1
+ (y
1
+q(Y
D
)/y
2
)
2
2
2
+q(Y )[
2
D
[ 2[
1
2
[[Y
2
D
[. (4.24)
It is easily seen that [q(Y
D
)[ <
t
4
t
4
+1
y
1
y
2
for Y {
t
. This implies
y
2
+q(Y
D
)/y
1
y
2
/(t
4
+ 1),
y
1
+q(Y
D
)/y
2
y
1
/(t
4
+ 1).
If we insert this into (4.24), we get
p
-estimates 109
h(, Y )
1
2(t
4
+ 1)
2
(y
2
2
2
1
+y
2
1
2
2
) +q(Y )[
2
D
[
+
1
2(t
4
+ 1)
2
y
2
2
2
1
8(t
4
+ 1)
2
y
1
y
2
[
1
2
[ +y
2
1
2
2
.
Thus for <
1
4(t
4
+1)
2
we nally obtain
h(, Y ) (y
2
2
2
1
+y
2
1
2
2
+q(Y )[
2
D
[).
.
Corollary 4.14. There exists an > 0 with the following property: For any
K
with q() 0 and any Y {

t
the inequality
[(, Y )[ (y
2
[
1
[ +y
1
[
2
[)
holds. Here {
t
denotes the set of Y dened by (4.18b)(4.18d).
Proof. For q() 0 one has 2(, Y )
2
2(, Y )
2
2
Y
2
. Hence, according to
Lemma 4.13 there is an > 0 such that
(, Y )
2
(y
2
2
2
1
+y
2
1
2
2
+q(Y )[
2
D
[) (y
2
2
2
1
+y
2
1
2
2
)
for all K
with q() 0 and all Y {

t
. This implies the assertion. .
Proposition 4.15 (Koecher principle). Let f : H
l
C be a holomorphic
function that satises
i) f(Z +k) = f(Z), for k K,
ii) f(Z) = f(Z), for (L) O(K), where O(K) is considered as a
subgroup of O(L).
Then f has a Fourier expansion of the form
f(Z) =
=(
1
,
2
,
D
)K
1
0; q()0
a()e((, Z)).
In particular, f is bounded on the Siegel domain o
t
.
Proof. The rst assertion can be proved in the same way as the Koecher
principle for Siegel modular forms (cf. [Fr1] chapter I Hilfssatz 3.5). Note
that the assumption l 3 is crucial. The group (L) O(K) plays the role
of SL(n, Z) in [Fr1]. Then the second assertion can be deduced using Lemma
4.13 or Corollary 4.14. .
Let t > 0 and g O
+
Q
(L). Let f : H
l
C be a -invariant measurable
function. We now consider the integral
I =
gS
t
[f(Z)[
p
dX dY
q(Y )
l
over the translated Siegel domain go
t
. It is convenient to use the identication
H
l
/
+
, Z [Z + z q(Z)z], and to rewrite I as the integral
g

S
t
[F(W)[
p
d,
where

o
t
= W = [Z + z q(Z)z]; Z o
t
, F denotes the function on /
+
corresponding to f, and d denotes the O
+
(V )-invariant measure on /
+
. We
will work with dierent coordinate systems on /
+
.
A 4-tuple = (
1
, . . . ,
4
) of vectors in L Q is called an admissible
index-tuple, if
i) q(
j
) = 0 for j = 1, . . . , 4,
ii) (
1
,
2
) = 1 and (
3
,
4
) = 1,
iii) (
i
,
j
) = 0 for i = 1, 2 and j = 3, 4,
iv) ((W,
3
)/(W,
1
)) > 0 for W /
+
.
Then Q
1
+Q
2
and Q
3
+Q
4
are orthogonal hyperbolic planes in LQ.
The intersection of (Q
1
+ Q
2
)
and L is a Lorentzian sublattice of L,

which will be denoted by K(). The intersection of (Q
1
+ +Q
4
)
and
L is a negative denite sublattice, denoted by D(). If X L R, then we
write X
K()
resp. X
D()
for the orthogonal projection of X to K() R
resp. D() R.
In the same way as in section 3.2 it can be seen that /
+
is biholomorphic
equivalent to
H
l
() = X +iY K() C; X, Y K() R, (Y,
3
) > 0, q(Y ) > 0,
the isomorphism being given by
H
l
() /
+
, Z [Z +
2
q(Z)
1
].
Let t > 0 and = (
1
, . . . ,
4
) be an admissible index-tuple. We denote by
o
t
() the subset of Z = X +iY H
l
() satisfying
(X,
3
)
2
+ (X,
4
)
2
q(X
D()
) < t
2
, (4.25a)
1/t < (Y,
3
), (4.25b)
(Y,
3
)
2
< t
2
q(Y ), (4.25c)
q(Y
D()
) < t
2
(Y,
3
)
2
. (4.25d)
We dene the Siegel domain

o
t
() attached to and t by
o
t
() = [Z +
2
q(Z)
1
]; Z o
t
() /
+
.
p
-estimates 111
Obviously
0
= (z, z, d,

d) is an admissible index-tuple. We have H
l
(
0
) = H
l
and o
t
(
0
) = o
t
. It is easily seen that the rational orthogonal group O
+
Q
(L)
acts transitively on admissible index-tuples. For g O
+
Q
(V ) the identity
g

o
t
(
1
, . . . ,
4
) =

o
t
(g
1
, . . . , g
4
) (4.26)
holds.
In order to prove that F L
p
(/
+
/), it suces to show that for any
t > 0 and any admissible index-tuple the integral
S
t
()
[F(W)[
p
d (4.27)
is nite.
Lemma 4.16. Let t > 0 and = (
1
, . . . ,
4
) be an admissible index-tuple.
For a, b Q
>0
there exist t
1
, t
2
R with 0 < t
1
< t < t
2
, such that
o
t
1
()

o
t
(a
1
,
1
a
2
, b
3
,
1
b
4
)

o
t
2
().
Proof. The proof is left to the reader. .
Lemma 4.17. Let
1
,
2
L Q with q(
1
) = q(
2
) = 0 and (
1
,
2
) = 1.
Let t
> 0 and
be an admissible index-tuple of the form (

1
, , , ). There
exist a t > 0 and an admissible index-tuple of the form (
1
,
2
, , ) such
that
o
t
(
)

o
t
().
Proof. Write
= (
1
,
2
,
3
,
4
) and put u =
2
2
. Then u LQ with
(u,
1
) = 0. Consider the Eichler transformation E(
1
, u) : L Q L Q,
E(
1
, u)(a) = a (a,
1
)u + (a, u)
1
q(u)(a,
1
)
1
.
It is easily checked that E(
1
, u) O
+
Q
(L). Put = E(
1
, u)
1
=
E(
1
, u)
. Then is clearly an admissible index-tuple. We have

E(
1
, u)(
1
) =
1
,
E(
1
, u)(
2
) =
2
,
E(
1
, u)(k) = k (k,
2
)
1
, k K().
The induced action of E(
1
, u) on H
l
() is simply the translation
E(
1
, u)Z = Z + (
2
)
K()
(Z H
l
()).
Hence there is a t > 0 such that E(
1
, u)o
t
() o
t
(). This implies
o
t
(
) =

o
t
(E(
1
, u)) = E(
1
, u)

o
t
()

o
t
().
.
Lemma 4.18. Let = (
1
,
2
,
3
,
4
) be an admissible index-tuple. More-
over, let t
> 0 and
be an admissible index-tuple of the form (

1
,
2
,
3
, ).
Then there exists a t > 0 such that
o
t
(
)

o
t
().
Proof. This can be proved similarly as Lemma 4.17. .
4.3 Modular forms whose zeros and poles lie on
Heegner divisors
Let F be a meromorphic modular form for (L) of weight r. Suppose that
its divisor is a linear combination of Heegner divisors
3
H(, m). Then there
are integral coecients c(, m) with c(, m) = c(, m) such that
(F) =
1
2
/L
mZ+q()
m<0
c(, m)H(, m).
The function f(Z) = log([F(Z)[q(Y )
r/2
) is (L)-invariant and has logarith-
mic singularities along Heegner divisors. In this section we show that up to
an additive constant f is equal to the regularized theta integral
1
8
/L
mZ+q()
m<0
c(, m)
,m
(Z).
In the proof we use a result due to S.-T. Yau on subharmonic functions on a
complete Riemann manifold that satisfy certain integrability conditions.
Let M be a complete connected Riemann manifold. Write for the
Laplace operator on M and (, ) for the inner product on square integrable
dierential forms on M.
Proposition 4.19. i) Let u be a smooth function on M. Suppose that u and
u are square integrable. Then du is square integrable, too.
ii) The Laplace operator is symmetric in the following sense: Let u and
v be smooth square integrable functions on M. Suppose that u and v are
also square integrable. Then
(u, v) = (u, v).
Proof. In the following proof we use some ideas of [Rh] '34, to simplify the
argument in [Ro].
3
For a precise denition of the Heegner divisor H(, m) see chapter 5.
4.3 Modular forms with zeros and poles on Heegner divisors 113
i) Since M is complete, there exists a proper C
function : M [0, )
and a constant C > 0 such that
[d(x)[ < C
for all x M. Such a function can be obtained by regularization from the
function x d(x, o), where d(x, o) denotes the geodesic distance of x from
a xed point o M (cf. [Rh] '34, '15). For any r > 0 the set B
r
= x
M; (x) r is compact and M =

