Are you sure?
This action might not be possible to undo. Are you sure you want to continue?
Pierre Colmez
C.N.R.S. Institut de Math´ematiques de Jussieu
i
These are notes from a course given at Tsinghua University during the
fall of 2004. The aim of the course was to explain how to construct padic L
functions using the theory of (ϕ, Γ)modules of Fontaine. This construction
is an adaptation of an idea of PerrinRiou. The content of the course is
well reﬂected in the table of contents which is almost the only thing that I
modiﬁed from the notes taken and typed by the students Wang Shanwen,
Chen Miaofen, Hu Yongquan, Yin Gang, Li Yan and Hu Yong, under the
supervision of Ouyang Yi, all of whom I thank heartily. The course runs in
parallel to a course given by Fontaine in which the theory of (ϕ, Γ)modules
was explained as well as some topics from padic Hodge theory which are used
freely in these notes, which means that they are not entirely selfcontained.
Also, as time runs short at the end, the last chapter is more a survey than a
course. For a bibliography and further reading, the reader is referred to my
Bourbaki talk of June 2003 published in Ast´erisque 294.
ii
Contents
I Classical padic Lfunctions: zeta functions and
modular forms 1
1 The padic zeta function of KubotaLeopoldt 3
1.1 The Riemann zeta function at negative integers . . . . . . . . 3
1.2 padic Banach spaces . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Continuous functions on Z
p
. . . . . . . . . . . . . . . . . . . 7
1.3.1 Mahler’s coeﬃcients . . . . . . . . . . . . . . . . . . . 7
1.3.2 Locally constant functions. . . . . . . . . . . . . . . . . 9
1.4 Measures on Z
p
. . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.4.1 The Amice transform . . . . . . . . . . . . . . . . . . . 10
1.4.2 examples of measures on Z
p
and of operations on mea
sures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.5 The padic zeta function . . . . . . . . . . . . . . . . . . . . . 15
1.5.1 Kummer’s congruences. . . . . . . . . . . . . . . . . . . 15
1.5.2 Restriction to Z
∗
p
. . . . . . . . . . . . . . . . . . . . . . 16
1.5.3 Leopoldt’s Γtransform. . . . . . . . . . . . . . . . . . 17
1.6 (
k
functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.6.1 Deﬁnition. . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.6.2 Mahler’s coeﬃcients of (
r
functions. . . . . . . . . . . . 21
1.7 locally analytic functions . . . . . . . . . . . . . . . . . . . . . 23
1.7.1 Analytic functions on a closed disk. . . . . . . . . . . . 23
1.7.2 Locally analytic functions on Z
p
. . . . . . . . . . . . . 25
1.8 Distributions on Z
p
. . . . . . . . . . . . . . . . . . . . . . . . 27
1.8.1 The Amice transform of a distribution. . . . . . . . . . 27
1.8.2 Examples of distributions. . . . . . . . . . . . . . . . . 29
1.8.3 Residue at s = 1 of the padic zeta function. . . . . . . 30
1.9 Tempered distributions . . . . . . . . . . . . . . . . . . . . . . 31
iii
iv CONTENTS
1.9.1 Analytic functions inside (
r
functions . . . . . . . . . . 31
1.9.2 Distributions of order r . . . . . . . . . . . . . . . . . . 33
1.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2 Modular forms 39
2.1 Generalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.1.1 The upper halfplane . . . . . . . . . . . . . . . . . . . 39
2.1.2 Deﬁnition of modular forms . . . . . . . . . . . . . . . 40
2.1.3 qexpansion of modular forms. . . . . . . . . . . . . . . 40
2.1.4 Cusp forms. . . . . . . . . . . . . . . . . . . . . . . . . 41
2.2 The case Γ = SL
2
(Z) . . . . . . . . . . . . . . . . . . . . . . . 42
2.2.1 The generators S and T of SL
2
(Z). . . . . . . . . . . . 42
2.2.2 Eisenstein series . . . . . . . . . . . . . . . . . . . . . . 43
2.2.3 The fundamental domain for SL
2
(Z) . . . . . . . . . . 44
2.2.4 The
k
12
formula. . . . . . . . . . . . . . . . . . . . . . . 46
2.2.5 Dimension of spaces of modular forms. . . . . . . . . . 47
2.2.6 Rationality results. . . . . . . . . . . . . . . . . . . . . 48
2.3 The algebra of all modular forms. . . . . . . . . . . . . . . . . 50
2.4 Hecke operators . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.4.1 Preliminary. . . . . . . . . . . . . . . . . . . . . . . . . 53
2.4.2 Deﬁnition of Hecke operators: R
n
, T
n
, n ≥ 1. . . . . . . 54
2.4.3 Action of Hecke operators on modular forms. . . . . . . 56
2.5 Petersson scalar product. . . . . . . . . . . . . . . . . . . . . . 58
2.6 Primitive forms . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3 padic Lfunctions of modular forms 63
3.1 Lfunctions of modular forms. . . . . . . . . . . . . . . . . . . 63
3.1.1 Estimates for the fourier coeﬃcients . . . . . . . . . . . 63
3.1.2 Dirichlet series and Mellin transform . . . . . . . . . . 65
3.1.3 Modular forms and Lfunctions . . . . . . . . . . . . . 66
3.1.4 Euler products . . . . . . . . . . . . . . . . . . . . . . 68
3.2 Higher level modular forms . . . . . . . . . . . . . . . . . . . . 69
3.2.1 Summary of the results . . . . . . . . . . . . . . . . . . 69
3.2.2 TaniyamaWeil Conjecture . . . . . . . . . . . . . . . . 71
3.3 Algebraicity of special values of Lfunctions . . . . . . . . . . 71
3.3.1 Modular symbols. . . . . . . . . . . . . . . . . . . . . . 71
3.3.2 The results . . . . . . . . . . . . . . . . . . . . . . . . 73
3.3.3 Rankin’s method . . . . . . . . . . . . . . . . . . . . . 74
CONTENTS v
3.4 padic Lfunctions of modular forms . . . . . . . . . . . . . . . 77
II Fontaine’s rings and Iwasawa theory 83
4 Preliminaries 85
4.1 Some of Fontaine’s rings . . . . . . . . . . . . . . . . . . . . . 85
4.1.1 Rings of characteristic p . . . . . . . . . . . . . . . . . 85
4.1.2 Rings of characteristic 0 . . . . . . . . . . . . . . . . . 87
4.2 (ϕ, Γ)modules and Galois representations. . . . . . . . . . . . 89
5 (ϕ, Γ)modules and Galois cohomology 91
5.1 Galois Cohomology . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2 The complex C
ϕ,γ
(K, V ) . . . . . . . . . . . . . . . . . . . . . 92
5.3 Tate’s EulerPoincar´e formula. . . . . . . . . . . . . . . . . . . 95
5.3.1 The operator ψ. . . . . . . . . . . . . . . . . . . . . . . 95
5.3.2 D
ψ=1
and D/(ψ −1) . . . . . . . . . . . . . . . . . . . 98
5.3.3 The Γmodule D
ψ=0
. . . . . . . . . . . . . . . . . . . . 100
5.3.4 Computation of Galois chomology groups . . . . . . . . 103
5.3.5 The EulerPoincar´e formula. . . . . . . . . . . . . . . . 104
5.4 Tate’s duality and residues . . . . . . . . . . . . . . . . . . . . 105
6 (ϕ, Γ)modules and Iwasawa theory 109
6.1 Iwasawa modules H
i
Iw
(K, V ) . . . . . . . . . . . . . . . . . . . 109
6.1.1 Projective limits of cohomology groups . . . . . . . . . 109
6.1.2 Reinterpretation in terms of measures . . . . . . . . . . 110
6.1.3 Twist by a character (`a la Soul´e) . . . . . . . . . . . . 111
6.2 Description of H
i
Iw
in terms of D(V ) . . . . . . . . . . . . . . 112
6.3 Structure of H
1
Iw
(K, V ) . . . . . . . . . . . . . . . . . . . . . . 115
7 Z
p
(1) and KubotaLeopoldt zeta function 117
7.1 The module D(Z
p
(1))
ψ=1
. . . . . . . . . . . . . . . . . . . . . 117
7.2 Kummer theory . . . . . . . . . . . . . . . . . . . . . . . . . . 118
7.3 Coleman’s power series . . . . . . . . . . . . . . . . . . . . . . 119
7.4 An explicit reciprocity law . . . . . . . . . . . . . . . . . . . . 122
7.5 Proof of the explicit reciprocity law . . . . . . . . . . . . . . . 123
7.5.1 Strategy of proof of Theorem 7.4.1 . . . . . . . . . . . 123
7.5.2 Explicit formulas for cocyles . . . . . . . . . . . . . . . 125
vi CONTENTS
7.5.3 Tate’s normalized trace maps . . . . . . . . . . . . . . 127
7.5.4 Applications to Galois cohomology . . . . . . . . . . . 130
7.5.5 No 2πi in C
p
! . . . . . . . . . . . . . . . . . . . . . . . 131
8 (ϕ, Γ)modules and padic Lfunctions 133
8.1 TateSen’s conditions . . . . . . . . . . . . . . . . . . . . . . . 133
8.1.1 The conditions (TS1), (TS2) and (TS3) . . . . . . . . . 133
8.1.2 Example : the ﬁeld C
p
. . . . . . . . . . . . . . . . . . 134
8.2 Sen’s method . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
8.2.1 Almost ´etale descent . . . . . . . . . . . . . . . . . . . 136
8.2.2 Decompletion . . . . . . . . . . . . . . . . . . . . . . . 138
8.2.3 Applications to padic representations . . . . . . . . . . 140
8.3 Overconvergent (ϕ, Γ)modules . . . . . . . . . . . . . . . . . . 141
8.3.1 Overconvergent elements . . . . . . . . . . . . . . . . . 141
8.3.2 Overconvergent representations . . . . . . . . . . . . . 145
8.3.3 padic Hodge theory and (ϕ, Γ)modules . . . . . . . . 147
8.3.4 A map of the land of the rings . . . . . . . . . . . . . . 148
8.4 Explicit reciprocity laws and padic Lfunctions . . . . . . . . 149
8.4.1 Galois cohomology of B
dR
. . . . . . . . . . . . . . . . 149
8.4.2 BlochKato’s dual exponential maps . . . . . . . . . . . 150
8.4.3 The explicit reciprocity law . . . . . . . . . . . . . . . 152
8.4.4 Cyclotomic elements and CoatesWiles morphisms. . . 154
8.4.5 Kato’s elements and padic Lfunctions of modular forms.155
Part I
Classical padic Lfunctions:
zeta functions and modular
forms
1
Chapter 1
The padic zeta function of
KubotaLeopoldt
1.1 The Riemann zeta function at negative
integers
We ﬁrst recall the deﬁnitions of Riemann zeta function and the classical
Gamma function:
ζ(s) =
+∞
¸
n=1
n
−s
=
¸
p
(1 −p
−s
)
−1
, if Re (s) > 1.
Γ(s) =
+∞
0
e
−t
t
s
dt
t
, if Re (s) > 0.
The Γfunction has the following properties:
(i) Γ(s+1) = sΓ(s), which implies that Γ has a meromorphic continuation
to C with simple poles at negative integers and 0.
(ii) Γ(n) = (n −1)! if n ≥ 1.
(iii) Γ(s)Γ(1 − s) =
π
sin(πs)
, which implies that
1
Γ(s)
is an entire(or holo
morphic) function on C with simple zeros at −n for n ∈ N.
(iv) Γ(
1
2
) =
√
π.
3
4CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Then we have the following formulas:
n
−s
=
1
Γ(s)
+∞
0
e
−nt
t
s
dt
t
,
ζ(s) =
1
Γ(s)
+∞
0
+∞
¸
n=1
e
−nt
t
s
dt
t
=
1
Γ(s)
+∞
0
1
e
t
−1
t
s
dt
t
.
Lemma 1.1.1. If f : R
+
→ C is a (
∞
function on R
+
, rapidly decreasing
(i.e., t
n
f(t) →0 when t →+∞ for all n ∈ N), then
L(f, s) =
1
Γ(s)
+∞
0
f(t)t
s
dt
t
, Re(s) > 0
has an analytic continuation to C, and
L(f, −n) = (−1)
n
f
(n)
(0).
Proof. Choose a (
∞
function φ on R
+
, such that φ(t) = 1 for t ∈ [0, 1] and
φ(t) = 0 for t ≥ 2.
Let f = f
1
+ f
2
, where f
1
= φf, f
2
= (1 − φ)f. Then
∞
0
f
2
(t)t
s dt
t
is
holomorphic on C, hence L(f
2
, s) is also holomorphic and L(f
2
, −n) = 0 =
f
(−n)
2
(0). Since, for Re (s) > 0,
L(f
1
, s) =
1
Γ(s)
[f
1
(t)
t
s
s
][
+∞
0
−
1
sΓ(s)
+∞
0
f
t
1
(t)t
s+1
dt
t
= −L(f
t
1
(t), s + 1) = (−1)
n
L(f
(n)
1
, s + n),
we get analytic continuation for f
1
and hence for f, moreover,
L(f, −n) = L(f
1
, −n) = (−1)
n+1
L(f
(n+1)
1
, 1)
= (−1)
n+1
+∞
0
f
(n+1)
1
(t)dt = (−1)
n
f
(n)
1
(0) = (−1)
n
f
(n)
(0).
We now apply the above lemma to the function f(t) =
t
e
t
−1
. Note that
f(t) =
∞
¸
0
B
n
t
n
n!
,
1.2. PADIC BANACH SPACES 5
where B
n
∈ Q is the nth Bernoulli number with value:
B
0
= 1, B
1
= −
1
2
, B
2
=
1
6
, B
3
= 0, B
4
= −
1
30
, B
5
= 0
Since f(t) −f(−t) = −t, we have B
2k+1
= 0 if k ≥ 1. Now :
ζ(s) =
1
Γ(s)
+∞
0
f(t)t
s−1
dt
t
=
1
s −1
L(f, s −1),
so we obtain the following result.
Theorem 1.1.2. (i) ζ has a meromorphic continuation to C. It is holomor
phic except for a simple pole at s = 1 with residue L(f, 0) = B
0
= 1.
(ii) If n ∈ N, then
ζ(−n) =
−1
n + 1
L(f, −n −1) =
(−1)
n
n + 1
f
(n+1)
(0)
= (−1)
n
B
n+1
n + 1
∈ Q
= −
B
n+1
n + 1
if n ≥ 2
.
Theorem 1.1.3 (Kummer). If p does not divide the numerators of ζ(−3),
ζ(−5), , ζ(2 −p), then the class number of Q(u
p
) is prime to p.
Remark. This theorem and a lot of extra work implies Fermat’s Last The
orem for these regular primes. We will not prove it in these notes, but we
will focus on the following result, also discovered by Kummer, which plays
an important role in the proof.
Theorem 1.1.4 (Kummer’s congruences). Let a ≥ 2 be prime to p. Let
k ≥ 1. If n
1
, n
2
≥ k such that n
1
≡ n
2
mod (p −1)p
k−1
, then
(1 −a
1+n
1
)ζ(−n
1
) ≡ (1 −a
1+n
2
)ζ(−n
2
) mod p
k
.
