Applying this, we have
Wν(s) = Z ∞
0
Γ(ν)e−2x
ν−1
∑
l=0
(2x)l l!
! exxsdx
x
=Γ(ν)
ν−1
∑
l=0
2l l!
Z ∞
0
e−xxs+ldx x
=Γ(ν)
ν−1
∑
l=0
2l
l!Γ(s+l).
By Stirling’s approximation for gamma function, for anyµ>0we have Γ(s)∼√
2πtσ−1/2e−π|t|/2as|t| →∞, (s=σ+it).
Wν(s)∼Γ(ν)
ν−1
∑
l=0
2l l!(√
2πtσ+l−1/2e−π|t|/2) =O(tσ+ν−3/2e−π|t|/2) =O(t−µ)as|t| →∞.
for fixedσ>0.
(3) By definition, we observe that Wν(s)is the Mellin transform of Γ(ν,2x)ex. So we only need to check the absolutely convergence of Γ(ν,2x)ex for x>0. Clearly Γ(ν,2x)ex is a continuous function. Since|Γ(ν,2x)ex| ≤Γ(ν)exfor x≥0, we haveΓ(ν,2x)ex=O(1)as x→0. In the proof of (1), we haveΓ(ν,2x)∼(2x)ν−1e−2xas|x| →∞, we get the required growth condition as x→∞.
Therefore the inversion integral ofΓ(ν,2x)exalong the vertical lineRe(s) =c>0exists.
Proof 28 As in the proof of Lemma 2, using the Euler-Gauss formula for gamma function, for n≥1we have
c+f(n)≤C1(−1)n
−v−1 n
,
for some constant C1>0. Next, since1−k≥1we can use(56)as follows. For n≥1, we have Γ(1−k,4πny) =Γ(1−k)e−4πny
−k
∑
l=0
(4πny)l
l! =
O(e−4πny), as y→0, O(y−ke−4πny), as y→∞.
Again by the Euler-Gauss formula, we have
|c−f(−n)Γ(1−k,4πny)| ≤C2(−1)ne−4πny
−k
∑
l=0
(4πny)l l!
−v−l−1 n
for some constant C2>0. Substituting these results to the Fourier expansion of f , we have
|f(z)| ≤C1
∞
∑
n=0
(−1)n
−v−1 n
e−2πny+c−f(0)y1−k+C2
−k
∑
l=0
(4πny)l l!
∞
∑
n=1
(−1)n
−v−l−1 n
e−2πny
≤C1(1−e−2πy)−v−1+c−f (0)y1−k+C2
−k
∑
l=0
(4πny)l
l! ((1−e−2πy)−v−l−1−1). (57) This shows that the given series f(z)is convergent absolutely and uniformly on any compact subset of H.
Since c−f(−n) =O(nv), we obtain that the formal Fourier series (1−k)c−f(0)−(4π)1−k
∑
n≥1
c−f(−n)n1−kqn
is uniformly convergent on any compact subset ofH and defines a holomorphic function. Moreover, by using Proposition 8 and(53), we have
∆k(f(z)) =−ξ2−k(ξk(f(z))) =0.
Now Theorem 11 tells us that f(z)is a real analytic function onH.
Since1−k≥1and(1−e−2πy)−v−1=O(y−v−1)as y→0, by using(57)we have f(z) =O(y−v+k−1)as y→0.
Next, we have
f(z)−c+f(0)−c−f(0)y1−k=e2πizg(z), where
g(z) =
∞
∑
n=0
c+f(n+1)qn+
∑
n≤−2
c−f(n+1)Γ(1−k,−4π(n+1)y)qn.