r>0
B
r
. Let : [0, ) [0, 1]
be a C
function equal to 1 on [0, 1] and 0 on [2, ). Then the function
r
(x) = ((x)/r) has its values in [0, 1]. Moreover, it satises
r
(x) = 1 for
x B
r
,
r
(x) = 0 for x / B
2r
, and
[d
r
(x)[ < D/r, (4.28)
where D > 0 is a constant which is independent of r.
Because
r
has compact support, we have
(u, u
2
r
) = (du,
2
r
du) + (
r
du, 2ud
r
).
Using the inequality [(f, g)[
1
2
(f, f) +
1
2
(g, g), which follows from the
Schwarz inequality, we nd
(
r
du,
r
du) (u, u
2
r
) +
1
2
(
r
du,
r
du) + 2(ud
r
, ud
r
)
(
r
du,
r
du) 2(u, u
2
r
) + 4(ud
r
, ud
r
)
2(u, u
2
r
) + 4D
2
r
2
(u, u).
In the last line we have used (4.28). As r we obtain (du, du) 2(u, u).
So du is square integrable.
ii) In the same way as in (i), since
r
has compact support, we have
(u,
r
v) = (du,
r
dv) + (du, vd
r
).
By (4.28), as r , the second term on the right hand side tends to zero.
This implies (u, v) = (du, dv). .
Theorem 4.20. Let u be a harmonic form on M. Suppose that u L
p
(M)
for some p > 1. Then u is closed and coclosed.
Proof. See [Yau] Proposition 1 on p. 663. The case p = 2 is also treated in
[Rh] Theorem 26. .
A continuous function u on M is called subharmonic, if it satises the local
harmonic maximum principle, i.e. if for any point a M and any smooth
harmonic function h dened in a neighborhood of a, the function uh has a
local maximum at a only if it is constant in a neighborhood of a (cf. [GW]).
If u is smooth, then it is subharmonic if and only if u 0 everywhere on
M. The following theorem is also due to S.-T. Yau (see [Yau] Theorem 3 on
p. 663 and Appendix (ii)).
Theorem 4.21. Let u be a non-negative continuous subharmonic function
on M. Suppose that u L
p
(M) (p > 1). Then u is constant.
Note that Theorem 4.20 for functions immediately follows from Theorem
4.21. If u is a harmonic function, then [u[ is subharmonic in the above sense.
Corollary 4.22. Let M be a complete connected Riemann manifold of nite
volume. Let f L
p
(M) (p > 1) be a smooth function that satises f = c,
where c is a real constant. Then f is constant.
Proof. Assume that f is a solution of f = c. Without any restriction we
may assume that c 0. So f is subharmonic. Let d > 0 be a large real
number. The function f+d is also subharmonic. Because M has nite volume,
f + d L
p
(M). Consider the function g = max(f + d, 0). It is continuous,
nonnegative, and belongs to L
p
(M). Moreover, it is subharmonic in the above
sense, since it is the maximum of two subharmonic functions (cf. [GW]).
According to Theorem 4.21, g is constant. This implies that f is constant on
x M; f(x) d. Since d was arbitrary, we nd that f is constant on
M.
In the case p = 2 we may also argue as follows: By Proposition 4.19 we
have
c
2
vol(M) = (c, c) = (f, c) = (f, c) = 0.
Hence f is a square integrable harmonic function. According to Theorem 4.20
it is constant. .
Theorem 4.23. Let F be a meromorphic modular form of weight r with
respect to (L), whose divisor is a linear combination of Heegner divisors
(F) =
1
2
/L
mZ+q()
m<0
c(, m)H(, m). (4.29)
Dene f(Z) = log([F(Z)[q(Y )
r/2
) (with Y = (Z)) and
(Z) =
1
8
/L
mZ+q()
m<0
c(, m)
,m
(Z).
Then f is constant.
Proof. Let us temporarily assume l 4. First, we will prove that f
L
2
(/
+
/(L)). It suces to show that
S
t
()
[f [
2
d <
for any t > 0 and any admissible index-tuple . According to Lemma 4.16
4.18 we may assume that = (z, z, d,

d), where z L is a primitive norm 0
vector, z
with (z
, z) = 1, z = z
q(z
)z, and d is a primitive norm 0

vector in K = L z
, d
with (d
, d) = 1, and

d = d
q(d
)d.
This is precisely the situation we considered in the previous sections. We may
view F, f, and as functions on H
l
= H
l
() and use the Fourier expansion
of which was determined in section 3.2. We have to prove that
S
t
[f(Z) (Z)[
2
dX dY
q(Y )
l
< .
Dene a meromorphic function on H
l
by
(Z) =
/L
mZ+q()
m<0
,m
(Z)
c(,m)/2
.
By Theorem 3.16 the quotient F/ is a holomorphic function on H
l
without
any zeros. Let
G = Log(F/)
be a holomorphic logarithm and g = 1(G) = log [F/[. Since F is a modular
form for (L) (with some multiplier system), the absolute value of F is K-
periodic, i.e. [F(Z + )[ = [F(Z)[ for any K. The Fourier expansions
(3.38) of the functions
L
,m
(Z) imply that log [[ is also K-periodic. Thus
g is K-periodic, too. By the maximum modulus principle we may infer that
there is a
0
K R, such that
G(Z +) = G(Z) + 2i(
0
, )
for any K. Hence the function
G(Z) = G(Z) 2i(

0
, Z)
is holomorphic on H
l
and K-periodic.
Let (L) O(K). Then acts linearly on H
l
and j(, Z) = 1. This
implies [F(Z)[ = [F(Z)[. According to Theorem 3.16 we have
1(

G(Z)) = log [F(Z)[ + 2(
0
, Y ) +
1
8
,m
c(, m)
L
,m
(Z) C
,m
.
In view of the Fourier expansion (3.38) of the
L
,m
we nd
1(

G(Z)) = 1(

G(Z)) +(Y ),
where (Y ) is (a priori) a piecewise linear real function of Y . Because 1(

G(Z))
and 1(

G(Z)) are real analytic, (Y ) is even linear on H
l
. Thus there exists
a
0
K R with
1(

G(Z)) = 1(

G(Z)) + 2(
0
, Y ).
Again, by the maximum modulus principle, there is a real constant C such
that
G(Z) =

G(Z) 2i(
0
, Z) +iC. (4.30)
Since

G(Z) is K-periodic,

G(Z) is K-periodic, too. Using (4.30), one easily
deduces that
0
= 0. If we insert the Fourier expansion
G(Z) =
a()e((, Z)), a() C,

of

G into (4.30) and compare the constant terms on both sides, we nd C = 0
and thereby
G(Z) =

G(Z).
But now, by the Koecher principle (Proposition 4.15),

G is bounded on o
t
.
We obtain
[g(Z)[ = [1(

G(Z)) 2(
0
, Y )[ 2[(
0
, Y )[ + const. (4.31)
Let us now consider the function
(Z) =
1
8
/L
mZ+q()
m<0
c(, m)
L
,m
(Z).
Using the asymptotic behavior (3.26) of the function 1
, the boundedness
(2.22) of the coecients b(, n) for n < 0, and Lemma 4.13, one easily obtains
that the sum over K
in the denition of
L
,m
(Z) (Def. 3.11) is bounded
on o
t
. By means of the Fourier expansion (3.12) of
K
,m
it can be deduced
that there is a K R with
[Y [
K
,m
(Y/[Y [)
[(, Y )[
on o
t
. Hence there exists a
0
K R such that
[(Z)[ [Blog q(Y )[ +[(
0
, Y )[ + const. (4.32)
on o
t
, where
B =
1
8
/L
mZ+q()
m<0
c(, m)b
,m
(0, 0).
By virtue of (4.31) and (4.32) we nd
[f(Z) (Z)[ =
log [F(Z)/(Z)[ (Z) +

r
2
log(q(Y ))
[g(Z)[ +[(Z)[ +[
r
2
log(q(Y ))[
< [(
0
, Y )[ +[(
0
, Y )[ + ([B[ +
|r|
2
)[ log(q(Y ))[ (4.33)
on o
t
. Using Lemma 4.11 and Lemma 4.12, we obtain that f
L
2
(/
+
/(L)).
Let us now consider the action of the invariant Laplace operator on
f . According to Theorem 4.7 we have
(Z) =
l
4
B.
Moreover, Lemma 4.1 implies that f(Z) =
rl
8
. Hence
(f(Z) (Z)) =
rl
8

l
4
B
is constant. Since H
l
/(L) is a complete Riemann manifold of nite volume,
we may apply Corollary 4.22. We nd that f is constant.
In the case l = 3 the same argument shows that f L
p
(/
+
/(L))
for p < 2. Again, by Corollary 4.22 it has to be constant. .
Corollary 4.24. Let F be a meromorphic modular form of weight r with
respect to (L). Suppose that its divisor is a linear combination of Heegner
divisors as in (4.29). Then
r =
1
4
/L
mZ+q()
m<0
c(, m)q(, m),
where the q(, n) denote the Fourier coecients of the Eisenstein series E
with constant term 2e
0
in M
,L
.
Proof. We use the same notation as in the proof of Theorem 4.23. By Theo-
rem 4.23 we know that f is constant. Thus
1(

G(Z)) 2(
0
, Y ) (Z) +
r
2
log(q(Y )) (4.34)
is constant. Comparing coecients we nd that the 0-th Fourier coecient
2(
0
, Y ) +
[Y [
8
/L
mZ+q()
m<0
c(, m)
K
,m
(Y/[Y [) + (
r
2
+B) log(q(Y ))
of (4.34) is constant.
This implies in particular that the function
2(
0
, Y/[Y [) +
[Y [
8
/L
mZ+q()
m<0
c(, m)
K
,m
(Y/[Y [)
is invariant under the action of the real orthogonal group O
+
(KR). Hence
it is constant.
We may infer that there are constants C, C
R such that
C[Y [ + (
r
2
+B) log(q(Y )) = C
.
But this is only possible, if C = 0 and
r
2
= B. Using Proposition 1.16, we
obtain the assertion. .
Remark 4.25. Let L
/L and m Z + q() with m < 0. If there is no

L
with (, q()) = (, m), then H(, m) = 0 in the divisor group of

H
l
/(L). In this case Theorem 4.23 implies that
,m
is constant.
5 Chern classes of Heegner divisors
Let X be a normal irreducible complex space. By a divisor D on X we mean
a formal linear combination
D =
n
Y
Y (n
Y
Z)
of irreducible closed analytic subsets Y of codimension 1 such that the sup-
port supp(D) =

n
Y
=0
Y is a closed analytic subset of everywhere pure
codimension 1. Then for any compact subset K X there are only nitely
many Y with Y K = and n
Y
= 0. We denote the group of divisors on X
by D(X).
If is a group of biholomorphic transformations of X acting properly
discontinuously, we may consider the inverse image
(D) of a divisor D
on X/ under the canonical projection : X X/. For any irreducible
component Y of the inverse image of supp(D) the multiplicity of Y with
respect to
(D) equals the multiplicity of (Y ) with respect to D. Then
(D) is a -invariant divisor on X. (So we do not take account of possibly

occurring ramication in the denition of
.)
Now let X = H
l
be the generalized upper half plane as in section 3.2, and
let be the orthogonal group (L) or a subgroup of nite index. If L
/L
and m Z +q() with m < 0, then
+L
q()=m
is a -invariant divisor on H
l
with support H(, m) (see (2.17)). It is the
inverse image under the canonical projection of an algebraic divisor on X/,
which we call the Heegner divisor of discriminant (, m). For simplicity we
denote it by H(, m), too. The multiplicities of all irreducible components
equal 2, if 2 = 0, and 1, if 2 = 0. Notice that H(, m) = H(, m).
We will use the following modied divisor class group

Cl(X/): Any
meromorphic modular form f with multiplier system with respect to
denes via its zeros and poles a -invariant divisor in D(X), which is the
inverse image of an algebraic divisor (f) in D(X/). We denote the subgroup
generated by these divisors (f) by

H(X/) and put
120 5 Chern classes of Heegner divisors
Cl(X/) = D(X/)
H(X/).
Moreover, we write Cl(X/) for the quotient of D(X/) modulo the sub-
group of divisors coming from meromorphic modular forms of weight 0 with
trivial character. If acts freely, then Cl(X/) coincides with the usual
notion of the divisor class group.
We now give an algebraic interpretation of Theorem 3.22. First we need
a rational structure on the space S
,L
of cusp forms of weight = 1 +l/2.
Throughout this section let
N = minn N; nq() Z for all L
denote the level of the lattice L. The representation