1.2 padic Banach spaces
Deﬁnition 1.2.1. A padic Banach space B is a Q
p
vector space with a
lattice B
0
(Z
p
module) separated and complete for the padic topology, i.e.,
B
0
· lim
←−
n∈N
B
0
/p
n
B
0
.
6CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
For all x ∈ B, there exists n ∈ Z, such that x ∈ p
n
B
0
. Deﬁne
v
B
(x) = sup
n∈N∪+∞¦
¦n : x ∈ p
n
B
0
¦.
It satisﬁes the following properties:
v
B
(x + y) ≥ min(v
B
(x), v
B
(y)),
v
B
(λx) = v
p
(λ) + v
B
(x), if λ ∈ Q
p
.
Then  x 
B
= p
−v
B
(x)
deﬁnes a norm on B, such that B is complete for 
B
and B
0
is the unit ball.
Example 1.2.2. (i) B = C
p
=
´
Q
p
, B
0
= O
Cp
, v
B
(x) = [v
p
(x)] ∈ Z;
(ii) The space B = (
0
(Z
p
, Q
p
) of continuous functions on Z
p
. B
0
=
(
0
(Z
p
, Z
p
) is a lattice, and v
B
(f) = inf
x∈Z
v
p
(f(x)) = −∞ because Z
p
is com
pact.
(iii) Let B = (
0
(Z
p
, C
p
), B
0
= C
0
(Z
p
, O
Cp
); v
B
(f) = inf
x∈Z
[v
p
(f(x))].
Deﬁnition 1.2.3. A Banach basis of a padic Banach space B is a family
(e
i
)
i∈I
of elements of B, satisfying the following conditions:
(i) For every x ∈ B, x =
¸
i∈I
x
i
e
i
, x
i
∈ Q
p
in a unique way with x
i
→ 0
when i →∞; equivalently for any C, the set ¦i [ v
p
(x
i
) ≤ C¦ is a ﬁnite set.
(ii) v
B
(x) = inf
i∈I
v
p
(x
i
).
Theorem 1.2.4. A family (e
i
)
i∈I
of elements of B is a Banach basis if and
only if
(i) e
i
∈ B
0
for all i;
(ii) the set ( ¯ e
i
)
i∈I
form a basis of B
0
/pB
0
as a F
p
vector space.
Proof. We leave the proof of the theorem as an exercise.
Let B and B
t
be two padic Banach spaces with Banach basis (e
i
)
i∈I
and
(f
j
)
j∈J
respectively, then B
´
¸
B
t
is a padic Banach space with Banach basis
(e
i
⊗f
j
)
(i,j)∈IJ
. Thus for all x ∈ B
´
¸
B
t
,
x =
¸
i,j
x
i,j
e
i
⊗f
j
(x
i,j
∈ Q
p
, x
i,j
→0 as (i, j) →∞)
=
¸
j
y
j
⊗f
j
(y
j
∈ B, y
j
→0 as j →∞)
=
¸
i
e
i
⊗z
i
(z
i
∈ B
t
, z
i
→0 as i →∞).
1.3. CONTINUOUS FUNCTIONS ON Z
P
7
Exercise. (
0
(Z
p
, C
p
) = C
p
´
¸
(
0
(Z
p
, Q
p
).
1.3 Continuous functions on Z
p
1.3.1 Mahler’s coeﬃcients
We have the binomial function:
x
n
=
1, if n = 0,
x(x −1) (x −n + 1)
n!
, if n ≥ 1.
Lemma 1.3.1. v
(
0(
x
n
) = 0.
Proof. Since
n
n
= 1, v
(
0(
x
n
) ≤ 0.
If x ∈ N, then
x
n
∈ N implies v
p
(
x
n
) ≥ 0. Hence for all x ∈ Z
p
,
v
p
(
x
n
) ≥ 0 because N is dense in Z
p
.
For all f ∈ (
0
(Z
p
, Q
p
), we write
f
[0]
= f, f
[k−1]
(x) = f
[k]
(x + 1) −f
[k]
(x)
and write the Mahler’s coeﬃcient
a
n
(f) = f
[n]
(0).
Hence:
f
[n]
(x) =
n
¸
i=0
(−1)
i
n
i
f(x + n −i),
a
n
(f) =
n
¸
i=0
(−1)
i
n
i
f(n −i).
Theorem 1.3.2 (Mahler). If f ∈ (
0
(Z
p
, Q
p
), then
(i) lim
n→∞
v
p
(a
n
(f)) = +∞,
(ii) For all x ∈ Z
p
, f(x) =
∞
¸
n=0
a
n
(f)
x
n
,
(iii) v
(
0(f) = inf v
p
(a
n
(f)).
8CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Proof. Let
∞
= ¦a = (a
n
)
n∈N
: a
n
∈ Q
p
bounded¦, v
∞
(a) = inf
n∈N
v
p
(a
n
).
Then
• f → a(f) = (a
n
(f))
n∈N
is a continuous map from (
0
(Z
p
, Q
p
) to
∞
.
and v
∞
(a(f)) ≥ v
C
0(f).
• The space
0
∞
= ¦(a
n
)
n∈N
: a
n
→0, as n →∞¦ is a closed subspace of
∞
and B = ¦f : a(f) ∈
0
∞
¦ is a close subspace of (
0
(Z
p
, Q
p
).
• For all a ∈
0
∞
,
f
a
=
+∞
¸
n=0
a
n
x
n
∈ (
0
(Z
p
, Q
p
)
because the series converges uniformly. Moreover, v
(
0(f
a
) ≥ v
∞
(a)
and as
x+1
n+1
−
x
n+1
=
x
n
,
f
[k]
a
=
+∞
¸
n=0
a
n+k
x
n
.
Hence we have: a
k
(f) = f
[k]
(0) = a
k
, which implies a(f
a
) = a.
• f → a(f) is injective. Since a(f) = 0 implies f(n) = 0 for all n ∈ N.
Hence f = 0 by the density of N in Z
p
.
Now for f ∈ B, a(f) ∈
0
∞
implies f − f
a(f)
= 0 because a(f − f
a(f)
) =
a(f) − a(f) = 0 and a is injective. So f ∈ B implies that f satisﬁes (ii).
Moreover, since
v
∞
(a(f)) ≥ v
(
0(f) = v
(
0(f
a(f)
) ≥ v
∞
(a(f)),
(iii) is also true. It remains to show that:
Claim: B = (
0
(Z
p
, Q
p
).
(a) First proof. We have a lemma:
Lemma 1.3.3. If f ∈ (
0
(Z
p
, Q
p
), then there exists k ∈ N such that
v
(
0(f
[p
k
]
) ≥ v
(
0(f) + 1.
1.3. CONTINUOUS FUNCTIONS ON Z
P
9
Proof. We have
f
[p
k
]
(x) = f(x+p
k
) −f(x) +
p
k
−1
¸
i=1
(−1)
i
p
k
i
f(x+p
k
−i) +(1 +(−1)
p
k
)f(x).
Now v
p
(
p
k
i
) ≥ 1, if 1 ≤ i ≤ p
k
−1 et v
p
(1+(−1)
p
k
) ≥ 1. Since Z
p
is compact,
f is uniformly continuous. For every c, there exists N, when v
p
(x −y) ≥ N,
we have v
p
(f(x) −f(y)) ≥ c. It gives the result for k = N.
First proof of the Claim. Repeat the lemma: for every c = v
(
0(f) +k, there
exists an N, such that v
(
0(f
[N]
) ≥ c. Hence, for all n ≥ N, v
p
(a
n
(f)) ≥ c.
1.3.2 Locally constant functions.
Choose a z ∈ C
p
, such that v
p
(z −1) > 0. Then
f
z
(x) =
+∞
¸
n=0
x
n
(z −1)
n
∈ (
0
(Z
p
, C
p
).
Note k ∈ N, f
z
(k) = z
k
. So we write, f
z
(x) = z
x
and we have z
x+y
= z
x
z
y
.
Example 1.3.4. (i) z
1
2
=
+∞
¸
n=0
1
2
n
(z − 1)
n
. z =
16
9
, z − 1 =
7
9
, the series
converges in R to
4
3
, and converges in Q
7
to −
4
3
.
(ii) If z is a primitive p
n
th root of 1, then
v
p
(z −1) =
1
(p −1)p
n−1
> 0.
Note that z
x+p
n
= z
x
for all x, then z
x
is locally constant( constant mod p
n
Z
p
).
The characteristic function of i + p
n
Z
p
is given by
1
i+p
n
Zp
(x) =
1
p
n
¸
z
p
n
=1
z
−i
z
x
since
¸
z
p
n
=1
z
x
=
p
n
if x ∈ p
n
Z
p
;
0 if not.
10CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Lemma 1.3.5. The set of locally constant functions LC(Z
p
, Q
p
) ⊂ B.
Proof. By compactness of Z
p
, a locally constant function is a linear combi
nation of 1
i+p
n
Z
pz
x
, z ∈ µ
p
∞, thus a linear combination of z
x
. But a
n
(z
x
) =
(z −1)
n
goes to 0, when n goes to ∞, hence z
x
∈ B.
Lemma 1.3.6. LC(Z
p
, Q
p
) is dense in (
0
(Z
p
, Q
p
).
Proof. For every f ∈ (
0
(Z
p
, Q
p
), let
f
k
=
p
k
−1
¸
i=0
f(i)1
i+p
k
Zp
.
Then f
k
→f in (
0
because f is uniformly continuous.
Second proof of the Claim. By the above two lemmas, LC(Z
p
, Q
p
) ⊂ B ⊂
(
0
(Z
p
, Q
p
), B is closed and LC(Z
p
, Q
p
) is dense in (
0
(Z
p
, Q
p
), hence B =
(
0
(Z
p
, Q
p
).
1.4 Measures on Z
p
1.4.1 The Amice transform
Deﬁnition 1.4.1. A measure µ on Z
p
with values in a padic Banach space
B is a continuous linear map f →
Zp
f(x)µ =
Zp
f(x)µ(x) from C
0
(Z
p
, Q
p
)
to B.
Remark. (i) If L ⊂ C
p
is a closed subﬁeld and B is an Lvector space, then
µ extends by continuity and Llinearity to (
0
(Z
p
, L) = L
´
¸
C
0
(Z
p
, Q
p
).
(ii) We denote T
0
(Z
p
, B) the set of the measure on Z
p
with values in B,
then T
0
(Z
p
, B) = T
0
(Z
p
, Q
p
)
´
¸
B.
Deﬁnition 1.4.2. The Amice transform of a measure µ is deﬁned to be the
map:
µ →A
µ
(T) =
Zp
(1 +T)
x
µ(x) =
+∞
¸
n=0
T
n
Zp
x
n
µ.
Lemma 1.4.3. If v
p
(z −1) > 0, A
µ
(z −1) =
Zp
z
x
µ(x).
1.4. MEASURES ON Z
P
11
Proof. Since z
x
=
+∞
¸
n=0
(z −1)
n
x
n
with normal convergence in (
0
(Z
p
, Q
p
), one
can exchange
¸
and
.
Deﬁnition 1.4.4. The valuation on T
0
is
v
T
0
(µ) = inf
f,=0
(v
p
(
Zp
fµ) −v
(
0(f)).
Theorem 1.4.5. The map µ → A
µ
is an isometry from T
0
(Z
p
, Q
p
) to the
set ¦
+∞
¸
n=0
b
n
T
n
, b
n
bounded, and b
n
∈ Q
p
¦ with the valuation v(
+∞
¸
n=0
b
n
T
n
) =
inf
n∈N
v
p
(b
n
).
Proof. On one hand, for all µ ∈ T
0
(Z
p
, Q
p
), write A
µ
(T) =
+∞
¸
n=0
b
n
(µ)T
n
, then
b
n
(µ) =
Zp
x
n
µ. Since v
(
0(
x
n
) = 0 by Lemma 1.3.1,
v
p
(b
n
(µ)) ≥ v
T
0
(µ) + v
(
0(
x
n
) ≥ v
T
0
(µ)
for all n, hence v(A
µ
) ≥ v
T
0
(µ).
On the other hand, if (b
n
)
n∈N
is bounded, f →
+∞
¸
n=0
b
n
a
n
(f)(by Mahler’s
theorem, a
n
(f) →0) gives a measure µ
b
whose Amice transform is
A
µ
b
(T) =
+∞
¸
n=0
T
n
Zp
x
n
µ
b
=
+∞
¸
n=0
T
n
(
+∞
¸
i=0
b
i
a
i
(
x
n
)) =
+∞
¸
n=0
b
n
T
n
since
a
n
(
x
i
) =
1 if i = n,
0 otherwise.
Hence
v
p
(
+∞
¸
n=0
b
n
a
n
(f)) ≥ min
n
(v
p
(b
n
) + v
p
(a
n
(f)))
≥ min
n
(v
p
(b
n
)) + min
n
(a
n
(f))
= v(
¸
b
n
T
n
) + v
(
0(f)
= v(A
µ
) + v
(
0(f).
Thus v
T
0
(µ
b
) ≥ v(A
µ
). Then we have v(A
µ
) = v
T
0
(µ).
12CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
By Lemma 1.3.6, we know that locally constant functions are dense in
(
0
(Z
p
, Q
p
). Explicitly, for all f ∈ (
0
(Z
p
, Q
p
), the locally constant functions
f
n
=
p
n
−1
¸
i=0
f(i)1
i+p
n
Zp
→f in (
0
.
Now if µ ∈ T
0
(Z
p
, Q
p
), set µ(i + p
n
Z
p
) =
Zp
1
i+p
n
Zp
µ. Then
Zp
f µ is
given by the following “Riemann sums”
Zp
fµ = lim
n→∞
p
n
−1
¸
i=0
f(i)µ(i + p
n
Z
p
) (1.1)
Note that v
p
(µ(i + p
n
Z
p
)) ≥ v
T
0
(µ).
Theorem 1.4.6. If µ is an additive bounded function on compact open sub
sets of Z
p
(by compactness of Z
p
is a ﬁnite disjoint union of i + p
n
Z
p
for
some n), then µ extends uniquely as a measure on Z
p
via (1.1).
Proof. Since µ is an additive function on compact open subsets, µ is linear
on locally constant functions. And µ is bounded, hence µ is continuous for
v
(
0. As the locally constant functions are dense in (
0
(Z
p
, Q
p
), we have µ as
a measure on Z
p
.
1.4.2 examples of measures on Z
p
and of operations on
measures.
Example 1.4.7. Haar measure: µ(Z
p
) = 1 and µ is invariant by translation.
We must have µ(i + p
n
Z
p
) =
1
p
n
which is not bounded. Hence, there exists
no Haar measure on Z
p
.
Example 1.4.8. Dirac measure: For a ∈ Z
p
, we deﬁne δ
a
by
Zp
f(x)δ
a
=
f(a). The Amice transform of δ
a
is A
δa
(T) = (1 +T)
a
.