By(50), for any n≤ −2we have
|Γ(1−k,−4π(n+1)y)qn| ≤((−4π(n+1)y)−ke4π(n+1)y)e−2πny
= (−4π(n+1)y)−ke2π(n+2)y as y→∞. Then we have
|g(z)| ≤C1
∞ n=0
∑
(−1)n+1
−v−1 n+1
e−2πny+
∑
n≤−2
(−(n+1))v(−4π(n+1)y)−ke2π(n+2)y
≤C1((1−e−2πy)−v−1−1) +
∑
n≤−2
(−(n+1))v(−4π(n+1)y)−ke2π(n+2)y. This shows that g(z)is bounded in a neighborhood of i∞. Therefore we have
f(z)−c+f(n)−c−f (n)y1−k=O(e−2πy)as y→∞.
The following lemma and the two specific functions serve to combine the two L-functionsΛN and ΩNwell in the proof of the converse theorem.
Lemma 16 Let k∈Z. Let f and g be two real analytic function onH such that g=f|kωN. Define H(z) =2iy∂f
∂x(z) +k f(z), I(z) =2iy∂g
∂x(z) +kg(z).
Then for any t>0, we have
H|kωN(it) =N−k/2(it)−kH
− 1 Nit
=−I(it).
Proof 29 Since g(z) = f|kωN(z) =Nk/2(Nz)−kf(−1/Nz), we have
∂g
∂x(z) =N−k/2∂f
∂x
z−kf
− 1 Nz
=N−k/2
−kz−k−1f
− 1 Nz
+z−k∂f
∂x
− 1 Nz
1 Nz2
=−kN−k/2z−k−1f
− 1 Nz
+N−k/2z−k∂f
∂x
− 1 Nz
1 Nz2. Multiplying both side by2iy and evaluating at z=it (x=0,y=t) for t>0. Then we get
2iy∂g
∂x(z) =−2kN−k/2iyz−k−1f
− 1 Nz
+2N−k/2iyz−k∂f
∂x
− 1 Nz
1 Nz2
⇒2iy∂g
∂x(z) +kN−k/2iyz−k−1f
− 1 Nz
=−kN−k/2iyz−k−1f
− 1 Nz
+2N−k/2iyz−k∂f
∂x
− 1 Nz
1 Nz2
⇒2it∂g
∂x(it) +kN−k/2(it)−kf
− 1 Nit
=−kN−k/2(it)−kf
− 1 Nit
−2iN−k/2(it)−k∂f
∂x
− 1 Nit
1 Nt.
Now the left-hand side is 2it∂g
∂x(it) +k
N−k/2(it)−kf
− 1 Nit
=2it∂g
∂x(it) +kg(it) =I(it), and the right-hand side is
−N−k/2(it)−k
k f
− 1 Nit
+2i 1
Nt
∂f
∂x
− 1 Nit
=−N−k/2(it)−k=−H|kωN(it).
This proves the lemma.
Using the above lemmas, we will prove the following Hecke-type converse theorem for the harmonic Maass forms of polynomial growth.
Theorem 15 (Theorem 4.5 in [34]) Let k be a negative integer and N be a positive integer. Let f and g be two functions defined onH by the formal Fourier series
f(z) =
∞ n=0
∑
c+f (n)qn+c−f(0)y1−k+
∑
n<0
c−f(n)Γ(1−k,−4πny)qn, g(z) =
∞
∑
n=0
c+g(n)qn+c−g(0)y1−k+
∑
n<0
c−g(n)Γ(1−k,−4πny)qn, with c±f (n),c±g(n) =O(|n|v), n∈Zfor some v≥0. Then the followings are equivalent.
(1) g= f|kωN
(2) The L-functionsΛN(f,s),ΛN(g,s),ΩN(f,s), andΩN(g,s)satisfy the following conditions (a) ΛN(f,s), ΛN(g,s), ΩN(f,s), and ΩN(g,s) admits meromorphic continuation to the whole
complex plane,
(b) the following functions are entire and bounded in vertical strips Λ∗N(f,s) =ΛN(f,s) +c+f(0)
s +c+g(0)ik
k−s + c−f (0) N(1−k)/2
1
s−k+1+ c−g(0)ik N(1−k)/2
1 1−s, Λ∗N(g,s) =ΛN(g,s) +c+g(0)
s +c+f(0)i−k
k−s + c−g(0) N(1−k)/2
1
s−k+1+c−f(0)i−k N(1−k)/2
1 1−s, Ω∗N(f,s) =ΩN(f,s) +k c+f(0)
s −c+g(0)ik
k−s + c−f (0) N(1−k)/2
1
s−k+1− c−g(0)ik N(1−k)/2
1 1−s
! ,
Ω∗N(g,s) =ΩN(g,s) +k c+g(0)
s −c+f(0)i−k
k−s + c−g(0) N(1−k)/2
1
s−k+1−c−f(0)i−k N(1−k)/2
1 1−s
! .