L
is trivial on
(N) = (M, ) Mp
2
(Z); M 1 (mod N)
and thereby factors through the nite group
Mp
2
(Z/NZ) = Mp
2
(Z)
(N).
If f =
/L
e
S
,L
, then each component f
lies in the space S
(N)
of cusp forms of weight for the group (N). In other words
S
,L
S
(N) C[L
/L].
Let
N
be a primitive N-th root of unity. It is easily seen that the co-
ecients of
L
are contained in Q(
N
). Indeed, it suces to check this on
the generators T and S. For T the assertion is trivial. For S one uses the
evaluation of the Gauss sum
/L
e(q()) =
[L
/L[
i
b
+
b
due to Milgram (cf. [Bo2] Corollary 4.2). The following Lemma is a slight
improvement of [Bo3] Lemma 4.2.
Lemma 5.1. If is integral, then the space S
,L
has a basis of cusp forms
whose Fourier coecients all lie in Z[
N
]. If is half-integral, then S
,L
has
a basis of cusp forms whose coecients all lie in Z[
N
], where N
is the least
common multiple of N and 8.
Proof. First, assume that is integral. We use the following well known
results on integral weight modular forms:
i) The space S
(N) has a basis of cusp forms with coecients in Z[

N
]
(cf. [Sh] section 5 or [DI] Corollary 12.3.9).
ii) If f S
(N) with coecients in Z[1/N,

N
] and (M, ) (N),
then f|
(M, ) also lies in S
(N) and has coecients in Z[1/N,

N
]. Here
5 Chern classes of Heegner divisors 121
f|
(M, ) denotes the usual Petersson operator of weight acting on func-

tions H C. This follows from the q-expansion principle (see [Ka] '1.6 or
[DI] section 12.3).
Consider the trace map
tr : S
(N) C[L
/L] S
,L
, f
(M,)Mp
2
(Z/NZ)
f[
(M, ).
It is obviously surjective. The image of f =
/L
e
(N)C[L
/L]
can also be written in the form
tr(f) =
(M,)Mp
2
(Z/NZ)
L
(M, )
1

L
/L
e
(M, ).
If f has coecients in Z[
N
], then by (ii) tr(f) has coecients in Q(
N
).
Hence by virtue of (i) we may infer that S
,L
has a set of generators with
coecients in Q(
N
). After multiplying by a common denominator, we obtain
a basis with coecients in Z[
N
].
Now assume that is half integral. We claim that (i) and (ii) also hold
in the half integral weight case with N replaced by N
.
To prove the claim we argue similarly as in [SeSt] '5. Let () =
nZ
e
in
2
be the Jacobi theta series of weight 1/2. It is well known that

S
1/2
(8). Moreover, for any (M, ) Mp
2
(Z) the Fourier coecients of
|
1/2
(M, ) lie in Z[
8
]. The space S
(N) can be embedded into S

2
(N
) by
multiplying with
2
. Hence it suces to show that the image can be dened
by linear equations with coecients in Q(
N
). Then the claim follows from
(i), (ii) applied to the space S
2
(N
).
Since has no zeros on H, a cusp form g belongs to the image if and
only if it vanishes with prescribed multiplicities at the cusps of (N
), i.e. if
certain coecients of g in the expansions at the various cusps vanish. By (ii)
these conditions are given by linear equations over Q(
N
).
Now we can proceed as in the integral weight case. .
Let f S
,L
be a cusp form with Fourier coecients c(, n) and
Gal(C/Q) an automorphism of C. Then we dene the -conjugate of f by
f
() =
/L
nZq()
c
(, n)e
(n),
where c
(, n) denotes the conjugate of c(, n). If is integral (resp. half-

integral) and Gal(C/Q(
N
)) (resp. Gal(C/Q(
N
))), then Lemma
5.1 implies that f
is also a cusp form in S

,L
. For general Gal(C/Q) we
only have f
(N) C[L
/L]. In this context it would be interesting to

know, if S
,L
always has a basis with rational Fourier coecients.
Let
o
,L
= Gal(C/Q) S
,L
be the space of all Galois conjugates of the elements of S
,L
and analogously
,L
= Gal(C/Q) M
,L
. If R is a subring of C then we write o
,L
(R)
resp. S
,L
(R) for the R-module of cusp forms in o
,L
resp. S
,L
, whose coef-
cients all lie in R. Observe that Lemma 5.1 implies
o
,L
=
[Gal(Q(
N
)/Q) S
,L
(Q(
N
))] C, for Z,
[Gal(Q(
N
)/Q) S
,L
(Q(
N
))] C, for
1
2
Z.
In particular o
,L
By construction o
,L
has a basis of cusp forms whose coecients are
rational integers. Hence o
,L
(Z)
Z
C = o
,L
. We will also consider the dual
Z-module o
,L
(Z) of o
,L
(Z). Special elements are the functionals
a
,m
: o
,L
(Z) Z; f =
,n
c(, n)e
(n) a
,m
(f) = c(, m)
( L
/L and m Z q(), m > 0). We write /

,L
(Z) for the submodule
of o
,L
(Z) consisting of nite linear combinations

,m
c(, m)a
,m
, where
c(, m) Z and c(, m) = c(, m). It is easily seen that /
,L
(Z) has nite
index in o
,L
(Z) and
/
,L
(Z)
Z
C = o
,L
,
where o
,L
denotes the C-dual space of o
,L
.
Theorem 5.2. Let A
L
= H
l
/(L). The assignment a
,m

1
2
H(, m)
denes a homomorphism
: /
,L
(Z)
Cl(A
L
). (5.1)
Proof. We have to show that is well dened. Suppose that
a =
/L
mZq()
m>0
c(, m)a
,m
(c(, m) Z and c(, m) = c(, m)) is a nite linear combination which is
equal to 0 in /
,L
(Z). Then we may also consider a as an element of the dual
space of S
,L
. According to Theorem 1.17 there exists a nearly holomorphic
modular form f of weight k = 1 l/2 with principal part
/L
mZq()
m>0
c(, m)e
(m).
Now Theorem 3.22 implies that there is a meromorphic modular form of
rational weight c(0, 0)/2 for (L) with divisor
5 Chern classes of Heegner divisors 123
1
2
/L
mZq()
m>0
c(, m)H(, m),
in other words (a) = 0 in

Cl(A
L
). .
As before, let X = H
l
be the generalized upper half plane and be
the orthogonal group (L) or a subgroup of nite index. Write A
L
for the
quotient H
l
/(L).
Let us briey recall some basic facts on Chern classes and the cohomology
of A
L
. For any divisor D on X/ one has a corresponding sheaf /(D). The
sections of /(D) over an open subset U X/ are meromorphic functions
f with (f) D on U.
We now temporarily assume that acts xed point freely on X. Then
X/ is an analytic manifold and every divisor D on X/ a Cartier divisor,
i.e. /(D) is a line bundle. The Chern class
c(D) = c(/(D)) H
2
(X/, C)
of /(D) can be constructed as follows: One chooses a meromorphic function f
on X such that (f) equals the inverse image
(D) of D under the canonical

projection . Then
J(, z) =
f((z))
f(z)
( )
is an automorphy factor, i.e. a 1-cocycle of in the ring of holomorphic
invertible functions on X. Hence a Hermitean metric on the bundle /(D) is
given by a positive C
-function h : X R with
h(z) = [J(, z)[h(z) for all .
Then the dierential form =
log(h) is -invariant and closed. It denes

(via de Rham isomorphism) a cohomology class in H
2
(X/, C). This is the
Chern class of D in the case that acts xed point freely on X.
In the general case one chooses a normal subgroup
0
of nite in-
dex and obtains the Chern class c(D) by the isomorphism H
2
(X/, C)

=
H
2
(X/
0
, C)
/
0
.
The construction of the Chern class gives rise to a homomorphism
c : Cl(A
L
) H
2
(A
L
, C)
into the second cohomology. Using the results of the previous chapters, we
may determine the images of the divisors H(, m) explicitly.
Before stating the theorem let us give an easy example. Let F : H
l
C
be a modular form of weight r with respect to (L). Then [F(Z)[q(Y )
r/2
is invariant under (L). We may view F as a trivialization of the inverse
image /(
((F))) of the sheaf attached to (F) D(A

L
), and q(Y )
r/2
as
a Hermitean metric on /((F)). Hence the Chern class of the divisor (F) is
given by
c((F)) =
r
2
log q(Y ),
a constant multiple of the K ahler class.
Theorem 5.3. The (1, 1)-form
h
,m
(Z) =
1
4
L
,m
(Z)
is a representative of the Chern class c(H(, m)) of the Heegner divisor
H(, m). It has the Fourier expansion
h
,m
(Z) =
[Y [
4
2

K
,m
(Y/[Y [)
b(0, 0)
4

log(Y
2
)
+
1
2
q()<0
0
/L
p()=+K
b(, q())
n1
1
n
e(n(, z
))
1
2k
(n[[[Y [, n(, Y )) e(n(, X)).
Here b(, n) = b(, n, 1 k/2) denote the Fourier coecients of the Poincare
series F
L
,m
(, 1 k/2) as in Proposition 1.10.
Proof. By Theorem 3.16 the function
,m
is meromorphic on H
l
. Its divisor
equals
(H(, m)). Since

L
,m
is real valued and real analytic, the function
e
L
,m
/4
is real analytic and positive. Moreover, by construction
[
,m
(Z)[e
L
,m
(Z)/4
= e
C
,m
/4
e
L
,m
(Z)/4
is invariant under (L). Thus e
L
,m
/4
denes a Hermitean metric on the sheaf
/(H(, m)). This implies the assertion. .
Remark 5.4. The above theorem, together with Theorem 2.12 on the singu-
larities of
,m
(Z), and Theorem 3.9 on its Fourier expansion, imply that
,m
(Z) is something as a Green current for the divisor H(, m) in the sense
of Arakelov geometry (see [SABK]). More precisely this function should dene
a Green object with log-log-growth in the extended arithmetic intersection
theory due to Burgos, Kramer, and K uhn [BKK]. It will be interesting to
investigate this connection in future.
5.1 A lifting into the cohomology 125
5.1 A lifting into the cohomology
Denote by

H
2
(A
L
, C) the quotient
H
2
(A
L
, C) = H
2
(A
L
, C)
log q(Y )
of the second cohomology H
2
(A
L
, C) and the span of the K ahler class
log q(Y ). The Chern class map Cl(A

L
) H
2
(A
L
, C) induces a homo-
morphism
c :
Cl(A
L
)

H
2
(A
L
, C).
If we combine c with the map constructed in Theorem 5.2, we obtain a
homomorphism
/
,L
(Z)

H
2
(A
L
, C).
After tensoring with C one gets a linear map
o
,L

H
2
(A
L
, C), (5.2)
which is characterized by a
,m

1
2
c(H(, m)). The image is the subspace
spanned by the classes of Heegner divisors. Dene a function () by
() =
1
2
/L
mZq()
m>0
c(H(, m))e
(m). (5.3)
Then () is a priori a formal power series with coecients in the subspace
of

H
2
(A
L
, C) spanned by the Heegner divisors. It is a consequence of the
existence of the Borcherds lift (Theorem 3.22), Serre duality (as in Theorem
1.17), and Lemma 5.1 that () is in fact a modular form in
,L
with
values in

H
2
(A
L
, C). This argument is due to Borcherds (see [Bo3]). Hence
() can be viewed as a kernel function for the map (5.2). The image of a
functional a o
,L
is given by
a(())