Example 1.4.9. Multiplication of a measure by a continuous function. For
µ ∈ T
0
, f ∈ (
0
, we deﬁne the measure fµ by
Zp
g fµ =
Zp
f(x)g(x)µ
for all g ∈ (
0
.
1.4. MEASURES ON Z
P
13
(i) Let f(x) = x, since
x
x
n
= (x −n + n)
x
n
= (n + 1)
x
n + 1
+ n
x
n
,
the Amice transform is
A
xµ
=
+∞
¸
n=0
T
n
Zp
x
n
xµ
=
+∞
¸
n=0
T
n
¸
(n + 1)
Zp
x
n + 1
µ + n
Zp
x
n
µ
¸
= (1 +T)
d
dT
A
µ
.
(ii) Let f(x) = z
x
, v
p
(z −1) > 0. For any y, v
p
(y −1) > 0, then
Zp
y
x
(z
x
µ) =
Zp
(yz)
x
µ = A
µ
(yz −1)
which implies that
A
z
x
µ
(T) = A
µ
((1 + T)z −1).
(iii) The restriction to a compact open set X of Z
p
: it is nothing but the
multiplication by 1
X
. If X = i + p
n
Z
p
, then 1
i+p
n
Zp
(x) = p
−n
¸
z
p
n
=1
z
−i
z
x
,
hence
A
Res
i+p
n
Zp
µ
(T) = p
−n
¸
z
p
n
=1
z
−i
A
µ
((1 + T)z −1).
Example 1.4.10. Actions of ϕ and ψ. For µ ∈ T
0
, we deﬁne the action of
ϕ on µ by
Zp
f(x)ϕ(µ) =
Zp
f(px)µ.
Hence
A
ϕ(µ)
(T) =
+∞
¸
n=0
T
n
Zp
px
n
µ = A
µ
((1 + T)
p
−1) = ϕ(A
µ
(T))
where ϕ : T →(1 +T)
p
−1 (compare this formula with (ϕ, Γ)modules). We
deﬁne the action of ψ by
Zp
f(x)ψ(µ) =
Zp
f(
x
p
)µ.
14CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Then A
ψ(µ)
= ψ(A
µ
) where
ψ(F)((1 + T)
p
−1) =
1
p
¸
z
p
=1
F((1 + T)z −1).
The actions ϕ and ψ satisfy the following properties:
(i) ψ ◦ ϕ = Id;
(ii) ψ(µ) = 0 ⇔µ has a support in Z
∗
p
;
(iii) Res
Z
∗
p
(µ) = (1 −ϕψ)µ.
The map ψ is very important in the theory of (ϕ, Γ)modules.
Example 1.4.11. Action of Γ. Let Γ = Gal(Q
p
(µ
p
∞)/Q
p
). Let χ : Γ
∼
→Z
∗
p
be the cyclotomic character. For γ ∈ Γ and µ ∈ T
0
, we let γµ be given by
Zp
f(x)γµ =
Zp
f(χ(γ)x)µ.
One can verify that A
γµ
(T) = A
µ
((1 + T)
χ(γ)
− 1) = γ(A
µ
(T)) for γ(T) =
(1 + T)
χ(γ)
−1. (Compare this formula with (ϕ, Γ)modules.)
For all γ ∈ Γ, γ commutes with φ and ψ.
Example 1.4.12. Convolution λ ∗ µ. Let λ, µ be two measures, their con
volution λ ∗ µ is deﬁned by
Zp
f(x)λ ∗ µ =
Zp
(
Zp
f(x + y)µ(x))λ(y).
Here we have to verify y →
Zp
f(x + y)µ(x) ∈ (
0
, which is a direct conse
quence of the fact f is uniformly continuous.
Let f(x) = z
x
, v
p
(z −1) > 0, then
Zp
z
x
λ ∗ µ =
Zp
z
x
µ(x)
Zp
z
y
λ(y),
thus A
λ∗µ
= A
λ
A
µ
.
1.5. THE PADIC ZETA FUNCTION 15
1.5 The padic zeta function
1.5.1 Kummer’s congruences.
Lemma 1.5.1. For a ∈ Z
∗
p
, there exists a measure λ
a
∈ T
0
such that
A
λa
=
Zp
(1 +T)
x
λ
a
=
1
T
−
a
(1 + T)
a
−1
.
Proof. This follows from Theorem 1.4.5 and the fact
a
(1 +T)
a
−1
=
a
¸
∞
n=1
a
n
T
n
=
1
T
1
1 +
¸
∞
n=2
a
−1
a
n
T
n−1
∈
1
T
+Z
p
[[T]]
since a
−1
a
n
∈ Z
p
. Moreover, we have v
T
0
(λ
a
) = 0.
Proposition 1.5.2. For every n ∈ N,
Zp
x
n
λ
a
= (−1)
n
(1 −a
1+n
)ζ(−n).
Proof. For a ∈ R
∗
+
, for T = e
t
−1, let
f
a
(t) = A
λa
(T) =
1
e
t
−1
−
a
e
at
−1
,
then f
a
is in (
∞
on R
+
and rapidly decreasing. Hence
L(f
a
, s) =
1
Γ(s)
+∞
0
f
a
(t)t
s
dt
t
= (1 −a
1−s
)ζ(s)
f
n
a
(0) = (−1)
n
L(f
a
, −n) = (−1)
n
(1 −a
1+n
)ζ(−n)
The identity f
n
a
(0) = (−1)
n
(1 − a
1+n
)ζ(−n) is algebric, so is true for all a,
hence even on Z
∗
p
. Thus
Zp
x
n
λ
a
= (
d
dt
)
n
(
Zp
e
tx
λ
a
)[
t=0
= (
d
dt
)
n
A
λa
(e
t
−1)[
t=0
= f
(n)
a
(0).
Corollary 1.5.3. For a ∈ Z
∗
p
, k ≥ 1 (k ≥ 2 if p = 2), n
1
, n
2
≥ k, n
1
≡
n
2
mod (p −1)p
k−1
, then
v
p
((1 −a
1+n
1
)ζ(−n
1
) −(1 −a
1+n
2
)ζ(−n
2
)) ≥ k.
16CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Proof. The left hand side LHS = v
p
((1 −a
1+n
1
)ζ(−n
1
) −(1 −a
1+n
2
)ζ(−n
2
))
is
v
p
(
Zp
(x
n
1
−x
n
2
)λ
a
) ≥ v
T
0
(λ
a
) + v
(
0(x
n
1
−x
n
2
).
From the proof of Lemma 1.5.1, v
T
0
(λ
a
) = 0, thus LHS ≥ v
(
0(x
n
1
−x
n
2
). It
suﬃces to show v
(
0(x
n
1
−x
n
2
) ≥ k. There are two cases:
If x ∈ pZ
p
, then v
p
(x
n
1
) ≥ k and v
p
(x
n
2
) ≥ k since n
1
, n
2
≥ k.
If x ∈ Z
∗
p
, v
p
(x
n
1
−x
n
2
) ≥ k because (Z/p
k
Z)
∗
has order (p −1)p
k−1
and
n
1
−n
2
is divisible by (p −1)p
k−1
.
Remark. The statement is not clean because of x ∈ pZ
p
.
1.5.2 Restriction to Z
∗
p
.
Lemma 1.5.4. ψ(
1
T
) =
1
T
.
Proof. Let F(T) = ψ(
1
T
), then
F((1 + T)
p
−1) =
1
p
¸
z
p
=1
1
(1 +T)z −1
=
−1
p
¸
z
p
=1
+∞
¸
n=0
((1 + T)z)
n
=−
+∞
¸
n=0
(1 +T)
pn
=
1
(1 +T)
p
−1
.
Proposition 1.5.5. ψ(λ
a
) = λ
a
.
Proof. We only need to show the same thing on the Amice transform, but
A
λa
(T) =
1
T
−
a
(1 +T)
a
−1
=
1
T
−a γ
a
(
1
T
)
where γ
a
∈ Γ is the inverse of a by χ : Γ →Z
∗
p
, i.e., χ(γ
a
) = a. Since ψ and
γ
a
commutes and ψ(
1
T
) =
1
T
, we have
ψ(A
λa
) =
1
T
−aγ
a
(
1
T
) = A
λa
.
1.5. THE PADIC ZETA FUNCTION 17
Corollary 1.5.6. (i) Res
Z
∗
p
(λ
a
) = (1 −φψ)λ
a
= (1 −φ)λ
a
,
(ii)
Z
∗
p
x
n
λ
a
=
Zp
x
n
(1 −φ)λ
a
= (−1)
n
(1 −a
n+1
)(1 −p
n
)ζ(−n).
Remark. The factor (1 −p
n
) is the Euler factor of the zeta function at p.
Theorem 1.5.7. For i ∈ Z/(p − 1)Z (or i ∈ Z/2Z if p = 2), there exists a
unique function ζ
p,i
, analytic on Z
p
if i = 1, and (s −1)ζ
p,1
(s) is analytic on
Z
p
, such that ζ
p,i
(−n) = (1 −p
n
)ζ(−n) if n ≡ −i mod p −1 and n ∈ N.
Remark. (i) If i ≡ 0 mod 2, then ζ
p,i
= 0 since ζ(−n) = 0 for n even and
≥ 2;
(ii) To get padic continuity, one has to modify ζ by some “Euler factor
at p”.
(iii) Uniqueness is trivial because N is inﬁnite and Z
p
is compact.
(iv) The existence is kind of a miracle. Its proof relies on Leopoldt’s
Γtransform.
1.5.3 Leopoldt’s Γtransform.
Lemma 1.5.8. (i) Every x ∈ Z
∗
p
can be written uniquely as x = ω(x)'x`,
with
ω(x) ∈ µ(Q
p
) =
¦±1¦ if p = 2,
µ
p−1
, if p = 2
and 'x` ∈ 1 + 2pZ
p
.
(ii) ω(xy) = ω(x)ω(y), 'xy` = 'x`'y`.
Proof. If p = 2, it is obvious.
If p = 2, ω(x) = lim
n→∞
x
p
n
= [¯ x].
Remark. (i) ω is the socalled Teichm¨ uller character;
(ii) 'x` = exp(log(x));
(iii) x
n
= ω(x)
n
'x`
n
, here 'x`
n
is the restriction to N of 'x`
s
which is
continuous in s, ω(x)
n
is periodic of period p − 1, which is not padically
continuous.
Proposition 1.5.9. If λ is a measure on Z
∗
p
, u = 1 +2p, then there exists a
measure Γ
(i)
λ
on Z
p
(Leopoldt’s transform) such that
Z
∗
p
ω(x)
i
'x`
s
λ(x) =
Zp
u
sy
Γ
(i)
λ
(y) = A
Γ
(i)
λ
(u
s
−1).
18CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Proof. We have
Z
∗
p
ω(x)
i
'x`
s
λ(x) =
¸
ε∈µ(Qp)
ω(ε)
i
ε+2pZp
'x`
s
λ(x)
=
¸
ε∈µ(Qp)
ω(ε)
i
1+2pZp
'xε`
s
γ
ε
−1 λ(x),
where γ
ε
∈ Γ is such that χ(γ
ε
) = ε. We have a isomorphism
α : 1 + 2pZ
p
· Z
p
x → y =
log(x)
log(u)
.
Then
Zp
f(y)α
∗
(γ
ε
−1λ) =
1+2pZp
f(α(x))γ
ε
−1λ.
Now 'x`
s
= exp(s log x) = exp(s log uy) = u
sy
and hence
¸
ε∈µ(Qp)
ω(ε)
i
1+2pZp
'xε`
s
λ(x) =
¸
ε∈µ(Qp)
ω(ε)
i
Zp
u
sy
α
∗
(γ
ε
−1 λ),
we just set Γ
(i)
λ
=
¸
ε∈µ(Qp)
ω(ε)
i
α
∗
(γ
ε
−1 λ).
Deﬁnition 1.5.10.
ζ
p,i
(s) =
−1
1 −ω(a)
1−i
'a`
1−s
Z
∗
p
ω(x)
−i
'x`
−s
λ
a
(x).
Proof of Theorem 1.5.7. If n ≡ −i mod p −1, then
ζ
p,i
(−n) =
1
1 −ω(a)
1−i
'a`
1+n
Z
∗
p
ω(x)
−i
'x`
n
λ
a
(x)
=
1
1 −ω(a)
1+n
'a`
1+n
Z
∗
p
ω(x)
n
'x`
n
λ
a
(x)
= (1 −p
−n
)ζ(−n).
The function ζ
p,i
is analytic if ω(a)
1−i
= 1, which can be achieved if i = 1.
If i = 1, there is a pole at s = 1.
1.6. (
K
FUNCTIONS 19
Remark. (i) A theorem of Mazur and Wiles (originally the Main conjecture
of Iwasawa theory) describes the zeros of ζ
p,i
(s) in terms of ideal class groups
of Q
p
(µ
p
n), n ∈ N.
(ii) Main open question: For i ≡ 1 mod 2, can ζ
p,i
(k) = 0, if k > 1 and
k ∈ N?
The case k = 1 is known. In this case, ζ
p,i
(1) is a linear combination
with coeﬃcients in
¯
Q
of log of algebraic numbers, hence by transcendental
number theory (Baker’s theorem), ζ
p,i
(1) = 0.
1.6 (
k
functions
1.6.1 Deﬁnition.
Let f : Z
p
→Q
p
be a given function. We deﬁne
f
0¦
(x) = f(x)
f
i¦
(x, h
1
, , h
i
)
=
1
h
i
(f
i−1¦
(x + h
i
, h
1
, , h
i−1
) −f
i−1¦
(x, h
1
, , h
i−1
))
=
1
h
1
h
i
(
¸
I⊂1, ,i¦
(−1)
i−[I[
f(x +
¸
j∈I
h
j
))
One notes that f
i¦
is the analogue of the usual derivation in ((R, C). In
fact, if f : R →C is in (
k
and i ≤ k, deﬁne f
i¦
by the above formula, then
f
i¦
(x, h
1
, , h
i
) =
[0,1]
i
f
(i)
(x + t
1
h
1
+ + t
i
h
i
)dt
1
dt
i
,
hence f
i¦
is continuous and f
i¦
(x, 0, , 0) = f
(i)
(x).
Deﬁnition 1.6.1. A function f : Z
p
→ Q
p
(or C
p
) is in (
k
if f
i¦
can be
extended as a continuous function on Z
i+1
p
for all i ≤ k.
Remark. If f ∈ (
0
and h
1
, , h
i
= 0, then we have:
v
p
(f
i¦
(x, h
1
, , h
i
)) ≥ v
(
0(f) −
i
¸
j=1
v
p
(h
j
).
20CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Example 1.6.2. The deﬁnition of (
k
here is diﬀerent than the usual case.
Here is an example. For all x in Z
p
, x =
+∞
¸
n=0
p
n
a
n
(x) with a
n
(x) ∈ ¦0, 1, , p−
1¦. Let f(x) =
+∞
¸
n=0
p
2n
a
n
(x), then v
p
(f(x) − f(y)) = 2v
p
(x − y). Hence
f
t
(x) = 0 for all x ∈ Z
p
, thus f is in (
∞
in the usual sense. But f is not (
2
in our case. In fact, let (x, h
1
, h
2
) = (0, p
n
, p
n
) and ((p − 1)p
n
, p
n
, p
n
), here
p = 2, we have:
f
2¦
(0, p
n
, p
n
) = 0;
f
2¦
((p −1)p
n
, p
n
, p
n
) = p −p
2
.