(c) satisfy the functional equations
ΛN(f,s) =ikΛN(g,k−s), ΩN(f,s) =−ikΩN(g,k−s).
Proof 30
(1)⇒(2) For n≥1, we have c−f(−n)
Z ∞
0
Γ(1−k,4πnt/√
N)e−2πnt/
√ Ntsdt
t =c−f(−n) 2πn
√N
−sZ ∞
0
Γ(1−k,2t)e−ttsdt t
=c−f(−n) 2πn
√ N
−s
W1−k(s).
Then forRe(s)>v+1, we obtain Z ∞
0
f it
√N
−c+f(0)−c−f (0) t
√N 1−k!
tsdt t
= Z ∞
0
∞
∑
n=1
c+f (n)e−2πnt/
√ N
! tsdt
t + Z ∞
0
∞
∑
n=1
c−f (−n)Γ(1−k,4πnt/√
N)e−2πnt/
√ N
! tsdt
t
= 2π
√ N
−s
(Γ(s)L+(f,s) +W1−k(s)L−(f,s))
=ΛN(f,s). (58)
Now forRe(s)>v+1, we have ΛN(f,s) =
Z ∞
0
f it
√ N
−c+f(0)−c−f(0) t
√ N
1−k! tsdt
t
= Z ∞
1
+ Z 1
0
f
it
√ N
−c+f(0)−c−f (0) t
√ N
1−k! tsdt
t
= Z ∞
1
f it
√ N
−c+f(0)−c−f(0) t
√ N
1−k! tsdt
t +
( Z 1
0
f it
√ N
tsdt
t −c+f(0)
s − c−f(0)
√N1−k 1 s−k+1
)
= Z ∞
1
f it
√ N
−c+f(0)−c−f(0) t
√ N
1−k! tsdt
t +ik
Z ∞
1
f i
√ Nt
tk−sdt
t −c+f(0)
s − c−f(0) N(1−k)/2
1 s−k+1. Using (1), forRe(s)>max{k,v+1}we have
ΛN(f,s) = Z ∞
1
f it
√ N
−c+f(0)−c−f (0) t
√ N
1−k! tsdt
t +ik
Z ∞
1
g it
√ N
tk−sdt
t −c+f(0)
s − c−f(0) N(1−k)/2
1 s−k+1
= Z ∞
1
f it
√ N
−c+f(0)−c−f (0) t
√ N
1−k! tsdt
t +ik
Z ∞
1
g it
√N
−c+g(0)−c−g(0) t
√N 1−k!
tk−sdt t
−c+f(0)
s − c−f(0) N(1−k)/2
1
s−k+1−c+g(0)ik
k−s − c−g(0)ik N(1−k)/2
1
1−s. (59)
By Lemma 15,Λ∗N(f,s)is entire and bounded in vertical strips. By the same computation, we see that Λ∗N(g,s) can be expressed by a similar integral representation as(59) and so is entire and bounded in vertical strips. Comparing two integral representations forΛN(f,s)andΛN(g,s), we obtainΛN(f,s) =ikΛN(g,k−s).
Next, we have
∂f
∂x(z) =2πi
∞
∑
n=0
nc+f (n)qn+
∑
n<0
nc−f(n)Γ(1−k,−4πny)qn
! ,
∂g
∂x(z) =2πi
∞
∑
n=0
nc+g(n)qn+
∑
n<0
nc−g(n)Γ(1−k,−4πny)qn
! .