H
2
(A
L
, C).
The purpose of this section is to rene the map (5.2) and to describe it
more precisely. First we will replace the image space

H
2
(A
L
, C) by a certain
space of harmonic (1, 1)-forms.
Let

Cl
H
(A
L
) denote the subgroup of

Cl(A
L
) generated by the classes of
the Heegner divisors H(, m). Write
1
1,1
(A
L
) (5.4)
for the vector space of square integrable harmonic (1, 1)-forms on A
L
. Ac-
cording to [Bl3] (Theorem 6.2 and '7) 1
1,1
(A
L
) is a nite dimensional space
of automorphic forms in the sense of [Ha]. In particular its elements are :-
nite, where : denotes the center of the universal enveloping algebra of the
Lie algebra of O(V ). By Theorem 4.20 any square integrable harmonic form
on a complete Riemann manifold is closed and thereby denes a cohomology
class via de Rham isomorphism.
Recall the basic facts on the Riemann geometry of H
l
summarized in
section 4.1. We saw that the l l matrix
h(Z) = h(Y ) =
1
4

2
y
i
y
j
log(q(Y ))
i,j
denes the (up to a constant multiple unique) O
+
(V )-invariant Hermitean
metric on H
l
and thereby a Hermitean metric on A
L
. Let h
1
(Y ) =
h
ij
(Y )
i,j
be the inverse of h(Y ). A computation shows that
h
1
(Y ) = 4
y
1
y
1
. . . . . . y
1
y
l
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
y
l
y
1
. . . . . . y
l
y
l
+ 2
0 2q(Y )
2q(Y ) 0
q(Y )
.
.
.
q(Y )
.
The natural scalar product for (1, 1)-forms on A
L
is given by
(f(Z) dz
i
d z
j
, g(Z) dz
m
d z
n
) = 4
X
L
f(Z)g(Z)h
im
(Y )h
jn
(Y )
dX dY
q(Y )
l
.
On the Siegel domain o
t
in H
l
(see Denition 4.9) the components of h
1
(Y )
satisfy the estimate
h
ij
(Y ) < q(Y ), if (i, j) = (2, 2).
The Kahler form
log q(Y ) is harmonic by (4.3). We leave it to the reader

to show that it is square integrable. We put
1
1,1
(A
L
) = 1
1,1
(A
L
)
log q(Y ). (5.5)

Notice that the natural map

1
1,1
(A
L
)

H
2
(A
L
, C) is in general not injec-
tive.
The following strengthening of Theorem 5.3 is crucial for our argument.
Theorem 5.5. i) The (1, 1)-form h
,m
(Z) =
1
4
L
,m
(Z) is a harmonic
square integrable representative of the Chern class of H(, m).
ii) The assignment
H(, m) h
,m
(Z)
denes a homomorphism

Cl
H
(A
L
)

1
1,1
(A
L
) such that the diagram
Cl
H
(A
L
)
//
c
&&
L
L
L
L
L
L
L
L
L
L
1
1,1
(A
L
)
H
2
(A
L
, C)
commutes.
Proof. i) Since
L
,m
(Z) =
L
,m
(Z), the (1, 1)-form
L
,m
(Z) is in-
variant under the action of (L). By Theorem 4.7 we know that
L
,m
is
constant. Because A
L
is K ahlerian, the Laplace operator commutes with
and

. Thus
L
,m
is harmonic.
We now prove that
L
,m
is square integrable. In the same way as in the
proof of Theorem 4.23 it suces to show that
L
,m
is square integrable on
any Siegel domain o
t
() (where is an admissible index tuple and t > 0).
By Lemmas 4.164.18 we may assume that = (z, z, d,

d), where z L is
a primitive norm 0 vector, z
with (z
, z) = 1, z = z
q(z
)z, and d
is a primitive norm 0 vector in K = L z
, d
with (d
, d) = 1,
and

d = d
q(d
)d. We put D = K d
and use the coordinates on

H
l
= H
l
() introduced in section 4.1 on page 95.
We consider the Fourier expansion of
L
,m
(see Theorem 5.3). The
constant term involves the function
K
,m
(Y/[Y [) (see Denition 3.3). Here the
contribution
D
,m
can be evaluated by Theorem 2.14. If we put everything
together we nd that
L
,m
(Z) is equal to
2
q()<0
0
/L
p()=+K
b(, q())
n1
1
n
e(n(, z
))
(n[[[Y [, n(, Y )) e(n(, X))

b(0, 0)
log(Y
2
) + 4
+D=p
2
()
q()=m
1
y
1
(, Y
D
)
2

2
Y
2
D
l 2
+
4
0
/K
p()=+D
b(, q())
n1
1
n
2
y
1
n[[[Y [
y
1
1
K
1
2n[[[Y [
y
1
n
(, Y
D
)
y
1
+n(, d
. (5.6)
(Since the computation is similar to the argument in the proof of Theorem
5.8 we omit the details.)
According to (2.22) the coecients b(, n) satisfy
b(, n) = O(1), n . (5.7)
Moreover, we will use the following estimates. Let > 0 and a, b N
0
. Then
b
1
(A, B) <
,a,b
e
2(1)
A
2
+B
2
(5.8)
uniformly on

A
2
+B
2
> (compare (3.26)), and
a
K
1
(y) <
,a
e
y
(5.9)
uniformly on y > (see [AbSt] chapter 9).
By means of (5.7), (5.8), and Lemma 4.13 one can show that the rst sum
over K
in (5.6) is rapidly decreasing on o

t
. Hence it is square integrable
over o
t
. The term b(0, 0)
log(Y
2
) is a multiple of the K ahler form. It is also
square integrable.
We now prove that the third term in (5.6) is square integrable over o
t
. For
l = 3 it vanishes identically. Hence we may temporarily assume that l 4. It
suces to show that
2
y
i
y
j
1
y
1
(, Y
D
)
2

2
Y
2
D
l 2
dz
i
d z
j
is square integrable for all i, j = 1, . . . , l. On o
t
the estimate
2
y
i
y
j
1
y
1
(, Y
D
)
2

2
Y
2
D
l 2
<
1/y
1
, if i = 2 and j = 2,
0, if i = 2 or j = 2,
holds. Thus it suces to show that the integral
I =
S
t
y
2
1
h
ii
(Y )h
jj
(Y )
dX dY
q(Y )
l
is nite for all i = 2 and j = 2. We have
I <
S
t
y
2
1
q(Y )
2
dX dY
q(Y )
l
<
S
t
y
2
2
dX dY
q(Y )
l
.
The latter integral is nite by Lemma 4.12.
Let us now consider the last term in (5.6). We prove that for any i, j =
1, . . . , l the (1, 1)-form
f
ij
dz
i
d z
j
=
0
/K
p()=+D
b(, q())
n1
1
n
2
2
y
i
y
j
y
1
n[[[Y [
y
1
1
K
1
2n[[[Y [
y
1
n
(, Y
D
)
y
1
+n(, d
dz
i
d z
j
(5.10)
is square integrable over o
t
. There is a > 0 such that
[[[Y [/y
1
> (5.11)
for all Z o
t
and all D
0. Let b Z. It is a consequence of (5.9) and

(5.11) that there exists a polynomial P(Y, ) of degree r in Y and degree r
in such that
2
y
i
y
j
y
1
[[[Y [
y
1
1
K
1
2[[[Y [
y
1
e((, Y
D
)/y
1
)
<
[P(Y, )[
y
r+1
1
y
2
1
[
2
[Y
2
b
e
|||Y |/y
1
for all Z o
t
and all D
0. Using (4.22) we nd that the right hand

side satises the estimate
[P(Y, )[
y
r+1
1
y
2
1
[
2
[Y
2
b
e
|||Y |/y
1
<
1
y
1
[P(Y, )[
y
r
2
[[
r
y
1
y
2
br
1
[
2
[
br
/2
e
|||Y |/y
1
<
1
y
1
y
1
y
2
br
1
[
2
[
br
/2
e
|||Y |/y
1
.
Hence the series in (5.10) has (up to a multiplicative constant) the majorant
n1
1
n
2
1
y
1
y
1
y
2
br
1
n
2
[
2
[
br
/2
e
n|||Y |/y
1
on o
t
. If we choose b large enough a priori, then this is bounded by y
1
/y
2
2
on
o
t
. Thus we nd
S
t
[f
ij
(Z)[
2
h
ii
(Y )h
jj
(Y )
dX dY
q(Y )
l
<
S
t
y
2
1
y
4
2
h
ii
(Y )h
jj
(Y )
dX dY
q(Y )
l
<
S
t
q(Y )
dX dY
q(Y )
l
.
The latter integral is nite for l 3 by Lemma 4.11. This concludes the proof
of the rst assertion.
ii) In order to prove that the above homomorphism is well dened it
suces to show: For any meromorphic modular form F whose divisor
(F) =
1
2
/L
mZ+q()
m<0
c(, m)H(, m)
(c(, m) Z with c(, m) = c(, m)) is a linear combination of Heegner
divisors, the corresponding linear combination
,m
c(, m)
L
,m
(Z)
equals zero in

1
1,1
(A
L
).
By Theorem 4.23 we have
log
[F(Z)[q(Y )
r/2
=
1
8
,m
c(, m)
L
,m
(Z),
where r denotes the weight of F. This implies that
,m
c(, m)
L
,m
(Z) =
,m
c(, m)
L
,m
(Z)
= 8
log
[F(Z)[q(Y )
r/2
= 4r
log (q(Y ))
is a multiple of the K ahler class. .
We now consider the generating series
L
(Z, ) =
1
4
log(Y
2
)e
0
+
1
2
/L
mZq()
m>0
h
,m
(Z)e
(m) (5.12)
of the special representatives h
,m
(Z) of the Chern classes of the Heegner
divisors. (The presence of the constant term will become clear in the proof of
Theorem 5.8.) By a similar argument as in the proof of Theorem 5.5 it can
be shown that
L
(Z, ) converges normally and is square integrable in Z.
Hence we may consider it as an element of 1
1,1
(A
L
) C[L
/L] for xed .