We deﬁne a valuation on (
k
functions by:
v
(
k
(f) = min
0≤i≤k
inf
(x,h
1
, ,h
i
)∈Z
i+1
p
v
p
(f
i¦
(x, h
1
, , h
i
)).
Let L(n, k) = max¦
i
¸
j=1
v
p
(n
j
), i ≤ k,
¸
n
j
= n, n
j
≥ 1¦
Theorem 1.6.3 (Barsky). p
L(n,k)
x
n
is a Banach basis of (
k
.
Exercise. there exists a C
k
, such that for all n ≥ 1,
k
log n
log p
−C
k
≤ L(n, k) ≤ k
log n
log p
.
Corollary 1.6.4. The following three conditions are equivalent:
(i)
+∞
¸
n=0
a
n
x
n
∈ (
k
,
(ii) lim
n→+∞
v
p
(a
n
) −k
log n
log p
= +∞,
(iii) lim
n→+∞
n
k
[a
n
[ = 0.
Deﬁnition 1.6.5. If r ≥ 0, f : Z
p
→Q
p
is in (
r
if
f =
+∞
¸
n=0
a
n
(f)
x
n
and
n
r
[a
n
(f)[ →0 when n →+∞.
(
r
becomes a Banach space with the valuation:
v
(
r (f) = inf
n∈N
¦v
p
(a
n
) −r
log(1 + n)
log p
¦.
1.6. (
K
FUNCTIONS 21
1.6.2 Mahler’s coeﬃcients of (
r
functions.
We need M¨ahler’s Theorem in several variables to prove Barsky’s theorem.
Let g(x
0
, x
1
, , x
i
) be a function deﬁned on Z
i+1
p
. We deﬁne the action
α
[k]
j
on g by the following formula:
α
[1]
j
g(x
0
, , x
i
) = g(x
0
, , x
j
+ 1, , x
i
) −g(x
0
, , x
i
),
α
[k]
j
= α
[1]
j
◦ α
[1]
j
◦ ◦ α
[1]
j
, k times .
We set
a
k
0
, ,k
i
(g) = α
[k
0
]
0
α
[k
i
]
i
g(0, , 0).
Recall that
(
0
(Z
i+1
p
, Q
p
) = (
0
(Z
p
, Q
p
)
¯
¯
(
0
(Z
p
, Q
p
).
Theorem 1.6.6 (M¨ahler). If g is continuous on Z
i+1
p
, then a
k
0
, ,k
i
(g) →0
when (k
0
, , k
i
) →∞ and we have the following identity:
g(x
0
, , x
i
) =
¸
k
0
, ,k
i
∈N
a
k
0
, ,k
i
(g)
x
0
k
0
x
i
k
i
(1.2)
Conversely, if a
k
0
, ,k
i
→0, then the function g via equation (1.2) is contin
uous on Z
i+1
p
, a
k
0
, ,k
i
(g) = a
k
0
, ,k
i
, and
v
(
0(g) = inf v
p
(a
k
0
, ,k
i
).
Proof of Theorem 1.6.3. Let g
T
(x) = (1 +T)
x
, then we have:
g
i¦
T
(x, h
1
, , h
i
) =
1
h
1
h
i
(
¸
I⊂1, ,i¦
(−1)
i−[I[
g
T
(x +
¸
j∈I
h
j
))
= (1 + T)
x
i
¸
j=1
(1 +T)
h
j
−1
h
j
Let P
n
=
x
n
. Since
1
x
x
n
=
1
n
x−1
n−1
and g
i¦
T
(x, h
1
, , h
i
) =
∞
¸
n=0
P
i¦
n
(x, h
1
, , h
i
)T
n
,
we have the following formulas:
P
i¦
n
(x
0
, h
1
, , h
i
) =
¸
n
0
+n
1
++n
i
=n,
n
1
, ,n
i
≥1
1
n
1
n
i
x
0
n
0
h
1
−1
n
1
−1
h
i
−1
n
i
−1
.
22CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Let
Q
n,i
(x
0
, , x
i
) = P
i¦
n
(x
0
, x
1
+ 1, , x
i
+ 1)
=
¸
n
0
+n
1
++n
i
=n,
n
1
, ,n
i
≥1
1
n
1
n
i
x
0
n
0
x
1
n
1
−1
x
i
n
i
−1
.
For all f ∈ (
0
(Z
p
, Q
p
), we have f(x) =
+∞
¸
n=0
a
n
(f)
x
n
. We denote
g
i
(x
0
, , x
i
) =
+∞
¸
n=0
a
n
(f)Q
n,i
(x
0
, , x
i
)
if x
j
+ 1 = 0, j ≥ 1. We have:
a
n
0
,n
1
−1, ,n
i
−1
(g
i
) =
+∞
¸
n=0
a
n
(f)a
n
0
,n
1
−1, ,n
i
−1
(Q
n,i
)
where
a
n
0
,n
1
−1, ,n
i
−1
(Q
n,i
) =
0 if n =
i
¸
j=0
n
j
,
1
n
1
n
i
if n =
i
¸
j=0
n
j
.
If f is in (
k
, i ≤ k, then g
i
is continuous on Z
i+1
p
, thus
a
n
0
+n
1
++n
i
(f)
n
1
n
i
→0.
Conversely, if
a
n
0
+n
1
++n
i
(f)
n
1
n
i
→0, then
+∞
¸
n=0
+∞
¸
n
0
+n
1
++n
i
=n
a
n
0
,n
1
, ,n
i
(f)
n
1
n
i
x
0
n
0
x
1
n
1
−1
x
i
n
i
−1
deﬁnes a continuous functions G
i
on Z
i+1
p
. But G
i
= g
i
on N
i+1
, hence
G
i
= g
i
, x
j
+ 1 = 0, for all j ≥ 1,hence f is in (
k
.
1.7. LOCALLY ANALYTIC FUNCTIONS 23
1.7 locally analytic functions
1.7.1 Analytic functions on a closed disk.
Lemma 1.7.1. Let (a
n
)
n∈N
with a
n
in C
p
be a sequence such that v
p
(a
n
) →∞
when n →∞, let f =
+∞
¸
n=0
a
n
T
n
. Then:
(i) If x
0
∈ O
Cp
, then f
(k)
(x
0
) converges for all k and
lim
n→∞
v
p
(
f
(k)
k!
(x
0
)) = ∞.
(ii) If x
0
, x
1
are in O
Cp
, then
f(x
1
) =
+∞
¸
n=0
f
(n)
(x
0
)
n!
(x
1
−x
0
)
n
and
inf
n∈N
v
p
(
f
(n)
(x
0
)
n!
) = inf
n∈N
v
p
(a
n
);
(iii) inf
n∈N
v
p
(a
n
) = inf
x∈C
Cp
v
p
(f(x)) and v
p
(f(x)) = inf
n
v
p
(a
n
) almost every
where (i.e.,outside a ﬁnite number of x
i
+ m
Cp
).
Proof. (i)
f
(k)
k!
=
+∞
¸
n=0
a
n+k
n+k
k
T
n
. Let T = x
0
; since v
p
(
n+k
k
) ≥ 0, v
p
(x
n
0
) ≥
0, we get (1) and also
v
p
(
f
(k)
(x
0
)
k!
) ≥ inf
n∈N
v
p
(a
n
) = inf
n∈N
v
p
(
f
(n)
(0)
n!
).
(ii)
f(x
1
) =
+∞
¸
n=0
a
n
x
n
1
=
+∞
¸
n=0
a
n
(
+∞
¸
k=0
n
k
(x
1
−x
0
)
k
x
n−k
0
)
=
+∞
¸
k=0
(
+∞
¸
n=0
a
n
n
k
x
n−k
0
)(x
1
−x
0
)
k
=
+∞
¸
n=0
f
(n)
(x
0
)
n!
(x
1
−x
0
)
n
.
So we can exchange the the roles of 0 and x
0
to get
inf
n∈N
v
p
(
f
(n)
(x
0
)
n!
) = inf
n∈N
v
p
(a
n
).
24CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
(iii) That inf
n∈N
v
p
(a
n
) ≤ inf
x∈C
Cp
v
p
(f(x)) is clear. As v
p
(a
n
) goes to +∞,
v
p
(a
n
) reaches its inﬁmum at some n
0
∈ N. So we can divide everything by
a
n
0
and we may assume that inf
n∈N
v
p
(a
n
) = 0. Let f(T) = f(T) mod m
Cp
∈
F
p
[T]. If x ∈ O
Cp
doesn’t reduce mod m
Cp
to a root of f, then f(x) = 0,
equivalently, v
p
(f(x)) = 0.
Corollary 1.7.2. Let f =
+∞
¸
n=0
a
n
T
n
, g =
+∞
¸
n=0
b
n
T
n
, then fg =
+∞
¸
n=0
c
n
T
n
,
where c
n
=
n
¸
i=0
a
i
b
n−i
. Suppose that v
p
(a
n
) and v
p
(b
n
) go to inﬁnity when n
goes to inﬁnity, then v
p
(c
n
) goes to inﬁnity and inf
n
v
p
(c
n
) = inf
n
(a
n
) +inf
n
(b
n
).
Deﬁnition 1.7.3. For x
0
∈ C
p
, r ∈ R, we deﬁne
D(x
0
, r) = ¦x ∈ C
p
, v
p
(x −x
0
) ≥ r¦.
Deﬁnition 1.7.4. A function f : D(x
0
, r) →C
p
is analytic if it is sum of its
Taylor expansion at x
0
or equivalently, if
lim
n→+∞
(v
p
(
f
(n)
(x
0
)
n!
) + nr) = +∞.
We deﬁne v
r¦
x
0
(f) = inf
n
(v
p
(
f
(n)
(x
0
)
n!
) + nr).
Proposition 1.7.5. If the function f : D(x
0
, r) →C
p
is analytic, then
(i) For all k ∈ N, f
(k)
is analytic on D(x
0
, r),
v
r¦
x
0
(
f
(k)
(x
0
)
k!
) + kr ≥ v
r¦
x
0
(f)
and goes to +∞ if k goes to +∞.
(ii) f is the sum of its Taylor expansion at any x ∈ D(x
0
, r).
(iii) v
r¦
x
0
(f) = inf
x∈D(x
0
,r)
v
p
(f(x)).
(iv) v
r¦
x
0
(fg) = v
r¦
x
0
(f) + v
r¦
x
0
(g).
Proof. If r ∈ Q, one can choose α ∈ C
p
, such that v
p
(α) = r. Let F(x) =
f(x
0
+ αx), x ∈ O
Cp
. Apply the previous lemma, we can get the result.
If r / ∈ Q, choose r
n
decreasing with the limit r, r
n
∈ Q. Use D(x
0
, r) =
∪
n
D(x
0
, r
n
) and the case r ∈ Q, we get the result.
1.7. LOCALLY ANALYTIC FUNCTIONS 25
1.7.2 Locally analytic functions on Z
p
.
Deﬁnition 1.7.6. Let h ∈ N be given. The space LA
h
(Z
p
, Q
p
) is the space of
f whose restriction to x
0
+p
h
Z
p
is the restriction of an analytic function f
x
0
on
D(x
0
, h), for all x
0
∈ Z
p
. The valuation of the space is v
LA
h
= inf
x
0
∈S
v
h¦
x
0
(f
x
0
),
S be any set of representations of Z
p
/p
h
Z
p
. (Use above proposition to prove
that this does not depend on S.)
Lemma 1.7.7. LA
h
is a Banach space. Moreover, let
e
n
= 1
i+p
h
Zp
(
x + i
p
h
)
m−1
, n = mp
h
−i, m ≥ 1, 1 ≤ i ≤ p
h
,
then e
n
’s are a Banach basis of LA
h
.
Theorem 1.7.8 (Amice). The functions [
n
p
h
]!
x
n
, n ∈ N are a Banach basis
of LA
h
.
Proof. The idea is to try to relate the g
n
= [
n
p
h
]!
x
n
to the e
n
.
(i) First step: For 1 ≤ j ≤ p
h
, we denote
g
n,j
(x) = g
n
(−j + p
h
x) = [
n
p
h
]!
1
n!
n−1
¸
k=0
(−j −k + p
h
x).
If v
p
(j + k) < h, then v
p
(−j − k + p
h
x) = v
p
(j + k), for all x in O
Cp
. If
v
p
(j + k) ≥ h, then v
p
(−j −k + p
h
x) ≥ h with equality if x / ∈ F
p
⊂ F
p
. So,
we get
v
0¦
0
(g
n,j
) = v
p
([
n
p
h
]!)−v
p
(n!)+
n−1
¸
k=0
inf(v
p
(j+k), h) =
∞
¸
i=1
#¦k : v
p
(k) ≥ i, 1 ≤ k ≤ n¦.
Since v
p
(n!) =
n
¸
k=1
v
p
(k) =
+∞
¸
i=1
[
n
p
i
], we have
v
p
(n!) −v
p
([
n
p
h
]!) =
h
¸
i=1
#¦k : v
p
(k) ≥ i, 1 ≤ k ≤ n¦ =
n
¸
k=1
inf(v
p
(k), h).
Thus,
v
0¦
0
(g
n,j
) =
n
¸
k=1
[inf(v
p
(j + k −1), h) −inf(v
p
(k), h)]
=
h
¸
l=1
([
n + j −1
p
l
] −[
j −1
p
l
] −[
n
p
l
]).
26CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
As [x +y] ≥ [x] +[y], we have v
0¦
0
(g
n,j
) ≥ 0, for all 1 ≤ j ≤ p
h
. So, we have
v
LA
h
(g
n
) ≥ 0.
(ii) Second step: we need a lemma
Lemma 1.7.9. Let n = mp
h
−i, g
n,j
∈ F
p
[x], then:
(i) g
n,j
= 0, if j > i,
(ii) deg g
n,j
= m−1, if j = i,
(iii) deg g
n,j
≤ m−1 if j < i.
The lemma implies the theorem: g
n
can be written in terms of the e
n
,
multiplying by an invertible upper triangular matrix. Now use the fact that
x
n
is a Banach basis if and only if x
n
is a basis of LA
0
h
/pLA
0
h
over F
p
.
Proof of Lemma 1.7.9. (i) If j > i, then j −1 ≥ i. Since
[
n + j −1
p
h
] −[
j −1
p
h
] −[
n
p
h
] = m−(m−1) = 1,
we have v
0¦
0
(g
n,j
) ≥ 1, then g
n,j
= 0.
(ii) and (iii):If j ≤ i, write
g
n,j
(x) =
n
¸
k=0
a
k
x
k
, a
k
∈ Z
p
.
The zeros of g
n,j
are the
j+k
p
h
, 0 ≤ k ≤ n −1 and
#¦zeros in Z
p
¦ = #¦k : v
p
(j + k) ≥ h¦ = [
n + j −1
p
h
] −[
j −1
p
h
] = m−1.