For n≥1, we have
2πinc−f(−n) Z ∞
0
Γ(1−k,4πnt/√
N)e2πnt/
√N
ts+1dt t
=2πinc−f(−n) 2πn
√N
−s−1Z ∞
0
Γ(1−k,2t)etts+1dt t
=i
√
Nc−f(−n) 2πn
√ N
−s
W1−k(s+1).
So we have
Z ∞
0
H it
√ N
−kc+f(0)−kc−f(0) t
√ N
1−k! tsdt
t
= 2i
√N Z ∞
0
∂f
∂x it
√N
ts+1dt t +k
Z ∞
0
f it
√ N
−c+f(0)−c−f (0) t
√ N
1−k! tsdt
t
=−2ΞN(f,s) +kΛN(f,s)
=ΩN(f,s).
In a similar way to the calculation ofΛN(f,s)with Lemma 16, we can obtain an integral repre- sentation ofΩN(f,s), and by Lemma 15, we can see thatΩ∗N(f,s)andΩ∗N(g,s) are entire and bounded in any vertical strips. Comparing the integral representations ofΩN(f,s)andΩN(g,s), we obtainΩN(f,s) =−ikΩN(g,k−s).
(2)⇒(1) From(58), for any t>0andα >v+1we have an inverse Mellin transform 1
2πi Z α+i∞
α−i∞
t−sΛN(f,s)ds= f it
√ N
−c+f(0)−c−f(0) t
√ N
1−k
.
Combining Stirling’s approximation, Lemma 14 (2), and Phragmén-Lindelöf principle with the proof of Theorem 5, we obtain the following boundary condition forΛN(f,s)in the specific verti- cal strip as follows. Chooseβ such that k−β>v+1. Then for anyµ>0,
ΛN(f,s) =O(|Im(s)|−µ) as|Im(s)| →∞onβ ≤Re(s)≤α. (60)
Now assume thatα >1(and soβ <k−1). The function t−sΛN(f,s)has possibly simple poles at s=0,k,1,k−1with the residues−c+f(0),ikc+g(0)t−k, c−g(0)ik
N(1−k)/2t−1,− c−f(0)
N(1−k)/2t1−krespectively.
So we can shift the integral path fromRe(s) =α toRe(s) =β using(60), we obtain 1
2πi Z β+i∞
β−i∞
ΛN(f,s)t−sds= f it
√ N
−ikc+g(0)t−k− c−g(0)ik N(1−k)/2t−1. By the functional equation forΛN, we have
f|kωN(it) =g(it). (61)
Similar calculations gives us that, for any t>0andα >v+1we have 1
2πi Z β+i∞
β−i∞ ΩN(f,s)t−sds=− 2t i√ N
∂f
∂x it
√ N
+k f
it
√ N
−kc+f (0)−kc−f(0)ik N(1−k)/2t1−k, and by the functional equation forΩN, we have
H|kωN(it) =−I(it). (62)
Finally, we will finish the proof to apply the method of proof of Theorem 12.
In the proof of Lemma 15, we observed that the given Fourier expansion of f and g satisfies
∆k(f) =∆k(g) =0. Letting G= f|kωN−g, we have∆k(G) =0. Since∆k is elliptic, by Theo- rem 11 the function G(z)is real analytic and so G has a power series expansion in x of the form G(z) =∑∞n=0bn(y)xn. From∆k(G) =0, we get the following recurrence relation
bn+2(y) =iky(n+1)bn+1(y)−kyb0n(y)−y2bn(y)
(n+1)(n+2)y2 . (63)
To prove G(z) =0, it is sufficient to show that b0(y) =b1(y) =0. But we have b0(t) =G(it) = f|kωN(it)−g(it),
b1(t) =∂G
∂x(it) = ∂
∂x
N−k/2z−kf
− 1 Nz
(it)−∂g
∂x(it).
By(61)and(62), we have b0(t) =b1(t) =0. Thus by(63)we have bn(t) =0for all n≥0. This concludes that g= f|kωN.