In the rest of this section we shall show that for xed Z the components
of
L
(Z, ) are contained in the space M
,L
. Thus
L
(Z, ) can be viewed
as the kernel function of a lifting
: S
,L
1
1,1
(A
L
); f
f(),
L
(Z, )

from elliptic cusp forms to 1
1,1
(A
L
). The idea of the proof is the same as
in [Za]: We write
L
(Z, ) as an explicit linear combination of Poincare and
Eisenstein series in M
,L
. This description is crucial for our further argument.
It can be used to evaluate the scalar product of
L
(Z, ) and a cusp form
f S
,L
explicitly. Thereby one obtains the Fourier expansion of (f) in
terms of the Fourier coecients of f. The map is one possible generalization
of the Doi-Naganuma map [DN, Na, Za]. In the O(2, 3)-case of Siegel modular
forms of genus 2 such a generalization was given by Piateskii-Shapiro using
representation theoretic methods ([PS1, PS2], see also [Wei]).
For technical reasons we rst introduce two linear operators between the
spaces S
,L
and S
,K
where K is a sublattice of L as in section 2.1.
Lemma 5.6. Let F =
/K
e
be a modular form in M
,K
. Dene a
function F [
L
K
: H C[L
/L] by
F [
L
K
=
0
/L
e
f
p()
.
Then F [
L
K
M
,L
.
Proof. Denote the components of F [
L
K
by g
( L
/L). It suces to check

the transformation behavior for the generators T, S of Mp
2
(Z). For T this is
immediately veried. Regarding S we have to show that
g
(1/) =
i
l2
[L
/L[
/L
e((, ))g
().
The right hand side obviously equals
i
l2
[L
/L[
0
/L
e((, ))f
p()
(). (5.13)
A set of representatives for L
0
/L is given by = (, )z/N+bz/N, where
runs through a set of representatives for K
/K and b runs modulo N. We

nd
(5.13) =
i
l2
[L
/L[
/K
e ((, (, )z/N)) f
()
b (N)
e((, z)b/N).
The latter sum is 0 if / L
0
/L, and N if L
0
/L. In the second case we
obtain
(5.13) =
i
l2
[K
/K[
/K
e ((p(), )) f
()
= f
p()
(1/) = g
(1/).
.
Lemma 5.7. Let F =

L
/L
e
be a modular form in M
,L
. Dene a
function F [
L
K
: H C[K
/K] by
F [
L
K
=
/K
e
0
/L
p()=
f
.
Then F [
L
K
M
,K
.
Proof. This can be proved in the same way as Theorem 2.6. .
It is easily seen that the operators
L
K
and
L
K
are adjoint with respect to
the Petersson scalar product. Moreover, they take cusp forms to cusp forms.
For the rest of this section we x a decomposition of the lattice L. Let
z L be a primitive norm 0 vector and dene z
, N, L
0
, p as in section
2.1. Then the lattice K = L z
has signature (1, l 1) and L R =

(K R) Rz
Rz. We assume that there also exists a primitive norm 0

vector d K and dene d
, M, K
0
, p analogous to z
, N, L
0
, p. Then the
lattice D = Kd
is negative denite and KR = (DR)Rd
Rd.
If LR then we write
K
resp.
D
for the orthogonal projection of to
K R resp. D R. Recall that k = 1 l/2 and = 1 +l/2.
L
(Z, ) dened in (5.12) can be written in
the form
L
(Z, ) =
K
(Y, ) [
L
K
+
1
8
log(Y
2
)E
L
()
1
4
( 1)
q()<0
0
/L
p()=+K
n1
1
n
e(n(, z
))
(n[[[Y [, n(, Y )) e(n(, X))P

L
,q()
(), (5.14)
where
K
(Y, ) =
1
8
/K
mZq()
m>0
[Y [
K
,m
(Y/[Y [)e
(m). (5.15)
The function
K
(Y, ) is equal to
(d, Y )
1
D
(Y
D
, ) [
K
D

1
2( 1)
(d, Y )
0
/K
p()=+D
n1
1
n
2
n[[[Y [
(d, Y )
1
K
1
2n[[[Y [
(d, Y )
n
(, Y )
(d, Y )
+n(, d
P
K
,q()
().
(5.16)
Here
D
(Y
D
, ) S
,D
denotes the theta series
D
(Y
D
, ) =
1
2
/D
+D
(, Y
D
)
2

2
Y
2
D
l 2
(q())
attached to the negative denite lattice D and the harmonic polynomial
(, Y
D
)
2

2
Y
2
D
l2
. Furthermore, P
K
,n
resp. P
L
,n
denote the Poincare series
in S
,K
resp. S
,L
, and E
L
the Eisenstein series with constant term 2e
0
in
M
,L
(cf. section 1.2).
In particular, for xed Z the components of
K
(Y, ) lie in S
,K
. Simi-
larly
L
(Z, ) can be viewed as an element of 1
1,1
(A
L
) M
,L
.
Proof. Throughout the proof we write b
L
,m
(, n) instead of b(, n, 1 k/2)
for the (, n)-th Fourier coecient of the Poincare series F
L
,m
(, 1 k/2) to
emphasize the dependence on L and (, m).
To prove (5.14) we insert the Fourier expansions of the dierential forms
h
,m
(Z) into (5.12) and exchange the order of summation. We nd
L
(Z, ) =
1
8
0
/L
mZq()
m>0
[Y [
K
p(),m
(Y/[Y [)e
(m)
+
1
4
log(Y
2
)e
0
1
8
log(Y
2
)
/L
mZq()
m>0
b
L
,m
(0, 0)e
(m)
+
1
4
q()<0
0
/L
p()=+K
n1
1
n
e(n(, z
))
(n[[[Y [, n(, Y ))
e(n(, X))
/L
mZq()
m>0
b
L
,m
(, q())e
(m). (5.17)
According to Proposition 1.16 one has
2e
0
/L
mZq()
m>0
b
L
,m
(0, 0)e
(m) = E
L
(),
/L
mZq()
m>0
b
L
,m
(, q())e
(m) =
1
( 1)
P
L
,q()
().
If we put this into (5.17), we obtain (5.14).
Using (3.12) it can be seen in the same way that
K
(Y, ) is equal to
1
16
0
/K
mZq()
m>0
Y
2
(d, Y )
2(d
, Y )
D
p(),m
+
8
(Y, d)
+D=p()
q()=m
(, Y )
2
(m)
1
2( 1)
0
/K
p()=+D
n1
(d, Y )
n
2
n[[[Y [
(d, Y )
1
K
1
2n[[[Y [
(d, Y )
n
(, Y )
(d, Y )
+n(, d
P
K
,q()
(). (5.18)
Since the
-operator annihilates all terms which are linear in Y , the rst

summand in (5.18) is equal to
1
16
(d, Y )
1

K
0
/K
mZq()
m>0
Y
2
D
D
p(),m
+ 8
+D=p()
q()=m
(, Y
D
)
2
(m).
(5.19)
According to Theorem 2.14 we have
D
p(),m
=
16m
l 2
# p() +D; q() = m .
Thus (5.19) can be rewritten as
(d, Y )
1

K
0
/K
mZq()
m>0
+D=p()
q()=m
(, Y
D
)
2

2
Y
2
D
l 2
(m)
=

2
(d, Y )
1

K
0
/K
p()+D
(, Y
D
)
2

2
Y
2
D
l 2
(q())
=
(d, Y )
1
D
(Y
D
, ) [
K
D
.
Putting this into (5.18) we get (5.16). The fact that
D
(Y
D
, ) is a cusp form
in S
,D
follows from the usual theta transformation formula ([Bo2] Theorem
4.1). .
Theorem 5.9. The lifting
: S
,L
1
1,1
(A
L
); f
f(),
L
(Z, )

has the following properties:
i) Let f =
,n
c(, n)e
(n) be a cusp form in S

,L
. Then the image of
f has the Fourier expansion
(f)(Z) =
0
(f)(Y )2
1/2
q()<0
n1
[[
22
n
1

L
0
/L
p()=+K
e(n(, z
))
c(, q())
(n[[[Y [, n(, Y )) e(n(, X)),

where the 0-th Fourier coecient
0
(f) is given by
0
(f)(Y ) =
f() [
L
K
,
K
(Y, )
.
ii) The diagram
S
,L
//
S
,L
//
1
1,1
(A
L
)
,L
OO
//
Cl
H
C
//

1
1,1
(A
L
)
(5.20)
commutes. Here o
,L
S
,L
is given by the restriction of functionals, and
S
,L
is identied with S
,L
by means of the Petersson scalar product (, f)
f. The right vertical arrow denotes the canonical projection.
Proof. i) This is an immediate consequence of Theorem 5.8 and Proposition
1.5.
ii) By Theorem 5.2 and Theorem 5.5 the image of the functional a
,m

o
,L
under the maps in the lower line is given by
1
2
h
,m
(Z). On the other
hand its image in S
,L
equals
(4m)
1
2( 1)
P
,m
.
According to Theorem 5.8, for xed Z the function (Z, )
1
8
E()
log(Y
2
)
is a cusp form. Using Proposition 1.5 we nd that the image of a
,m
in
1
1,1
(A
L
) is equal to
1
2
h
,m
(Z) +
1
8
b
,m
(0, 0)
log(Y
2
).
5.1.1 Comparison with the classical theta lift
We now compare the map constructed above with the theta lifting from el-
liptic cusp forms to holomorphic cusp forms on the orthogonal group O
+
(V )
due to Oda [Od] and Rallis-Schimann [RS]. We use the description of
Borcherds [Bo2], because it ts most easily into our setting. Let us rst recall
how holomorphic cusp forms of weight l contribute to the cohomology of A
L
.
It is easily veried that the determinant of the Jacobi matrix of the trans-
formation Z (Z) ( O
+
(V )) at Z H
l
is equal to
j(, Z)
l
(see [Fr1] chapter I Hilfssatz 1.6 for the case of Siegel modular forms). Thus,
if f is a holomorphic modular form of weight l for (L), then
f
= f dz
1
dz
l
is a holomorphic dierential form of type (l, 0), which is invariant under (L).
If f is a cusp form, then
f
is square integrable.
If r Q, we denote by S
r
((L)) the space of holomorphic cusp forms of
weight r for the group (L). Moreover, we write 1
l,0
(A
L
) for the space of
square integrable holomorphic l-forms on the quotient A
L
.
Lemma 5.10. The assignment f
f
denes an isomorphism
S
l
((L))

//
1
l,0
(A
L
).
Proof. This can be proved in the same way as Satz 2.6 in chapter III of [Fr1].
.
Note that the natural map 1
l,0
(A
L
) H
l
(A
L
, C) to the middle coho-
mology of A
L
is injective.
We write L for the lattice L as a Z-module, but equipped with the
quadratic form q(). Then the representation
L
is the dual representation
of
L
. This follows immediately from the denition.
The classical theta lift of Oda and Rallis-Schimann in weight = 1+l/2
is a map from S
,L
to S
l
((L)). It can be viewed as a lifting
: S
,L
1
l,0
(A
L
)
or as a lifting to the middle cohomology of A
L
. According to Theorem 14.3
in [Bo2], if f S
,L
with Fourier coecients c(, n) ( L
/L and n Z+
q()), then there is a square integrable holomorphic l-form (f) 1
l,0
(A
L
)
with Fourier expansion
(f)(Z) =
q()>0
(,d)>0
n|
n
l1

L
0
/L
p()=/n+K
e(n(, z
))c(, q()/n
2
)e((, Z)) dz
1
. . . dz
l
.
Here the second sum

n|
runs over all n N with /n K
. The map
5.2 Modular forms with zeros and poles on Heegner divisors II 137
: S
,L
1
1,1
(A
L
)
(cf. Theorem 5.9) can be regarded as a lifting to the second cohomology of
A
L
. If f S
,L
with Fourier coecients c(, n) ( L
/L and n Zq()),
then there exists a square integrable harmonic (1, 1)-form (f) with Fourier
expansion
(f)(Z) =
0
(f)(Y )2
1/2
q()<0
[[
l
n|
n
l1

L
0
/L
p()=/n+K
e(n(, z
))
c(, q()/n
2
)
([[[Y [, (, Y )) e((, X)).