Let ¦α
i
: 1 ≤ i ≤ m−1¦ be the set of the roots with α
1
, , α
m−1
in Z
p
and
α
m
, , α
n
not in Z
p
. Then
g
n,j
= c
m−1
¸
l=1
(x −α
l
)
n
¸
l=m
(1 −α
−1
l
x), (c is a constant ).
Since v
p
(α
−1
l
) > 0 when l ≥ m, then v
p
(a
m−1
) = v
p
(c) = v
0¦
0
(g
n,j
). It implies
c ∈ Z
p
. Hence
g
n,j
= c
m−1
¸
l=1
(x −α
l
).
1.8. DISTRIBUTIONS ON Z
P
27
It remains to prove v
0¦
0
(g
n,i
) = 0. Since
v
0¦
0
(g
n,i
) =
h
¸
l=1
([
mp
h
−1
p
l
] −[
i −1
p
l
] + [
mp
h
−i
p
l
])
and −[
−i
a
] = [
i−1
a
] + 1, we get the result.
Let LA = ¦locally analytic functions on Z
p
¦. Because Z
p
is compact,
LA = ∪LA
h
and is an inductive limit of Banach spaces. So
(i) A function ϕ : LA →B is continuous if and only if ϕ[
LA
h
: LA
h
→B
is continuous for all h.
(ii) A sequence f
n
→f converges in LA if and only if there exists h, such
that for all n, f
n
∈ LA
h
and f
n
→f in LA
h
.
Since
1
n
v
p
([
n
p
h
]!) ∼
1
(p−1)p
h
, we have the following theorem:
Theorem 1.7.10. The function f =
+∞
¸
n=0
a
n
x
n
is in LA if and only if there
exists r > 0, such that v
p
(a
n
) −rn →+∞ when n →+∞.
1.8 Distributions on Z
p
1.8.1 The Amice transform of a distribution.
Deﬁnition 1.8.1. A distribution µ on Z
p
with values in B is a continuous
linear map f →
Zp
fµ from LA to B. We denote the set of distributions
from LA to B by T(Z
p
, B).
Remark. (i) µ[
LA
h
is continuous for all h ∈ N. Set
v
LA
h
(µ) = inf
f∈LA
h
(v
B
(
Zp
fµ) −v
LA
h
(f)).
Then v
LA
h
is a valuation on T(Z
p
, B) for all h, and T(Z
p
, B) is complete for
the Fr´echet topology deﬁned by v
LA
h
, h ∈ N which means that µ
n
goes to µ
if and only if v
LA
h
(µ
n
−µ) →+∞ for all h.
(ii) T(Z
p
, B) = T(Z
p
, Q
p
)
´
¸
B. From now on, we will denote T(Z
p
, Q
p
)
by T.
28CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Let 1
+
be the ring of analytic functions deﬁned on D(0, 0
+
) = ¦x ∈
C
p
, v
p
(x) > 0¦. A function f ∈ 1
+
can be written as f =
+∞
¸
n=0
a
n
T
n
, a
n
∈ Q
p
for all n ∈ N.
Let v
h
=
1
(p−1)p
h
= v
p
(ε −1), where ε is a primitive p
h+1
root of 1.
If F(T) =
+∞
¸
n=0
b
n
T
n
∈ 1
+
, we deﬁne v
(h)
(F) to be
v
(h)
(F) = v
v
h
¦
0
(F) = inf
n∈N
v
p
(b
n
) + nv
h
.
Then, for F, G ∈ 1
+
,
v
(h)
(FG) = v
(h)
(F) + v
(h)
(G).
We put on 1
+
the Fr´echet topology deﬁned by the v
(h)
, h ∈ N.
Deﬁnition 1.8.2. The Amice transform of a distribution µ is the function:
A
µ
(T) =
+∞
¸
n=0
T
n
Zp
x
n
µ =
Zp
(1 +T)
x
µ.
Note that the last identity in the above deﬁnition is only a formal identity
here. However, we have
Lemma 1.8.3. If v
p
(z) > 0, then
Zp
(1 + z)
x
µ = A
µ
(z)
Proof. Choose h such that v
h
< v
p
(z). Then
v
p
(
z
n
[
n
p
h
]!
) →+∞,
therefore
+∞
¸
n=0
z
n
x
n
converges to (1 + z)
x
in LA
h
.
Theorem 1.8.4. The map µ → A
µ
is an isomorphism of Fr´echet spaces
from T to 1
+
. moreover,
v
(h)
(A
µ
) ≥ v
LA
h
(µ) ≥ v
(h+1)
(A
µ
) −1.
1.8. DISTRIBUTIONS ON Z
P
29
Proof. Let A
µ
(T) =
+∞
¸
n=0
b
n
T
n
. Since b
n
=
Zp
x
n
µ and v
p
(n!) ≤
n
p−1
, then we
have:
v
p
(b
n
) = v
p
(b
n
) −v
LA
h
(
x
n
) + v
LA
h
(
x
n
)
≥ v
LA
h
(µ) + v
LA
h
(
x
n
) = v
LA
h
(µ) −v
p
([
n
p
h
]!)
≥ v
LA
h
(µ) −
n
(p −1)p
h
= v
LA
h
(µ) −nv
h
.
Hence A
µ
∈ 1
+
and v
(h)
(A
µ
) ≥ v
LA
h
(µ).
Conversely, for F ∈ 1
+
, F =
+∞
¸
n=0
b
n
T
n
, then for all h,
v
p
([
n
p
h
]!b
n
) = v
p
(b
n
) +
n
(p −1)p
h
→∞.
So f →
+∞
¸
n=0
b
n
a
n
(f) is a continuous map on LA
h
. Denote the left hand side
by
Zp
fµ, this deﬁnes a distribution µ ∈ T. Moreover,
v
LA
h
(µ) = inf
n∈N
v
p
([
n
p
h
]!b
n
) ≥ inf
n∈N
v
p
([
n
p
h+1
]!b
n
)
≥ inf
n∈N
(v
p
(b
n
) +
n
(p −1)p
h+1
) −1 = v
(h+1)
LA
h
(A
µ
) −1.
1.8.2 Examples of distributions.
(i) Measures are distributions and T
0
⊂ T.
(ii) One can multiply a distribution µ ∈ T by g ∈ LA, and one gets
• A
xµ
= ∂A
µ
, ∂ = (1 + T)
d
dT
;
• A
z
x
µ
(T) = A
µ
((1 + T)z −1);
• A
Res
a+p
n
Zp
µ
(T) = p
−n
¸
z
p
n
=1
z
−a
A
µ
((1 + T)z −1)
30CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
(iii) one gets actions ϕ, ψ, Γ with the same formulas than on measures.
(iv) Convolution of distributions: If f ∈ LA
h
and for all y ∈ y
0
+ p
h
Z
p
,
f(x + y) =
+∞
¸
n=0
p
nh
f
(n)
(x + y
0
)
n!
(
y −y
0
p
h
)
n
∈ LA
h
(x)
¯
LA
h
(y),
and v
LA
h
(
p
nh
f
(n)
(x+y
0
)
n!
) goes to +∞, when n →+∞. Hence
Zp
(
Zp
f(x + y)µ(x))λ(y) =
Zp
f λ ∗ µ
is well deﬁned, A
λ∗µ
= A
λ
A
µ
.
(v) The derived distribution: µ →dµ given by
Zp
fdµ =
Zp
f
t
µ. Easy to
check A
dµ
(T) = log(1+T)A
µ
(T). µ can’t be integrated because log(1+T) = 0
if T = ε −1, ε ∈ µ
p
∞.
(vi) Division by x, the Amice transform A
x
−1
µ
of x
−1
µ is a primitive(or
called antiderivative) of (1 + T)
−1
A
µ
, so A
x
−1
µ
is deﬁned up to αδ
0
, α ∈ Q
p
(we have xδ
0
= 0).
1.8.3 Residue at s = 1 of the padic zeta function.
The KubotaLeopoldt distribution µ
KL
given by A
µ
KL
(T) =
log(1+T)
T
. Then
Zp
x
n
µ
KL
=
d
dt
n
t=0
(
Zp
e
tx
µ
KL
) =
d
dt
n
t=0
A
µ
KL
(e
t
−1)
=
d
dt
n
t=0
(
t
e
t
−1
) = (−1)
n
nζ(1 −n), for all n ∈ N.
Since
ψ(
1
T
) =
1
T
and ϕ(log(1 + T)) = p log(1 + T),
we get ψ(µ
KL
) =
1
p
µ
KL
and
Z
∗
p
x
n
µ
KL
= (1 −p
n−1
)
Zp
x
n
µ
KL
= (−1)
n
n(1 −p
n−1
)ζ(1 −n);
ζ
p,i
(s) =
(−1)
i−1
s −1
Z
∗
p
ω(x)
1−i
'x`
1−s
µ
KL
.
The integral is analytic in s by the same argument as for measures.
1.9. TEMPERED DISTRIBUTIONS 31
Proposition 1.8.5. lim
s→1
(s −1)ζ
p,1
(s) =
Z
∗
p
µ
KL
= 1 −
1
p
, ( compare with
lim
s→1
(s −1)ζ(s) = 1).
Proof. It follows from the following lemma.
Lemma 1.8.6.
a+p
n
Zp
µ
KL
= p
−n
, for all n, for all a ∈ Z
p
(almost a Haar
measure but µ ∗ δ
a
= µ).
Proof.
a+p
n
Zp
µ
KL
= p
−n
¸
z
p
n
=1
z
−a
A
µ
KL
(z −1) = p
−n
(1 +
¸
z
p
n
=1,z,=1
log z
z −1
),
and
log z
z−1
= 0, if z
p
n
= 1, z = 1.
1.9 Tempered distributions
1.9.1 Analytic functions inside (
r
functions
Theorem 1.9.1. For all r ≥ 0, LA ⊂ (
r
. Moreover there exists a constant
C(r) depending on r, such that for all h ∈ N and for all f in LA
h
,
v
(
r (f) ≥ v
LA
h
(f) −rh −C(r).
Proof. Since v
LA
h
(f) = inf
n
(v
p
(a
n
(f)) −v
p
([
n
p
h
]!)), we have
v
(
r (f) = inf
n
(v
p
(a
n
(f))−r
log(1 + n)
log p
) ≥ v
LA
h
(f)+inf
n
(v
p
([
n
p
h
]!)−r
log(1 + n)
log p
).
We have a formula for every a:
v
p
(a!) = [
a
p
] + + [
a
p
h
] + ≥
a
p −1
−
log(1 + a)
log p
.
Write n = p
h
a + b, 0 ≤ b ≤ p
h
−1, then we have
v
(
r (f) −v
LA
h
(f) ≥ inf
n
(v
p
([
n
p
h
]!) −r
log(1 + n)
log p
)
= inf
a∈N
0≤b≤p
h
−1
(v
p
(a!) −r
log(ap
h
+ b + 1)
log p
)
≥
a
p −1
−(r + 1)
log(a + 1)
log p
−rh.
32CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
The function −
a
p−1
+ (r + 1)
log(a+1)
log p
of a is bounded above, we just let C(r)
be its maximum.
Observe that the function log is well deﬁned on Z
∗
p
. First if v
p
(x−1) > 0,
let
log x =
+∞
¸
n=1
(−1)
n+1
n
(x −1)
n
;
in general, if x = ω(x)'x`, let log x = log'x`. If x = p, let log p = 0. By the
formula log xy = log x+log y, log is well deﬁned in Q
p
−¦0¦. This log is the
socalled Iwasawa’s log, or log
0
.
However, we can deﬁne the value at p arbitrarily. For L ∈ Q
p
, deﬁne
log
/
p = L, then log
/
x = log
0
x +Lv
p
(x).
Theorem 1.9.2. Choose a L in C
p
. Then there exists a unique log
/
: C
∗
p
→
C
p
satisfying:
(i) log
/
x =
+∞
¸
n=1
(−1)
n−1
n
(x −1)
n
, here v
p
(x −1) > 0,
(ii) log
/
xy = log
/
x + log
/
y,
(iii) log
/
= L.
Proposition 1.9.3. If r ≥ 0, j > r, then x
j
log
/
x ∈ (
r
.
Proof. We have
x
j
log
/
x =
+∞
¸
n=0
p−1
¸
a=1
1
p
n
a+p
n+1
Zp
x
j
log
/
x.
Let f
n,a
= 1
p
n
a+p
n+1
Zp
x
j
log
/
x. We have to prove the sum converges in (
r
.
On p
n
a + p
n+1
Z
p
, we have
x
j
log
/
x = (x −p
n
a + p
n
a)
j
log
/
(p
n
a + (x −p
n
a))
= p
nj
(a + p
x −p
n
a
p
n+1
)
j
(log
/
p
n
a + log
0
(1 + p
x −p
n
a
p
n+1
a
)).
So f
n,a
∈ LA
n+1
, v
LA
n+1
(f
n,a
) ≥ nj. Use the previous theorem, we get
v
(
r (f
n,a
) ≥ nj −r(n + 1) −C(r) and it goes to +∞.
1.9. TEMPERED DISTRIBUTIONS 33
1.9.2 Distributions of order r
Deﬁnition 1.9.4. Let r ≥ 0 and B be a Banach space. A distribution
µ ∈ T(Z
p
, B) is a distribution of order r if f →
Zp
fµ is a continuous
map from (
r
(Z
p
, Q
p
) to B. We denote the set of distributions of order r by
T
r
(Z
p
, B). We deﬁne a valuation on T
r
(Z
p
, B) by
v
t
Tr
(µ) = inf
f∈(
r
(v
p
(
Zp
fµ) −v
(
r (f)).
Remark. (i) Under the above valuation, T
r
(Z
p
, B) is a padic Banach space
and T
r
(Z
p
, B) = T
r
(Z
p
, Q
p
)
´
¸
B. We denote T
r
(Z
p
, Q
p
) by T
r
.
(ii) T
temp
= ∪T
r
=set of tempered distributions.
(iii) Since LA
h
⊂ (
r
, and for f ∈ LA
h
, v
(
r (f) ≥ v
LA
h
(f) −rh −C(r), we
get, for µ ∈ T
r
⊂ LA
∗
h
,
v
LA
∗
h
(µ) = inf
f∈LA
h
(v
p
(
Zp
fµ) −v
LA
h
(f)) ≥ v
t
Tr
(µ) −rh −C(r).
Theorem 1.9.5. µ ∈ T, the following are equivalent: (i) µ ∈ T
r
i.e. µ can
be extended by continuity to (
r
.
(ii) There exists a constant C, such that v
p
(
Zp
x
n
µ) ≥ C −r
log(1+n)
log p
, for
all n.
(iii) There exists a constant C, such that v
p
(
a+p
h
Zp
(x−a)
j
µ) ≥ C+h(j −
r), for all a ∈ Z
p
, j ∈ N, h ∈ N.
(iv) There exists a constant C, such that v
LA
h
(µ) ≥ C−rh, for all h ∈ N.