The similarity of the Fourier expansions of the two liftings is striking. In
both cases the arithmetic part of the coecient with index (and q() = 0)
is given by
n|
n
l1

L
0
/L
p()=/n+K
e(n(, z
))c(, [q()[/n
2
).
We see that in some cases an elliptic modular formf S
,L
can contribute
to both, the middle cohomology and the second cohomology of A
L
. For in-
stance, if the level of L is 1 or 2, then
L
=
L
and therefore S
,L
= S
,L
.
In a series of papers Kudla and Millson constructed liftings from the
cohomology with compact support of locally symmetric spaces associated to
orthogonal (or unitary) groups to classical holomorphic Siegel modular forms
on congruence subgroups [KM1, KM2, KM3]. Serious diculties arise, if one
tries to generalize their construction to obtain liftings from the cohomol-
ogy with arbitrary support to classical modular forms. In particular the lifts
are no longer holomorphic and the geometric meaning of the Fourier coef-
cients of the non-holomorphic part remains unclear. A rst step towards
such a generalization was made by J. Funke in his thesis [Fu]. In our case the
(generalized) lifting of Kudla and Millson would go from H
2
(A
L
, C) to (non-
holomorphic) elliptic modular forms of weight . One should expect that it
is in some sense adjoint to our map . It would be interesting to understand
this connection in detail.
5.2 Modular forms whose zeros and poles lie on
Heegner divisors II
We pick up again our study of meromorphic modular forms whose divisor is
a linear combination of Heegner divisors. In section 4.3 we saw that such a
modular form F is (up to an additive constant) equal to a regularized theta
lift of a linear combination of Maass-Poincare series F
,m
.
In this section we show, how Theorem 5.9 can be used to answer the
question, whether F is already the Borcherds lift of a nearly holomorphic
modular form as in Theorem 3.22.
Recall that a hyperbolic plane is a lattice which is isomorphic to the lattice
Z
2
with the quadratic form q((a, b)) = ab.
We dene two subspaces of S
,L
by
S
,L
= f =
,n
c(, n)e
(n) S
,L
; c(, q()) = 0 for all L
,
(5.21)
S
+
,L
=
,L
. (5.22)
If the lattice L splits a hyperbolic plane over Z, then S
,L
= 0 and S
+
,L
=
S
,L
. It is an immediate consequence of Theorem 5.9 (i) that the lifting
vanishes identically on S
,L
.
Theorem 5.11. Assume that the restricted map
: S
+
,L

1
1,1
(A
L
)
is injective. Then every meromorphic modular form F with respect to (L),
whose divisor is a linear combination of Heegner divisors
(F) =
1
2
q()<0
c()H(, q()) (5.23)
(c() Z with c() = c() and c() = 0 for all but nitely many ),
is a Borcherds product; i.e. there exists a nearly holomorphic modular form
f M
!
k,L
with principal part
q()<0
c()e
(q()) (5.24)
such that F is the Borcherds lift of f as in Theorem 3.22.
Proof. Let F be a meromorphic modular form for the group (L) with divisor
(5.23). Then
1
2
c()c(H(, q())) = 0
in

1
1,1
(A
L
). According to Theorem 5.9 this implies that
c()(q())
1
P
,q()
= 0
5.2 Modular forms with zeros and poles on Heegner divisors II 139
in

1
1,1
(A
L
). Using the assumption on we nd that
c()(q())
1
P
,q()
is contained in S
,L
. On the other hand, by Proposition 1.5 this linear com-
bination of Poincare series is orthogonal to S
,L
. Therefore it vanishes iden-
tically. But this is equivalent to saying
c()a
,q()
= 0
in S
,L
. By Theorem 1.17 there exists a nearly holomorphic modular form
f M
!
k,L
with principal part (5.24). Let B(f) be the Borcherds lift of f
as in Theorem 3.22. Then B(f)/F is a holomorphic modular form (with
a multiplier system of nite order) without any zeros on H
l
. Hence it is
constant. .
Theorem 5.12. Let L be an even lattice of signature (2, l), that splits two
orthogonal hyperbolic planes over Z. Then every meromorphic modular form
for the group (L), whose divisor is a linear combination of Heegner divisors,
is a Borcherds product as in Theorem 3.22.
Proof. We write L as an orthogonal sum L = D H
1
H
2
, where D is a
negative denite lattice and H
1
, H
2
are hyperbolic planes. Put K = DH
1
.
According to Theorem 5.11 it suces to prove that the map : S
+
,L

1
1,1
(A
L
) is injective.
Let f S
+
,L
with (f) = 0. Denote the Fourier coecients of f by c(, n).
We have to show that f S
,L
.
For a suitable choice of the vectors z and z
the Fourier expansion of (f)

is given by
(f)(Z) =
0
(f)(Y ) 2
1/2
q()<0
n|
[[
22
n
23
c(/n, q()/n
2
)
([[[Y [, (, Y )) e((, X)).

The assumption (f) = 0 implies that
[[
22
n|
n
23
c(/n, q()/n
2
) = 0
for all K
with q() < 0. By an easy inductive argument we nd that

c(, q()) = 0 for all K
with q() < 0.

Recall that there is an isomorphism between H
1
H
2
and the lattice
M
2
(Z) of integral 2 2 matrices such that the quadratic form q corresponds
to the determinant on M
2
(Z). The group SL
2
(Z) acts on M
2
(Z) by multipli-
cation from both sides. This gives rise to a homomorphism
SL
2
(Z) SL
2
(Z) O
+
(H
1
H
2
)
(cf. [FH] Lemma 4.4). Hence, by the theorem of elementary divisors for
SL
2
(Z), for any L
there is a O
+
(H
1
H
2
) (L) such that
() K
.
Thereby we obtain c(, q()) = 0 for all L
with q() < 0. .

It seems likely that Theorem 5.11 can also be used to prove a more general
version of Theorem 5.12. However, this seems to be a non-trivial problem,
which might require additional arguments like some newform theory or Hecke
theory for the space S
,L
and the lifting .
Example 5.13. Fix a positive integer t. Let L be the lattice L = DH
1
H
2
,
where H
1
, H
2
are hyperbolic planes and D denotes the lattice Z with the
2
. Obviously L has signature
(2, 3), k = 1/2, and = 5/2. The discriminant group L
/L

= D
/D has
order 2t. In the same way as in [EZ] '5 it can be seen that the space S
5/2,L
is isomorphic to the space .
cusp
3,t
of skew-holomorphic Jacobi cusp forms of
weight 3 and index t (see [Sk], [Ko]).
The group (L) is isomorphic to the paramodular group of level t (cf. [GN]
Lemma 1.9 and [GrHu]). Moreover, the quotient A
L
= H
l
/(L) is the moduli
space of Abelian surfaces with a (1, t)-polarization. The Heegner divisors
H(, m) are known as Humbert surfaces. Since L splits two hyperbolic planes
over Z, any H(, m) is a prime divisor in D(A
L
) (cf. [FH] Lemma 4.4).
By Theorem 5.9 we have an injective map
: .
cusp
3,t
1
1,1
(A
L
)
from .
cusp
3,t
into the space of square integrable harmonic (1, 1)-forms on A
L
.
Observe that the 0-th Fourier coecient
0
(f)(Y ) of the lift of f .
cusp
3,t
can be identied with the Shimura lift of f. According to Theorem 5.12 any
meromorphic modular form with respect to (L), whose only zeros and poles
lie on Humbert surfaces, is a Borcherds product in the sense of Theorem 3.22.
In this case the classical theta lift is essentially the Maass lift. It gives rise
to an injective map
: J
cusp
3,t
1
3,0
(A
L
)
from holomorphic Jacobi cusp forms of weight 3 and index t into 1
3,0
(A
L
).
We obtain a partial answer to a question raised by W. Kohnen ([Ko] Problem
(ii) in '2). Holomorphic Jacobi forms of weight 3 contribute via Maass lift
to the middle cohomology of A
L
, whereas skew-holomorphic Jacobi forms
contribute to the second cohomology. Here it is natural to ask, if there is a
similar connection in higher weights.
References
[AbSt] M. Abramowitz and I. Stegun, Pocketbook of Mathematical Functions,
Verlag Harri Deutsch (1984).
[Bo1] R. E. Borcherds, Automorphic forms on O
s+2,2
(R) and innite products,
Invent. Math. 120 (1995), 161213.
[Bo2] R. E. Borcherds, Automorphic forms with singularities on Grassmannians,
Invent. Math. 132 (1998), 491562.
[Bo3] R. E. Borcherds, The Gross-Kohnen-Zagier theorem in higher dimensions,
Duke Math. J. 97 (1999), 219233.
[Bo4] R. E. Borcherds, Sporadic groups and string theory, First E. C. M., Vol. I
(Paris 1992), 411-421, Progr. Math. 119 (1994), Birkhauser, Basel.
[Bl1] A. Borel, Introduction aux groupes arithmetiques, Hermann (1969).
[Bl2] A. Borel, Linear Algebraic Groups. In: Algebraic groups and discontinuous
subgroups, Proceedings of Symposia in Pure Mathematics 9, American
Mathematical Society (1966), 319.
[Bl3] A. Borel, Cohomology of arithmetic groups, Proc. Int. Congr. of Mathe-
matics, Vancouver 1974, Vol. 1 (1975), 435442.
[Br1] J. H. Bruinier, Borcherds products and Chern classes of Hirzebruch-Zagier
divisors, Invent. math. 138 (1999), 5183.
[Br2] J. H. Bruinier, Borcherdsprodukte und Chernsche Klassen von Hirze-
bruch-Zagier-Zykeln, Dissertation, Universitat Heidelberg (1998).
[BK] J. H. Bruinier and M. Kuss, Eisenstein series attached to lattices and mod-
ular forms on orthogonal groups, preprint (2000), to appear in Manuscr.
Math.
[BKK] J. I. Burgos, J. Kramer, and U. K uhn , manuscript, in preparation (2001).
[DI] F. Diamond and J. Im, Modular forms and modular curves, Canadian
Mathematical Society Conference Proceedings 17 (1995), 39133.
[DN] K. Doi and H. Naganuma, On the functional equation of certain Dirichlet
series, Invent. Math. 9 (1969), 114.
[E1] A. Erdelyi, W. Magnus, F. Oberhettinger and F. G. Tricomi, Higher Tran-
scendental Functions, McGraw-Hill (1953).
[E2] A. Erdelyi, W. Magnus, F. Oberhettinger and F. G. Tricomi, Tables of
Integral Transforms, vol. I, McGraw-Hill (1954).
[EZ] M. Eichler and D. Zagier, The Theory of Jacobi Forms, Progress in Math.
55 (1985), Birkhauser.
[Fa] J. Fay, Fourier coecients of the resolvent for a Fuchsian group, J. reine
angew. Math. 294 (1977), 143203.
[Fr1] E. Freitag, Siegelsche Modulfunktionen, Springer-Verlag (1983).
[Fr2] E. Freitag, Singular Modular Forms and Theta Relations, Lecture Notes
in Mathematics 1487, Springer-Verlag (1991).
142 References
[FH] E. Freitag and C. F. Hermann, Some modular varieties of low dimension,
Adv. Math. 152 (2000), 203287.
[Fu] J. Funke, Rational quadratic divisors and automorphic forms, PhD-thesis,
University of Maryland (1999).
[GH] P. Griths and J. Harris, Principles of algebraic geometry, John Wiley
(1978).
[GrHu] V. Gritsenko and K. Hulek, Minimal Siegel modular threefolds, Math.
Proc. Camb. Phil. Soc. 123 (1998), 461485.
[GN] V. Gritsenko and V. Nikulin, Automorphic forms and Lorentzian Kac-
Moody algebras. Part II, Intern. J. of Math. 9 (1998), 201275.
[GR] R. C. Gunning and H. Rossi, Analytic functions of several complex vari-
ables, Prentice-Hall (1965).
[GW] R. E. Greene and H. Wu, C
approximations of convex, subharmonic,

and plurisubharmonic functions, Ann. scient.