Remark. It follows that
v
Tr
(µ) = inf
a∈Zp
j∈N,n∈N
(v
p
(
a+p
h
Zp
(x −a)
j
µ) −h(j −r))
is equivalent to v
t
Tr
.
Proof. (i) ⇔ (ii) is just the deﬁnition of v
t
Tr
. (iii) ⇔ (iv) is true by the
deﬁnition of LA
h
(with some C). Remains to prove (ii) ⇔ (iv). We have
v
(h)
(A
µ
) ≥ v
LA
h
(µ) ≥ v
(h+1)
(A
µ
) −1, hence the proof is reduce to the follow
ing lemma with F = A
µ
.
34CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Lemma 1.9.6. Suppose F ∈ 1
+
, F =
+∞
¸
n=0
b
n
T
n
, the following are equivalent:
(i) there exists C, such that v
(h)
(F) ≥ C −rh, for all h ∈ N,
(ii) there exists C
t
, such that v
p
(b
n
) ≥ C
−r
log(1+n)
log p
for all n.
Proof. Let
C
0
= inf
h∈N
(v
(h)
(F) + rh) = inf
h∈N
(inf
n∈N
(v
p
(b
n
) +
n
(p −1)p
h
) + rh),
C
1
= inf
n∈N
(v
p
(b
n
) + r
log(1 + n)
log p
).
Let h = [
log(1+n)
log p
], then
v
p
(b
n
) ≥ C
0
−rh −
n
(p −1)p
h
≥ C
0
−r
log(1 + n)
log p
−2,
which implies C
1
≥ C
0
−2.
Now, if h is ﬁxed, then C
1
− r
log(1+n)
log p
+
n
(p−1)p
h
is minimal for (1 + n) =
(p −1)p
h
r. Hence,
C
1
−r
log(1 + n)
log p
+
n
(p −1)p
h
≥ C
1
−rh −
log(p −1)r
log p
.
Thus, C
0
≥ C
1
−r
log(p−1)r
log p
.
For N ≥ 0, let LP
[0,N]
be the set of the locally polynomial functions of
degree no more than N on Z
p
.
Theorem 1.9.7. Suppose r ≥ 0, N > r −1. If f →
Zp
fµ is linear function
from LP
[0,N]
to a Banach space B , such that there exists C,
v
p
(
a+p
n
Zp
(x −a)
j
µ) ≥ C + (j −r)n
for all a ∈ Z
p
and n, j ∈ N, then µ extends uniquely to an element of T
r
.
Remark. (i) Let r = 0, N = 0, we recover the construction of measures as
bounded additive functions on open compact sets.
1.9. TEMPERED DISTRIBUTIONS 35
(ii) We deﬁne a new valuation on T
r
v
T
r,N
(µ) = inf
a∈Zpn∈N,j∈N
v
p
(
a+p
n
Zp
(x −a)
j
µ) −n(j −r),
then v
p
(
Zp
fµ) ≥ v
LA
h
(f) + v
T
r,N
(µ) −rn for all f ∈ LP
[0,N]
∩ LA
h
;
(iii) The open mapping theorem in Banach spaces implies that v
T
r,N
is
equivalent to v
Tr
.
Proposition 1.9.8. If f ∈ LA, r ≥ 0, N > r −1, put
f
n
=
p
n
−1
¸
i=0
1
i+p
n
Zp
(
N
¸
k=0
f
(k)
(i)
k!
(x −i)
k
) ∈ LP
[0,N]
,
then f
n
→f in (
r
. Hence LP
[0,N]
is dense in (
r
.
Proof. There exists h, such that f ∈ LA
h
. We assume n ≥ h, then
v
LA
h
(f −f
n
) = inf
0≤i≤p
n
−1
inf
k≥N+1
v
p
(p
nk
f
(k)
(i)
k!
).
f ∈ LA
h
implies v
p
(
p
hk
f
(h)
(i)
h!
) ≥ v
LA
h
(f). Hence
v
LA
h
(f −f
n
) ≥ v
LA
h
(f) + (N + 1)(n −h).
Then
v
(
r (f −f
n
) ≥ v
LA
h
(f −f
n
) −rn −C(r)
≥ v
LA
h
(f) −C(r) −(N + 1)h + (N + 1 −r)n →+∞,
because N + 1 −r > 0.
Proof of Theorem 1.9.7. The proposition implies the uniqueness in the
theorem. We only need to prove the existence.
We show that if f ∈ LA
h
, then lim
n→∞
Zp
f
n
µ exists:
v
p
(
Zp
(f
n+1
−f
n
)µ) ≥ v
LA
n+1
(f
n
−f
n+1
) + v
T
r,N
(µ) −r(n + 1)
≥ inf(v
LA
n+1
(f −f
n
), v
LA
n+1
(f −f
n+1
)) + v
T
r,N
(µ) −r(n + 1)
≥ v
T
r,N
(µ) + v
LA
h
(f) −r(h −1) + (n −h)(N + 1 −r) →+∞.
36CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Set
Zp
fµ = lim
n→+∞
Zp
f
n
µ, then
v
p
(
Zp
fµ) ≥ inf(v
p
(
Zp
f
n
µ), v
p
(inf
n≥h
Zp
(f
n−1
−f
n
)µ))
≥ v
LA
h
(f) −rh + (v
T
r,N
(µ) −r).
This implies that µ ∈ T
r
.
1.10 Summary
To summarize what we established:
(i) We have the inclusions:
(
0
⊃ (
r
⊃ LA ⊃ LA
h
T
0
⊂ T
r
⊂ T ⊂ LA
∗
h
.
Now, if f is a function on Z
p
and µ is a linear form on polynomials, then we
have:
f →a
n
(f) =
n
¸
i=0
(−1)
i
n
i
f(n −i)
µ →b
n
(µ) =
Zp
x
n
µ
(ii) For f a function,
• f ∈ (
0
if only if v
p
(a
n
(f)) →+∞ and
v
(
0(f) = inf
x∈Zp
v
p
(f(x)) = inf
n
v
p
(a
n
(f)).
• f ∈ (
r
if only if v
p
(a
n
(f)) −r
log(1+n)
log p
→+∞ and
v
(
r (f) = inf
n
v
p
(a
n
(f) −r
log(1 + n)
log p
).
• f ∈ LA if only if there exists r > 0 such that v
p
(a
n
(f)) − rn → +∞.
LA is not a Banach space; it is a compact inductive limit of Banach
spaces.
1.10. SUMMARY 37
• f ∈ LA
h
if and only if v
p
(a
n
(f)) −v
p
([
n
p
h
]!) →+∞ and
v
LA
h
(f) = inf
x∈Zp
inf
k∈N
v
p
(
p
kh
f
(k)
(x)
h!
) = inf
n
(v
p
(a
n
(f)) −v
p
([
n
p
h
]!)).
(iii) For µ a distribution,
• µ ∈ T
0
if and only if v
T
0
(µ) = inf
n
v
p
(b
n
(µ)) > −∞.
• µ ∈ T
r
if and only if v
t
Tr
(µ) = inf
n
v
p
(b
n
(µ)) + r
log(1+n)
log p
> −∞.
• µ ∈ T if and only if for all r > 0, inf
n
v
p
(b
n
(µ)) + rn > −∞.
(iv) f =
+∞
¸
n=0
a
n
(f)
x
n
and
Zp
fµ =
+∞
¸
n=0
a
n
(f)b
n
(µ).
38CHAPTER 1. THE PADIC ZETAFUNCTIONOF KUBOTALEOPOLDT
Chapter 2
Modular forms
2.1 Generalities
2.1.1 The upper halfplane
By SL
2
we mean the group of 2 2 matrices with determinant 1. We write
SL
2
(A) for those elements of SL
2
with entries in a ring A. In practice, the
ring A will be Z, Q, R.
Let γ =
a b
c d
in SL
2
(R), z in C −¦−
d
c
¦, let γz =
az+b
cz+d
, then
Im(γz) =
(ad −bc)
[cz + d[
2
Im(z) =
Imz
[cz + d[
2
.
We denote H = ¦z, Imz > 0¦ the upper half plane. It is stable under z →γz
and one can verify (γ
1
γ
2
)z = γ
1
(γ
2
z).
Proposition 2.1.1. The transform action z →γz deﬁnes a group action of
SL
2
(R) on H.
Proposition 2.1.2.
dx∧dy
y
2
is invariant under SL
2
(R).
(hint : dx ∧ dy =
i
2
dz ∧ dz and z →γz is holomorphic.)
Deﬁnition 2.1.3. Let f : H → C be a meromorphic function and γ =
a b
c d
be in SL
2
(R). If k in Z, we deﬁne the weight k action of SL
2
(R) by
(f[
k
γ)(z) = (cz + d)
−k
f(γz).
Exercise. (f[
k
γ
1
)[
k
γ
2
= f[
k
γ
1
γ
2
.
39
40 CHAPTER 2. MODULAR FORMS
2.1.2 Deﬁnition of modular forms
Deﬁnition 2.1.4. Let Γ be a subgroup of SL
2
(Z) of ﬁnite index, χ is a
ﬁnite order character of Γ (i.e. χ(Γ) ⊂ µ
N
). f : H →C is a modular form of
weight k, character χ for Γ, if:
(i) f is holomorphic on H;
(ii) f[
k
γ = χ(γ)f, if γ ∈ Γ;
(iii) f is slowly increasing at inﬁnity, i.e. for all γ ∈ Γ`SL
2
(Z), there
exists C(γ) and r(γ) such that [ f[
k
γ(z) [≤ y
r(γ)
, if y ≥ C(γ).
Deﬁnition 2.1.5. Γ is a congruence subgroup if Γ ⊃ Γ(N) = Ker (SL
2
(Z) →
SL
2
(Z/NZ)) for some N in N.
Example 2.1.6.
Γ
0
(N) = ¦
a b
c d
∈ SL
2
(Z) : c ≡ 0 mod N¦ ⊃ Γ(N).
Any character χ : (Z/NZ)
∗
→C
∗
extends to a congruence character
χ : Γ
0
(N) →C
∗
χ(
a b
c d
) →χ(d).
Let M
k
(Γ, χ) be the set of modular forms of weight k, character χ for Γ.
Then M
k
(Γ, χ) is a Cvector space.
Remark. (i) If
−1 0
0 −1
= −I ∈ Γ and χ(−I) = (−1)
k
, then M
k
(Γ, χ) = 0;
(ii) f ∈ M
k
(Γ, χ), g ∈ SL
2
(Z), f[
k
g ∈ M
k
(g
−1
Γg, χ
g
) where χ
g
(γ) =
χ(gγg
−1
).
2.1.3 qexpansion of modular forms.
Lemma 2.1.7. If Γ is a subgroup of ﬁnite index of SL
2
(Z) and χ : Γ →C
∗
is of ﬁnite order, then there exists M in N − ¦0¦, such that
1 M
0 1
∈ Γ and
χ(
1 M
0 1
) = 1.
Proof. We can replace Γ by Ker χ and assume χ = 1. There exists n
1
= n
2
,
such that
1 n
1
0 1
and
1 n
2
0 1
have the same image in Γ`SL
2
(Z), then M =[
n
1
−n
2
[ satisfy the condition.
For M ∈ N−¦0¦, let q
M
(z) = e
2πiz
M
. Then z →q
M
(z) gives a holomorphic
bijection MZ`H · T
∗
= ¦0 <[ q
M
[< 1¦.
2.1. GENERALITIES 41
Corollary 2.1.8. If f ∈ M
k
(Γ, χ), then there exists M = 0, M ∈ N, such
that f(z + M) = f(z). Thus there exists
˜
f holomorphic on T
∗
, such that
f(z) =
˜
f(q
M
).
Now
˜
f has a Laurent expansion
˜
f(q
M
) =
¸
n∈Z
a
n
q
n
M
with
a
n
= e
2πny
M
1
M
M
2
−
M
2
f(x + iy)e
−2πinx
M
dx
for all y. If n < 0, when y → ∞, the right hand side goes to 0, so a
n
= 0.
Hence we get the following result.
Proposition 2.1.9. If f is in M
k
(Γ, χ), there exists M ∈ N − ¦0¦, and
elements a
n
(f) for each n ∈
1
M
N, such that
f =
¸
n∈
1
M
N
a
n
(f)q
n
, where q(z) = e
2πiz
,
which is called the q expansion of modular forms.
2.1.4 Cusp forms.
Deﬁnition 2.1.10. (i) v
∞
(f) = inf¦n ∈ Q, a
n
(f) = 0¦ ≥ 0 and we say
that f has a zero of order v
∞
(f) at ∞. We say that f has a zero at ∞ if
v
∞
(f) > 0.
(ii) A modular form f is a cusp form if f[
k
γ has a zero at ∞ for all γ
in Γ`SL
2
(Z). We denote S
k
the set of cusp form of weight k. S
k
(Γ, χ) ⊂
M
k
(Γ, χ).
Remark. If f is a cusp form, then f is rapidly decreasing at ∞ since
[ (f[
k
γ)(z) [= O
e
−v∞(f[
k
γ)2πy
.
Theorem 2.1.11. S
k
(Γ, χ) and M
k
(Γ, χ) are ﬁnite dimensional Cvector
spaces with explicit formulas for the dimensions( if k ≥ 2).
Remark. ⊕
k,χ
M
k
(Γ, χ) = M(Γ) is an algebra.
The study of M
k
(Γ, χ) for congruence subgroup and congruence charac
ters (Ker χ congruence subgroup ) can be reduced to the study of M
k
(Γ
0
(N), χ)
for a simple group theoretic reason. From now on, we write
M
k
(N, χ) = M
k
(Γ
0
(N), χ), S
k
(N, χ) = S
k
(Γ
0
(N), χ).
42 CHAPTER 2. MODULAR FORMS
2.2 The case Γ = SL
2
(Z)
2.2.1 The generators S and T of SL
2
(Z).
Let M
k
(1) = M
k
(SL
2
(Z), 1), S
k
(1) = S
k
(SL
2
(Z), 1). Let
S =
0 −1
1 0
, T =
1 1
0 1
.
It is easy to verify
T
n
=
1 n
0 1
for any n ∈ Z.
So Sz = −
1
z
, T
n
z = z + n.
Proposition 2.2.1. (i) If (a, b) = 1, then there exists n = n(a, b), (a
0
, b
0
) =
(1, 0), (a
1
, b
1
) = (0, 1), (a
n
, b
n
) = (a, b), such that
a
l
a
l+1
b
l
b
l+1
∈ SL
2
(Z) for any l.
(ii) SL
2
(Z) = ' S, T` .
Proof. (i) We prove it by induction on [a[ +[b[.
If [a[ +[b[ = 1, one can do it by hand:
I =
1 0
0 1
, S =
0 −1
1 0
, S
2
=
−1 0
0 −1
, S
3
=
0 1
−1 0
.
If [a[ +[b[ ≥ 2, there exists µ, ν ∈ Z, such that bµ−aν = 1, and [ν[ < [b[,
which implies [µ[ ≤ [a[. Then we have
µ a
ν b
∈ SL
2
(Z) and [µ[ +[ν[ < [a[ +[b[.