Ec. Norm. Sup. 12 (1979),
4784.
[Ha] Harish-Chandra, Automorphic Forms on Semisimple Lie Groups, Lecture
Notes in Mathematics 62, Springer-Verlag (1968).
[He] D. A. Hejhal, The Selberg Trace Formula for PSL(2, R), Lecture Notes in
Mathematics 1001, Springer-Verlag (1983).
[Hel] S. Helgason, Dierential geometry and symmetric spaces, Academic Press
(1962).
[HM] J. Harvey and G. Moore, Algebras, BPS states, and strings, Nuclear Phys.
B 463 (1996), no. 2-3, 315368.
[HZ] F. Hirzebruch und D. Zagier, Intersection Numbers of Curves on Hilbert
Modular Surfaces and Modular Forms of Nebentypus, Invent. Math. 36
(1976), 57113.
[Ho] R. Howe, -series and invariant theory, Proceedings of Symposia in Pure
Mathematics 33, part 1, American Mathematical Society (1979), 275285.
[Ka] N. Katz, p-adic properties of modular schemes and modular forms. In:
Modular Functions of One Variable III, Lecture Notes in Math. 350,
Springer-Verlag (1973), 69190.
[Ko] W. Kohnen, Jacobi forms and Siegel modular forms: Recent results and
problems, LEns. Math. 39 (1993), 121136.
[Kon] M. Kontsevich, Product formulas for modular forms on O(2, n), Seminaire
Bourbaki 821 (1996).
[KM1] S. Kudla and J. Millson, The theta correspondence and harmonic forms
I, Math. Ann. 274, (1986), 353378.
[KM2] S. Kudla and J. Millson, The theta correspondence and harmonic forms
II, Math. Ann. 277, (1987), 267314.
[KM3] S. Kudla and J. Millson, Intersection numbers of cycles on locally sym-
metric spaces and Fourier coecients of holomorphic modular forms in
several complex variables, IHES Publi. Math. 71 (1990), 121172.
[Ma1] H. Maass, Modular functions of one complex variable, Spinger-Verlag
(1983).
[Ma2] H. Maass, Automorphe Funktionen in mehreren Veranderlichen und
Dirichletsche Reihen, Abh. Math. Sem. Univ. Hamburg 16 (1949), 72
100.
[Ma3] H. Maass,

Uber ein Analogon zur Vermutung von Saito-Kurokawa, Invent.
Math. 60 (1980), 85104.
References 143
[Mar] G. A. Margulis, Discrete Subgroups of Semisimple Lie Groups, Springer-
Verlag (1991).
[Me] O. Meyer,

Uber die Satakekompaktizierung arithmetischer Quotienten
von beschrankten hermiteschen Bereichen des Types IV
n
, Diplomarbeit,
Universitat Heidelberg (1998).
[Na] H. Naganuma, On the coincidence of two Dirichlet series associated with
cusp forms of Heckes Neben-type and Hilbert modular forms over a real
quadratic eld, J. Math. Soc. Japan 25 (1973), 547555.
[Nak] S. Nakajima, On Invariant Dierential Operators on Bounded Symmetric
Domains of Type IV, Proc. Japan Acad. 58 Ser. A (1982), 235238.
[Ni] D. Niebur, A class of nonanalytic automorphic functions, Nagoya Math.
J. 52 (1973), 133145.
[No] A. Nobs, Die irreduziblen Darstellungen der Gruppen SL
2
(Z
p
), insbeson-
dere SL
2
(Z
2
). I. Teil, Comment Math. Helvetici 51 (1976), 465489.
[Od] T. Oda, On Modular Forms Associated with Indenite Quadratic Forms
of Signature (2, n 2), Math. Ann. 231 (1977), 97144.
[PS1] I. Piateskii-Shapiro, On the Saito-Kurokawa lifting, Invent. Math. 71
(1983), 309338.
[PS2] I. Piateskii-Shapiro, Cuspidal automorphic representations associated to
parabolic subgroups and Ramanujan conjecture. In: Number theory re-
lated to Fermats last theorem, Prog. Math. 26, Birkhauser (1982), 143
151.
[RS] S. Rallis and G. Schimann, On a relation between SL
2
cusp forms and
cusp forms on tube domains associated to orthogonal groups, Trans. AMS
263 (1981), 158.
[Rh] G. de Rham, Dierentiable Manifolds, Springer-Verlag (1984).
[Ro] W. Roelcke,

Uber den Laplace-Operator auf Riemannschen Manigfaltig-
keiten mit diskontinuierlichen Gruppen, Math. Nachr. 21 (1960), 131149.
[Sc] B. Schoeneberg, Elliptic Modular Functions, Springer-Verlag (1974).
[SeSt] J.-P. Serre and H. M. Stark, Modular forms of weight 1/2. In: Modular
Functions of One Variable VI, Lecture Notes in Math. 627, Springer-
Verlag (1977), 2767.
[Sh] G. Shimura, Introduction to the Arithmetic Theory of Automorphic Func-
tions, Princeton University Press (1971).
[Sk] N.-P. Skoruppa, Developments in the theory of Jacobi forms. In: Proceed-
ings of the conference on automorphic funtions and their applications,
Chabarovsk (eds.: N. Kuznetsov and V. Bykovsky), The USSR Academy
of Science (1990), 167185. (see also MPI-preprint 89-40, Bonn (1989).)
[Sn] T. Shintani, On construction of holomorphic cusp forms of half integral
weight, Nagoya Math. J. 58 (1975), 83126.
[SABK] C. Soule, D. Abramovich, J.-F. Burnol, and J. Kramer, Lectures on
Arakelov Geometry, Cambridge Studies in Advanced Mathematics 33,
Cambridge University Press (1992).
[Wei] R. Weissauer, The Picard group of Siegel modular threefolds, J. reine
angew. Math. 430 (1992), 179211.
[Wel] R. O. Wells, Dierential analysis on complex manifolds, Springer-Verlag
(1980).
[Yau] S.-T. Yau, Some function-theoretic properties of complete Riemannian
Manifolds and their applications to geometry, Indiana Univ. Math. J. 25
(1976), 659670.
144 References
[Za] D. Zagier, Modular Forms Associated to Real Quadratic Fields, Invent.
Math. 30 (1975), 146.
Notation
The principal branch of the square root

f() <g() [f()[ C[g()[
f() <
g() [f()[ C()[g()[

f g See p. 52
[
The Petersson slash operator of weight with representation

L
(p. 17)
[
The Petersson slash operator of weight with representation
L
(p. 17)
'a, b` The standard scalar product on C[L
/L] for a, b C[L
/L]
(a, b) The bilinear form on L for a, b L
(f, g) The Petersson scalar product of f, g S
,L
(p. 22)
L
K
See p. 131
L
K
See p. 132
2
= (, ) for V
[[ = [(, )[
1/2
for V
The orthogonal complement of
v
The orthogonal projection of V onto v
a
,n
A certain functional in S
,L
(p. 36) or o
,L
(Z) (p. 122)
/
,L
(Z) See p. 122
b(, n, s) The Fourier coecients of F
,m
(, s) (p. 30)
b
,m
(, n) = b(, n) = b(, n, 1 k/2)
B
r
(x) The r-th Bernoulli polynomial (p. 63)
B
r
(x) See p. 64
c(D) The Chern class of the divisor D in the second cohomology
c(, n; y, s) The Fourier coecients of F
,m
(, s) (p. 57)
C The positive cone in K R (p. 79)
C
,m
A certain constant (p. 75)
Cl(X/) The divisor class group of X/ (p. 119)
Cl(X/) The modied divisor class group of X/ (p. 119)
Cl
H
(A
L
) The subgroup of

Cl(A
L
) generated by the Heegner divisors H(, m)
C The complex numbers
C[L
/L] The group algebra of the discriminant group L
/L
(
s=a
[f(s)] The constant term of the Laurent expansion of f at s = a (p. 51)
A multiplier system or character
d Often a primitive isotropic vector in K
146 Notation
d
An element of K
with (d, d
) = 1
d Usually d
q(d
)d
D = K d
, or a divisor
,
The Kronecker-delta
k
The Laplace operator of weight k (p. 29)
e(z) = e
2iz
for z C
e
An element of the standard basis (e
)
L
/L
of C[L
/L]
e
() = e
e()
E
L
() An Eisenstein series in M
,L
(p. 23)
E() = E
0
() = E
L
0
()
E(u, v) An Eichler transformation (p. 107)
See p. 122
F
L
,m
(, s) A non-holomorphic Poincare series of weight k (p. 29)
F
,m
(, s) See p. 30
F(a, b, c; z) The Gauss hypergeometric function as in [AbSt]
F
K
,m
(, s; r, t) See p. 46
T The standard fundamental domain for SL
2
(Z)
T
u
The truncated fundamental domain (p. 47)
Gal(E/F) The Galois group of a eld extension E/F
Gr(L) The Grassmannian of L
Usually a subgroup of nite index of (L)
(N) The principal congruence subgroup of level N
(y) The Gamma function
(a, x) The incomplete Gamma function as in [AbSt] (p. 28)
1
= SL
2
(Z)
= (
1 n
0 1
) ; n Z SL
2
(Z)
= ((
1 n
0 1
) , 1); n Z Mp
2
(Z)
(L) = O
+
(V ) O
d
(L)
H(, m) The Heegner divisor of index (, m) (p. 48 and p. 119)
h
,m
(Z) The representative
1
4
L
,m
(Z) of the Chern class of H(, m)
H
c
(, m, , n) A generalized Kloosterman sum (p. 19)
H
c
(, m, , n) A generalized Kloosterman sum (p. 30)
H
2
(A
L
, C) The second cohomology of A
L
with coecients in C
H
2
(A
L
, C) See p. 125
1
1,1
(A
L
) The space of square integrable harmonic (1, 1)-forms on A
L
(p. 125)
1
1,1
(A
L
) See p. 126
H The complex upper half plane C; () > 0
H
l
The generalized upper half plane (p. 79)
H
l
() The generalized upper half plane attached to (p. 110)
1 The upper half space model of hyperbolic space (p. 67)
I
(y) The I-Bessel function as in [AbSt] or [E1]

() The imaginary part
J
(y) The J-Bessel function as in [AbSt] or [E1]