Therefore the conclusion is obtained by the inductive assumption.
(ii) Let γ =
a b
c d
∈ SL
2
(Z), there exists n = n(a, b), (a
0
, b
0
) = (1, 0),
(a
1
, b
1
) = (0, 1), (a
n
, b
n
) = (a, b), such that
γ
l
=
a
l
a
l+1
b
l
b
l+1
∈ SL
2
(Z) for any l.
As γ
1
= I and
γ
−1
l+1
γ
l
=
n
l
1
−1 0
= T
−n
l
S
3
,
then γ =
¸
(γ
−1
l+1
γ
l
)
−1
∈ ' T, S `.
2.2. THE CASE Γ = SL
2
(Z) 43
Corollary 2.2.2. Let f =
+∞
¸
n=0
a
n
q
n
, where q = e
2πiz
, then f ∈ M
k
(1) if and
only if the following two conditions hold:
(i)
+∞
¸
n=0
a
n
q
n
converges if [q[ < 1.
(ii) f(−
1
z
) = z
k
f(z).
2.2.2 Eisenstein series
Proposition 2.2.3. If k ≥ 3, then G
k
∈ M
k
(1), where
G
k
(z) =
1
2
Γ(k)
(−2πi)
k
¸
m,n
t
1
(mz + n)
k
∈ M
k
(1),
and
¸
t
means the summation runs over all pairs of integers (m, n) distinct
from (0, 0).
Proof. As [mz + n[ ≥ min(y, y/[z[) sup([m[, [n[), the series converges uni
formly on compact subsets of H and is bounded at ∞.
Let γ =
a b
c d
∈ SL
2
(Z), since
(cz + d)
−k
¸
m,n
t
1
(m
az+b
cz+d
+ n)
k
=
¸
m,n
t
1
((am + cn)z + (bm + dn))
k
,
and
(m, n) →(am + cn, bm + dn)
is a bijection of Z
2
−¦(0, 0)¦, it follows that G
k
[
k
γ = G
k
.
Proposition 2.2.4.
G
k
(z) =
Γ(k)
(−2πi)
k
ζ(k) +
+∞
¸
n=1
σ
k−1
(n)q
n
,
where σ
s
(n) =
¸
d[n, d≥1
d
s
, and k is even (if k is odd, M
k
(1) = 0, since −I ∈
SL
2
(Z)).
Proof.
G
k
(z) =
Γ(k)
(−2πi)
k
ζ(k) +
Γ(k)
(−2πi)
k
+∞
¸
m=1
A
k
(mz),
44 CHAPTER 2. MODULAR FORMS
where
A
k
(z) =
¸
n∈Z
1
(z + n)
k
=
¸
l∈Z
ˆ
φ(l)q
l
for the last identity given by the Poisson summation formula of Fourier trans
forms, and (by residue computation)
ˆ
φ(l) =
+∞
−∞
e
−2πilx
(x + iy)
k
dx =
0, if l ≤ 0,
(−2πi)
k
(k−1)!
l
k−1
, if l ≥ 0.
It follows that
G
k
(z) =
Γ(k)
(−2πi)
k
ζ(k) +
+∞
¸
m=1
+∞
¸
l=1
l
k−1
q
lm
=
Γ(k)
(−2πi)
k
ζ(k) +
+∞
¸
n=1
σ
k−1
(n)q
n
.
Remark. (i) G
2
(z) =
Γ(2)
(−2πi)
2
ζ(2) +
¸
+∞
n=1
σ
1
(n)q
n
is not a modular form, but
it is almost one. Let
G
∗
2
(z) = G
2
(z) +
1
8πy
=
1
2
Γ(2)
(−2πi)
2
lim
s→0
¸
m,n
t
1
(mz + n)
2
y
s
[mz + n[
2s
,
G
∗
2
is not holomorphic, but G
∗
2
[
2
γ = G
∗
2
, for any γ ∈ SL
2
(Z).
(ii) Let E
k
=
G
k
a
0
(G
k
)
, so that a
0
(E
k
) = 1.
2.2.3 The fundamental domain for SL
2
(Z)
Theorem 2.2.5. Let D denotes the shadows in Figure 1.1. Then it is a
fundamental domain for PSL
2
(Z). Moreover, the stabilizer of z ∈ D is
 ¦I¦ if z = i, ρ;
 ¦I, S¦ if z = i;
 ¦I, TS, (TS)
2
¦ if z = ρ.
2.2. THE CASE Γ = SL
2
(Z) 45
Figure 2.1: The Fundamental Domain.
Proof. Let z
0
∈ H,
γ =
a b
c d
∈ SL
2
(Z)
Since Im(γz
0
) =
z
0
[cz
0
+d[
2
tends to zero, as (c, d) tends to inﬁnity, there exists
γ
0
such that Im(γ
0
z
0
) is maximal. There exists a unique n such that:
−
1
2
< Re (γ
0
z
0
) + n ≤
1
2
.
Let γ
1
= T
n
γ
0
, then
Im(γ
1
z
0
) = Im(γ
0
z
0
) ≥ Im(Sγ
1
z
0
) =
Im(γ
1
z
0
)
[γ
1
z
0
[
2
which implies [γ
1
z
0
[ ≥ 1. Therefore D contains a fundamental domain.
If z
1
, z
2
∈ D, and there exists γ ∈ SL
2
(Z), such that z
1
= γz
2
, we
want to show z
1
= z
2
. By symmetry, we may assume Im(z
2
) ≥ Im(z
1
). If
γ =
a b
c d
, Im(z
2
) ≥
Im(z
2
)
[cz
2
+d[
2
implies [cz
2
+d[
2
≤ 1. As Im(z
2
) ≥
√
3
2
, we have
c ≤ 1, d ≤ 1. It remains only ﬁnite number of cases to check.
If c = 0, then d = ±1, and γ is the translation by ±b. Since
−
1
2
< Re (z
1
), Re (z
2
) ≤
1
2
,
this implies b = 0, and γ = ±I.
If c = 1, the fact [z
2
+ d[ ≤ 1 implies d = 0 except if z
2
= ρ, in which
case we can have d = 0, −1. The case d = 0 gives [z
2
[ ≤ 1, hence [z
2
[ = 1; on
the other hand, γ ∈ SL
2
(Z) implies b = −1, hence z
1
= γz
2
= a −1/z
2
∈ D,
46 CHAPTER 2. MODULAR FORMS
Figure 2.2: The Route C(M, ε) of Integration.
which implies a = 0, and z
1
= z
2
= i. The case z
2
= ρ, and d = −1 gives
a +b + 1 = 0 and z
1
= γz
2
= a −
1
ρ−1
= a +ρ ∈ D, which implies a = 0 and
z
1
= z
2
= ρ.
If c = −1, we have similar argument as c = 1.
This completes the proof of the Theorem.
2.2.4 The
k
12
formula.
The following proposition is usually called “the
k
12
formula”.
Proposition 2.2.6. Let f ∈ M
k
−¦0¦, then
v
∞
(f) +
1
2
v
i
(f) +
1
3
v
ρ
(f) +
¸
z∈D−i,ρ¦
v
z
(f) =
k
12
.
Proof. Apply Cauchy residue formula to d log f over the path showed in
Figure 1.2. As M →+∞, and ε →0, we have:
1
2πi
C(M,ε)
d log f =
¸
z∈D−i,ρ¦
v
z
(f),
lim
M→+∞
1
2πi
C∞(M)
d log f = lim
M→+∞
−
1
2πi
[z[=e
−2πM
d log
¸
a
n
(f)z
n
= −v
∞
(f),
lim
ε→0
1
2πi
C(i,ε)
d log f = −
1
2
v
i
(f),
2.2. THE CASE Γ = SL
2
(Z) 47
lim
ε→0
1
2πi
C(ρ,ε)
d log f = −
1
6
v
ρ
(f) = −
1
6
v
ρ
2(f) = lim
ε→0
1
2πi
C(ρ
2
,ε)
d log f
1
2πi
(
i
ρ
2
d log f +
ρ
i
d log f) =
1
2πi
i
ρ
2
(d log f −d log f(−
1
z
))
= −
1
2πi
i
ρ
2
(d log f −d log z
k
f(z))
= −
k
2πi
i
ρ
2
d z
z
= −
k
2πi
(log i −log ρ
2
) =
k
12
.
Putting all these equations together, we get the required formula.
Corollary 2.2.7. G
4
has its only zero on D at z = ρ, G
6
has its only zero
on D at z = i.
∆ = ((
G
4
a
0
(G
4
)
)
3
−(
G
6
a
0
(G
6
)
)
2
)
1
3a
−1
0
(G
4
) −2a
−1
0
(G
6
)
= q + ∈ M
12
(1)
does not vanish on D (v
∞
(∆) = 1).
Remark. One can prove ∆ = q
+∞
¸
n=1
(1 −q
n
)
24
.
2.2.5 Dimension of spaces of modular forms.
Theorem 2.2.8. (i) M
k
(1) = 0, if k is odd or k = 2.
(ii) dimM
k
(1) = 1, if k = 0 or k is even and 2 < k ≤ 10. In this case
M
k
(1) = C G
k
(We have G
0
= 1).
(iii) M
k+12
(1) = C G
k+12
⊕∆ M
k
(1).
Proof. If f ∈ M
k+12
, then
f =
a
0
(f)
a
0
(G
k+12
)
G
k+12
+ ∆g,
where g ∈ M
k
(1), because ∆ does not vanish on H, v
∞
(∆) = 1 and v
∞
(f −
a
0
(f)
a
0
(G
k+12
)
G
k+12
) ≥ 1.
48 CHAPTER 2. MODULAR FORMS
Corollary 2.2.9. If k is even, dim
C
M
k
(1) =
[
k
12
], k ≡ 2 mod 12,
[
k
12
] + 1, if not.
Remark. Finite dimensionality of spaces of modular forms has many com
binatorical applications. For example, let
θ(z) =
¸
n∈Z
q
n
2
2
=
¸
n∈Z
e
πin
2
z
,
Γ
θ
= ¦γ ∈ SL
2
(Z), γ ≡ I or γ ≡ S mod 2¦,
χ
θ
: Γ
θ
→¦±1¦. χ
θ
(γ) =
1 if γ ≡ I
−1 if γ ≡ S
One can check that dimM
2
(Γ
θ
, χ
θ
) ≤ 1, θ
4
∈ M
2
(Γ
θ
, χ
θ
), and 4G
∗
2
(2z) −
G
∗
2
(
z
2
) ∈ M
2
(Γ
θ
, χ
θ
), so we have
4G
∗
2
(2z) −G
∗
2
(
z
2
) =
3ζ(2)Γ(2)
(−2πi)
2
θ
4
,
hence
[¦(a, b, c, d) ∈ Z
4
: a
2
+ b
2
+ c
2
+ d
2
= n¦[ = 8
¸
d[n,4d
d,
from which we can deduce that any positive integer can be written as a sum
of 4 squares.
2.2.6 Rationality results.
As M
8
(1) and M
10
(1) are of dimension 1, we have
a
0
(G
8
)G
2
4
= a
0
(G
4
)
2
G
8
, a
0
(G
10
)G
4
G
6
= a
0
(G
4
)a
0
(G
6
)G
10
. (∗)
Let
α =
Γ(4)
(−2πi)
4
ζ(4), β =
Γ(8)
(−2πi)
8
ζ(8).
Substituting
G
4
= α + q + 9q
2
+ , G
8
= β + q + 129q
2
+
2.2. THE CASE Γ = SL
2
(Z) 49
in (∗), compare the coeﬃcients of q and q
2
, we have the following equations:
2αβ = α
2
β(1 + 18α) = 129α
2
The solution is: α =
1
240
, β =
1
480
. In particular, α, β ∈ Q, which implies G
4
and G
8
have rational qexpansions, and
ζ(4)
π
4
∈ Q,
ζ(8)
π
8
∈ Q.
Exercise. a
0
(G
6
) = −
1
504
, which implies
ζ(6)
π
6
∈ Q.
Let A be a subring of C, let
M
k
(Γ, A) = ¦f ∈ M
k
(Γ), a
n
(f) ∈ A, for all n¦,
then M(Γ, A) =
¸
k
M
k
(Γ, A) is an Aalgebra.
Theorem 2.2.10. (i) M(SL
2
(Z), Q)
∼
−→Q[X, Y ], where X = G
4
, Y = G
6
.
(ii) M(SL
2
(Z), C) = C ⊗M(SL
2
(Z), Q).
Proof. If
¸
k
f
k
= 0, where f
k
∈ M
k
(SL
2
(Z), C), then for any z, for any
a b
c d
∈ SL
2
(Z), we have
¸
k
(cz +d)
k
f
k
(z) = 0. Therefore
¸
k
(Xz +Y )
k
f
k
(z)
is identically zero because it (as a polynomial in X and Y ) has too many
zeros. Hence f
k
(z) = 0, which implies that
M(SL
2
(Z), C) =
k
M
k
(SL
2
(Z), C).
Now if k = 12n, G
3n
4
, G
3(n−1)
4
∆, , ∆
n
is a basis of M
k
(1); if k = 12n+2,
G
3(n−1)+2
4
G
6
, G
3(n−2)+2
4
G
6
∆, , G
2
4
G
6
∆
n−1
is a basis of M
k
(1), and so on,
∆ = aG
3
4
+ bG
2
6
, a, b ∈ Q. As G
4
, G
6
∈ M(SL
2
(Z), Q), this proves both
results.
Corollary 2.2.11. Let f ∈ M
k
(1), σ ∈ Aut(C), then f
σ
=
¸
a
n
(f)
σ
q
n
∈
M
k
(1). Moreover,
ζ(k)
(−2πi)
k
∈ Q if k is even and k ≥ 4.
Proof. The ﬁrst assertion is a direct consequence of Theorem 2.2.10 (ii). For
any σ ∈ Aut(C), we have
G
σ
k
−G
k
= a
0
(G
k
)
σ
−a
0
(G
k
) ∈ M
k
(1).
This implies a
0
(G
k
)
σ
= a
0
(G
k
) for any σ ∈ Aut(C), therefore a
0
(G
k
) ∈
Q.
50 CHAPTER 2. MODULAR FORMS
Remark. When k = 2, we can use
4G
∗
2
(2z) −G
∗
2
(
z
2
) ∈ M
2
(Γ
θ
, Q)
to deduce
ζ(2)
π
2
∈ Q.
Remark. (i) The zeta function ζ is a special case of Lfunctions, and ζ(k)
are special values of Lfunctions (i.e. values of Lfunctions at integers).
Siegel used the above method to prove rationality of special values of
Lfunctions for totally real ﬁelds.
(ii) With a lot of extra work, we can prove integrality results. As
G
k
(z) =
Γ(k)
(−2πi)
k
ζ(k) +
+∞
¸
n=1
σ
k−1
(n)q
n
,
and σ
k−1
(n) =
Zp
x
k−1
(
¸
d[n
δ
d
), we have all a
n
(G
k
) are given by measures on
Z
p
, therefore a
0
(G
k
) is also given by measures. From which we can deduce
other constructions of KubotaLeopoldt zeta functions (the work of Serre,
Deligne, Ribet).