Notation 147
j(, Z) The automorphic factor ((Z
L
), z) (p. 85)
J
k,t
The space of holomorphic Jacobi forms of weight k and index t
.
k,t
The space of skew-holomorphic Jacobi forms of weight k and index t
K
(y) The K-Bessel function as in [AbSt] or [E1]

k Usually 1 l/2
Usually 1 +l/2
K = L z
/ See p. 78
/
+
A component of /
/
+
The cone over /
+
(p. 84)
l An integer 3
L An even lattice of signature (b
+
, b
)
L
The dual lattice of L

L
0
= L
; (, z) 0 (mod N)
L
k
A Maass dierential operator (p. 97)
L
p
(H
l
/) See p. 103
Usually an admissible index-tuple (p. 110)
M = (M,
c +d) for M =

a b
c d
SL
2
(R)
M
,
(y) The M-Whittaker function as in [AbSt] (p. 27)
s
(y) = y
k/2
M
k/2, s1/2
(y)
M
,L
The space of modular forms of weight with respect to
L
and Mp
2
(Z)
M
!
k,L
The space of nearly holomorphic modular forms of weight k (p. 34)
Mp
2
(R) The metaplectic cover of SL
2
(R) (p. 15)
Mp
2
(Z) The integral metaplectic group (p. 16)
,L
= Gal(C/Q) M
,L
(p. 122)
Usually a certain vector in V z
(p. 42)
N The unique positive integer with (z, L) = NZ, or the level of L
A The zero-quadric in P(V (C)) (p. 78)
N = 1, 2, 3, . . .
N
0
= N 0
O(V ) The (special) orthogonal group of V (p. 40)
O
+
(V ) The connected component of the identity of O(V )
O(L) = g O(V ); gL = L
O
d
(L) The discriminant kernel of O(L), see p. 40
O
+
Q
(L) The rational orthogonal group of L (p. 103)
The invariant Laplace operator on H
l
(p. 96)
() A kernel function, see p. 125
L
(Z, ) The kernel function of , see p. 130
K
(Y, ) A kernel function, see p. 132
p Usually a certain projection L
0
K
(p. 41)
P
L
,m
() The Poincare series of index (, m) in S
,L
(p. 19)
p
,m
(, n) The Fourier coecients of P
L
,m
P(V (C)) The projective space of V (C)
L
,m
(v, s) The regularized theta lift of F
,m
(, s) (p. 47)
148 Notation
,m
(v) = (
s=1k/2
[
,m
(v, s)], the regularized theta lift of F
,m
(, 1 k/2)
L
,m
See p. 68 and p. 81
,m
(Z) See p. 82
q(a) =
1
2
(a, a), the quadratic form on L
q
(, n) The Fourier coecients of E

L
()
q(, n) The Fourier coecients of E()
Q The rational numbers
R
k
A Maass dierential operator (p. 97)
R The real numbers
{
t
The set of Y dened by (4.18b)(4.18d)
1() The real part
L
The Weil representation attached to (L
/L, q) (p. 16)
(M, ) A coecient of the representation

L
(p. 17)
L
The dual representation of
L
,m
(W) See p. 69
f
(W) See p. 88
s A complex variable
S =

0 1
1 0
Mp
2
(Z)
SL
2
(R) The group of real 2 2-matrices of determinant 1
SL
2
(Z) The group of integral 2 2-matrices of determinant 1
S
,L
The subspace of cusp forms in M
,L
S
,L
The dual space of S
,L
S
,L
A certain subspace of S
,L
(p. 138)
S
+
,L
A certain subspace of S
,L
(p. 138)
supp(D) The support of the divisor D
o
t
A Siegel domain in H
l
(p. 103)
o
t
() A Siegel domain in /
+
(p. 110)
o(, m, U) See p. 48 and p. 83
o
,L
= Gal(C/Q) S
,L
(p. 121)
o
,L
(Z) The module of cusp forms in o
,L
whose coecients all lie in Z
o
,L
(Z) The dual module of o
,L
(Z)
Often the real part of s
T = ((
1 1
0 1
) , 1) Mp
2
(Z)
A variable in H
L
(, v; r, t) The Siegel theta function attached to L (p. 39)
(, v; r, t) A component of
L
(, v; r, t)
A lifting, see p. 130
0
(f)(Y ) The 0-th Fourier coecient of (f) (p. 135)
D
(Y
D
, ) A theta series, see p. 133
v Usually an element of Gr(L)
V L R
V (C) = V C
1
(A, B) A special function, see p. 74

w The orthogonal complement of z
v
in v
Notation 149
W A Weyl chamber (p. 63), or a complex variable
W
,
(y) The W-Whittaker function as in [AbSt] (p. 27)
)
s
(y) = [y[
k/2
W
k/2 sgn(y), s1/2
([y[)
x The real part of
X The real part of Z H
l
, or a normal irreducible complex space
A
L
= H
l
/(L)
L
,m
See p. 68 and p. 81
y The imaginary part of
Y Usually the imaginary part of Z H
l
z Usually a primitive isotropic vector in L
z
An element of L
with (z, z
) = 1
z Usually z
q(z
)z
Z =

1 0
0 1
, i
Mp
2
(Z), or a variable in H
l
Z
L
Usually (Z, 1, q(Z) q(z
))

/
+
for Z H
l
Z The integers
A vector in L with (, z) = N
(s) The Riemann zeta function
N
A primitive N-th root of unity
Index
admissible index-tuple, 110
arrow-operators, 131
automorphic form, 85
automorphy factor, 85, 123
Bernoulli
number, 26
polynomial, 63
Bessel function, 19, 30, 64
Borcherds product, 87, 88, 139
generalized, 63, 82
Cartier divisor, 123
Casimir element, 101
character, 85
Chern class
of a divisor, 123
of a Heegner divisor, 124, 126
cocycle relation, 85
complex structure, 78
cusp form, 18
discriminant group, 16
discriminant kernel, 40
divisor, 119
class group, 119
group, 119
Doi-Naganuma lifting, 131
dual lattice, 16
dual representation, 17
Eichler transformation, 86, 107, 111
Eisenstein series, 23
error function, 54
Euler-Mascheroni constant, 75
Fourier coecients, 18, 30
Fourier expansion, 18, 29, 86
of the theta lift, 63
fundamental domain, 22
truncated, 47
Gauss hypergeometric function, 54
Gram determinant, 41
Grassmannian, 39
Green current, 124
Green object, 124
Hankel integral, 25
Heegner divisor, 80, 119
Hermitean metric, 123, 126
Hermitean symmetric space, 73, 95
Humbert surface, 93
Humbert surfaces, 140
hyperbolic plane, 91, 138
hyperbolic space, 63
hyperboloid model, 66, 79
upper half space model, 67
incomplete Gamma function, 28
Jacobi form, 18
skew-holomorphic, 18, 140
weak, 92
Kahler class, 124
Kahler metric, 96
kernel function, 125, 130
Kloosterman sum, 19, 30
Koecher principle, 85, 86, 109
Laplace operator
for O(2, l), 96
hyperbolic, 72
of weight k, 28
on a Riemann manifold, 112
lattice, 16
152 Index
Lorentzian, 63
Laurant expansion, 47
Lehmer conjecture, 94
level, 120
Maass operators, 97
metaplectic group, 15
modular form, 18
elliptic , 25
nearly holomorphic, 34
on O(2, l), 85
multiplier system, 85
orthogonal group, 40
paramodular group, 140
Petersson coecient formula, 22
Petersson slash operator, 17
pluriharmonic, 82
Poincare series, 19
non-holomorphic, 29
principal part, 34
projection, 41, 42
quadratic form, 16
Ramanujan sum, 25
Ramanujan-Petersson conjecture, 20
reduction theory, 103, 104
Riemann manifold, 112
Riemann metric, 95
invariant, 96
scalar product, 17
on (1, 1)-forms, 126
Petersson, 22
Serre duality, 36, 125
Shimura lift, 140
Siegel domain, 103, 110
Siegel theta function, 39
singularity, 52
skew-holomorphic Jacobi form, 18, 140
subharmonic, 113
theta function, 39
theta integral, 46
theta lift, 39, 135
regularized, 47
theta transformation formula, 40
tube domain, 79
unfolding trick, 23, 54, 55, 57
upper half plane, 15
generalized, 79
upper half space model, 67
weak Jacobi form, 92
Weil representation, 16, 100
Weyl chamber, 63, 80, 88
Weyl vector, 69, 88
Whittaker dierential equation, 27
Whittaker function, 27
asymptotic properties, 50
generalized, 73
zero-quadric, 78

Bruinier O2l Forms

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Bruinier O2l Forms

Hochgeladen von

Copyright:

Verfügbare Formate

To my parents

z) (/2, /2]. For any integer k we put

is the standard generator of the center of Mp

= x L Q; (x, y) Z for all y L

/L is a nite Abelian group, the so-

/L whose associated bilinear form is the modulo 1 re-

/L]. If we denote the standard basis of C[L

/L, q) (cf. [No]). It factors through the nite group

is even, and through a double cover of SL

a( +r, +r) 2(, +r) +d(, )

/L]-valued function on H. For (M, )

/L] is a holomorphic function which is

. The invariance of f under T implies

() are periodic with period 1. Thus f has a

()e(n) dx. Using the abbreviation

e() we may write

/L] the coecients can be expressed

/L] is called a modular

(M, ) = f for all (M, ) Mp

/L and m Z q() with m > 0. Then e

/L] which is invariant under the [

(m) is also invariant under the action of Z

= 'T`). Since 5/2, the usual argument shows that

/L]. We will often omit the superscript L, if

with lower row (c d

d (mod c). Observe that the expression

are universally bounded, because

/L, n Z q(), and c Z 0.

/L and n Z q(). According to (1.10)

(m) is invariant under the action of Z

is bounded on the standard

is bounded on the whole upper half plane H.) We dene the

/L and m Z q() with m > 0. Let f be a

/L with q() Z. Then the vector e

/L], is invariant under the [

has the Fourier expansion

/L and n Z q(). By (1.10) we have

is invariant under the action of Z

, and the fact that 2 (mod 4), we get

equals the space

where runs through a set of representatives of

/L; q() Z modulo the action of 1. Comparing constant terms one

with as above is already linearly

/L; q() Z, 2 = 0 +L. (1.20)

1.3 Non-holomorphic Poincare series of negative weight 27

) = (2, l), (1, l 1), or (0, l 2) with l 3. Put k = 1 l/2 and

/L] component-wise. Let L

/L] and any (M, ) Mp

/L] be a continuous function with the property f [

(nx)` dx. (1.37)

(z) the usual Bessel functions as dened in [AbSt] Chap. 9.

with lower row (c d

d (mod c). Observe that the expression

with c = 0. The latter sum will be denoted by H(, s). Since

(mx), we nd that the rst term equals

/L and n Z+q() and denote the (, n)-th Fourier coecient of

+d/c and substitute x by x d/c in the

/L] is called a nearly holomorphic

/L, n Z+q()). Then

/L, m Z+q() and n Z+q().

/L, m Z+q() and n Z+q().

/L, n Z+q(), m Z+q(), and m < 0.

(, m) is the (, m)-th coecient of the Eisenstein series E