2.3 The algebra of all modular forms.
Let A be a subring of C, let
´
k
(A) =
¸
[SL
2
(Z):Γ]<+∞
M
k
(Γ, A) =
¸
a
n
q
n
∈ M
k
(Γ, C), a
n
∈ A, n ∈ N
¸
.
Let ´(A) = ⊕´
k
(A), then it is an Aalgebra. Let
´
cong
(A) =
¸
Γ congruence subgroup
M(Γ, A).
Theorem 2.3.1. (i) If f ∈ ´(C), and σ ∈ Aut(C), then f
σ
∈ ´(C).
(ii) ´(C) = C ⊗
Q
´(Q) = C ⊗
Q
´(Q).
(iii) Let Π
Q
= Aut(´(Q)/ M(SL
2
(Z), Q)), G
Q
= Gal(Q/Q), then we
have an exact sequence:
1
SL
2
(Z)
∧
Π
Q
G
Q
1
2.3. THE ALGEBRA OF ALL MODULAR FORMS. 51
where G
∧
lim
←−
[G:Γ]<∞
Γ normal
(G/Γ), and G
Q
→Π
Q
is induced by the action on Fourier
coeﬃcients.
(iv) ´
cong
(Q
ab
) is stable by Π
Q
, and
Aut(´
cong
(Q
ab
)/ M(SL
2
(Z), Q))
∼
−→GL
2
(
ˆ
Z).
Moreover, we have the following commutative diagram:
1
SL
2
(Z)
∧
Π
Q
G
Q
1
1
SL
2
(
ˆ
Z)
GL
2
(
ˆ
Z)
ˆ
Z
∗
.
1
where G
Q
→
ˆ
Z
∗
is the cyclotomic character, GL
2
(
ˆ
Z) →
ˆ
Z
∗
is the determinant
map, and
ˆ
Z
∗
→GL
2
(
ˆ
Z) maps u to
1 0
0 u
.
Remark. (i) SL
2
(Z)
∧
is much bigger than SL
2
(
ˆ
Z).
(ii) We can get an action of G
Q
on SL
2
(Z)
∧
by inner conjugation in
Π
Q
. This is a powerful way to study G
Q
(Grothendieck, “esquisse d’un pro
gramme”).
(iii) There are padic representations of G
Q
attached to modular forms
(by Deligne) for congruence subgroups. They come from the actions of G
Q
on H
1
(SL
2
(Z)
∧
, W), where W = Sym
k−2
V
p
⊗
Zp
Z
p
[SL
2
(Z)/Γ], V
p
is Q
2
p
with
actions of Π
Q
through GL
2
(Z
p
) and are cut out using Hecke operators on
these spaces.
Proof of Theorem 2.3.1 (i). Let N(Γ, A) denote the set of holomorphic func
tions f : H →C satisfying the following conditions:
(a) for any γ ∈ Γ, f(γz) = f(z),
(b) for any γ ∈ Γ` SL
2
(Z), f ◦ γ =
¸
n≥n
0
(γ,f)
n∈
1
M
Z
a
n
q
n
, and a
n
∈ A for any n.
As ∆ ∈ S
12
(SL
2
(Z), Q) does not vanish on H, ∆
1
12
∈ S
1
(SL
2
(Z), χ, Q), where
χ : SL
2
(Z) →µ
12
. Let Γ
0
= Ker χ. If f ∈ M
k
(Γ, A), ∆
−
k
12
f ∈ N(Γ ∩ Γ
0
, A).
If f ∈ N(Γ, A), ∆
k
f ∈ M
12k
(Γ, A), where k + n
0
(γ, f) ≥ 0 for any γ ∈
52 CHAPTER 2. MODULAR FORMS
Γ` SL
2
(Z). Therefore knowing N(Γ, A) is equivalent to knowing M(Γ, A). So
it suﬃces to prove if
f =
¸
n≥n
0
a
n
q
n
∈ ^(C) =
¸
Γ
N(Γ, C)
and σ ∈ Aut C, then f
σ
∈ ^(C).
Let j =
G
3
4
a
0
(G
3
4
) ∆
= q
−1
+ ∈ N(SL
2
(Z), Q).
Proposition 2.3.2. (i) N(SL
2
(Z), Q) = Q[j], N(SL
2
(Z), C) = C[j].
(ii) j : SL
2
(Z)`H →C is bijective.
(iii) j(z) −j(α) has a zero at z = α of order e(α) =
3 if α ∈ SL
2
(Z)ρ
2 if α ∈ SL
2
(Z)i,
1 otherwise.
(iv) j(i), j(ρ) ∈ Q.
Proof. (i) Note that G
3a
4
, G
3(a−1)
4
∆, , ∆
a
is a basis of M
12a
(SL
2
(Z), Q).
(ii) and (iii): For any β ∈ C, f = (j − β) ∆ ∈ M
12
(SL
2
(Z), C), with
v
∞
(f) = 0. As D = SL
2
(Z)`H, and
¸
z∈D−ρ,i¦
γ
z
(f) +
1
2
γ
i
(f) +
1
3
(f) = 1,
we can deduce the required results.
(iv) G
4
(ρ) = 0, G
6
(i) = 0.
Let f ∈ N(Γ, C),
P
f
(X) =
¸
δ∈Γ\ SL
2
(Z)
(X −f ◦ δ) ∈ N(SL
2
(Z), C)[X] ⊂ C((q))[X]
P
f
σ(X) =
¸
δ∈Γ\ SL
2
(Z)
(X −(f ◦ δ)
σ
) ∈ N(SL
2
(Z), C)[X] ⊂ C((q))[X]
Denote P
f
(X) =
n
¸
l=0
g
l
X
l
, P
f
σ(X) =
n
¸
l=0
g
σ
l
X
l
, where g
l
∈ N(SL
2
(Z), C), and
g
σ
l
∈ N(SL
2
(Z), C) thanks to the Corollary 2.2.11. We give the proof in two
steps.
2.4. HECKE OPERATORS 53
Step 1: Prove that f
σ
is holomorphic on H, by the Proposition 2.3.2. We
have
P
f
(X) =
n
¸
l=0
P
l
(j)X
l
, P
f
σ(X) =
n
¸
l=0
P
σ
l
(j)X
l
.
The roots of P
f
are the f ◦ δ’s, where δ ∈ Γ` SL
2
(Z). They are holomorphic
on H. The roots of P
f
σ are multivalued holomorphic functions on H. In order
to prove that are single valued, it suﬃces to show there is no ramiﬁcation.
Let α be an arbitrary element in H. we have, around α, n distinct formal
solutions
+∞
¸
k=0
a
l,k
(α)(j −j(α))
k
e(α)
(1 ≤ l ≤ n)
of P
f
(X) = 0 as (j −j(α))
1
e(α)
is a local parameter around α by Proposition
2.3.2. Let β
σ
∈ H satisﬁes j(β
σ
) = j(α)
σ
, then we have e(β
σ
) = e(α).
Therefore
+∞
¸
k=0
a
l,k
(α)
σ
(j −j(β
σ
))
k
e(βσ)
, (1 ≤ l ≤ n)
are n distinct formal solutions around β
σ
. It follows that there is no ramiﬁ
cation around β
σ
, for any β
σ
. Hence the roots of P
f
σ are holomorphic on H.
In particular, f
σ
is holomorphic on H.
Step 2: Prove that there exists Γ
t
⊂ SL
2
(Z) of ﬁnite index, such that f
σ
◦γ =
f
σ
for any γ ∈ Γ. For any γ ∈ SL
2
(Z),
P
f
σ(f
σ
◦ γ) =
n
¸
l=0
g
σ
l
(f
σ
◦ γ)
l
=
n
¸
l=0
g
σ
l
◦ γ(f
σ
◦ γ)
l
= P
f
σ(f
σ
) ◦ γ = 0
So f
σ
◦ γ belongs to the ﬁnite set of roots of P
f
σ, which leads to the required
conclusion.
2.4 Hecke operators
2.4.1 Preliminary.
Let Γ ⊂ G be groups (for example, Γ = SL
2
(Z), G = GL
2
(Q)
+
), let x ∈ G,
xΓ = ¦xγ : γ ∈ Γ¦, Γx = ¦γx : γ ∈ Γ¦.
54 CHAPTER 2. MODULAR FORMS
Let A be a ring, deﬁne A[Γ`G/Γ] to be the set of φ : G → A satisfying the
following two conditions:
(i) φ(γx) = φ(xγ) = φ(x), for all x ∈ G, γ ∈ Γ.
(ii) There exists a ﬁnite set I such that φ =
¸
i∈I
λ
i
1
Γx
i
.
Remark. (i) We impose x
i
to be distinct in Γ`G, in this situation, the
decomposition is unique, λ
i
’s are unique.
(ii) For any γ ∈ Γ, 1
Γx
i
γ
(x) = 1
Γx
i
(xγ
−1
). So φ =
¸
i∈I
λ
i
1
Γx
i
∈ A[Γ`G/Γ]
implies
¸
i∈I
λ
i
1
Γx
i
γ
(x) =
¸
i∈I
λ
i
1
Γx
i
(xγ
−1
) = φ(xγ
−1
) = φ(x) =
¸
i∈I
λ
i
1
Γx
i
(x)
Therefore there exists a permutation: σ : I → I, and for any i ∈ I, there
exists γ
i
∈ Γ, such that λ
σ(i)
= λ
i
, x
i
γ = γ
i
x
σ(i)
.
Proposition 2.4.1. (i) If φ =
¸
i∈I
λ
i
1
Γx
i
, φ
t
=
¸
j∈J
µ
j
1
Γy
j
∈ A[Γ`G/Γ], then
φ ∗ φ
t
=
¸
(i,j)∈IJ
λ
i
µ
j
1
Γx
i
y
j
∈ A[Γ`G/Γ],
and it does not depend on the choices.
(ii) (A[Γ`G/Γ], +, ∗) is an associative Aalgebra with 1
Γ
as a unit.
(iii) If M is a right Gmodule with G action m → m ∗ g, and φ =
¸
λ
i
1
Γx
i
∈ A[Γ`G/Γ], then for any m ∈ M
Γ
, m∗ φ =
¸
i∈I
λ
i
m∗ x
i
does not
depend on the choices of x
i
. Moreover, m∗φ ∈ M
Γ
, m∗(φ
1
∗φ
2
) = (m∗φ
1
)∗φ
2
,
m∗ (φ
1
+ φ
2
) = (m∗ φ
1
) + (m∗ φ
2
).
Proof. Exercise, using the previous remark.
Remark. If Γ = 1, then A[Γ`G/Γ] = A[G] is commutative if and only if G
is commutative.
2.4.2 Deﬁnition of Hecke operators: R
n
, T
n
, n ≥ 1.
Let G = GL
2
(Q)
+
, Γ = SL
2
(Z).
Lemma 2.4.2. Let g ∈ G ∩ M
2
(Z), then there exists a unique pair (a, d) ∈
N −¦0¦, and b ∈ Z unique mod dZ, such that Γg = Γ
a b
0 d
2.4. HECKE OPERATORS 55
Proof. Let g =
α β
γ δ
, there exists µ, ν ∈ Z, such that (µ, ν) = 1, and
µα + νγ = 0. And there exists x, y ∈ Z, such that xν − µy = 1, Let
γ
0
=
x y
µ ν
if xα + yγ ≥ 0; γ
0
= −
x y
µ ν
if xα + yγ < 0. Then γ
0
g =
a b
0 d
,
where a > 0. Thus completes the proof of existence.
If γ
1
, γ
2
∈ Γ satisﬁes
γ
1
g =
a
1
b
1
0 d
1
γ
2
g =
a
2
b
2
0 d
2
then
(γ
1
g)(γ
2
g)
−1
=
a
1
a
2
a
2
b
1
−a
1
b
2
a
2
d
2
0
d
1
d
2
∈ SL
2
(Z)
This implies a
1
= a
2
, d
1
= d
2
, b
1
−b
2
divisible by d
1
.
Lemmadeﬁnition 2.4.3. For any n ≥ 1,
R
n
= 1
Γ(
n
0
0
n
)
∈ Z[Γ`G/Γ],
T
n
= 1
g∈M
2
(Z),det g=n¦
∈ Z[Γ`G/Γ].
Proof. Left and right invariance come from det gg
t
= det g det g
t
. And Lemma 2.4.2
implies T
n
=
¸
ad=n,a≥1
b mod d
1
Γ(
a
0
b
d
)
, so get the ﬁniteness needed.
Remark. If p is prime, Then T
p
= 1
Γ(
p
0
0
1
)Γ
by elementary divisors for prin
ciple ideal domains.
Theorem 2.4.4. (i) For any n ≥ 1 and l ≥ 1, R
n
R
l
= R
nl
= R
l
R
n
,
R
n
T
l
= T
l
R
n
.
(ii) If (l, n) = 1, T
l
T
n
= T
ln
= T
n
T
l
.
(iii) If p is prime and r ≥ 1, T
p
r T
p
= T
p
r+1 + pR
p
T
p
r−1.
(iv) Let T
Z
be the subalgebra of Z[Γ`G/Γ] generated by R
n
and T
n
(n ≥ 1).
It is a commutative algebra.
Proof. (i) It is trivial.
(ii) We have
T
n
T
l
=
¸
ad=n,a≥1
b mod d
¸
a
d
=n,a
≥1
b
mod d
1
Γ
“
aa
0
ab
+bd
dd
”
.
56 CHAPTER 2. MODULAR FORMS
As (n, l) = 1, (a, a
t
) = 1, (a, d
t
) = 1. This implies ¦aa
t
: a[n, a
t
[l¦ =
¦a
tt
: a
tt
[nl¦. Therefore in order to show T
n
T
l
= T
nl
, it suﬃces to verify
that ¦ab
t
+ bd
t
¦ is a set of representatives of Z/(dd
t
)Z, where b is a set of
representatives of Z/dZ, b
t
is a set of representatives of Z/d
t
Z. It suﬃces to
show the injectivity under the mod dd
t
Z map. If
ab
t
1
+ b
1
d
t
≡ ab
t
2
+ b
2
d
t
,
then b
t
1
≡ b
t
2
mod d
t
, so b
t
1
= b
t
2
, which leads to the required conclusion.
(iii) We have
T
p
r =
r
¸
i=0
¸
b mod p
i
1
Γ
“
p
r−i
0
b
p
i
”
, T
p
= 1
Γ(
p
0
0
1
)
+
¸
c mod p
1
Γ(
1
0
c
p
)
Then
T
p
r T
p
=
r
¸
i=0
¸
b mod p
i
1
Γ
“
p
r+1−i
0
b
p
i
”
+
r
¸
i=0
¸
b mod p
i
¸
c mod p
1
Γ
“
p
r−i
0
pb+p
r−i
c
p
i+1
”
=T
p
r+1 + R
p
(
r−1
¸
i=0
¸
b mod p
i
¸
c mod p
1
Γ
“
p
r−1−i
0
b+p
r−1−i
p
i
”
) = T
p
r+1 + pR
p
T
p
r−1.
(iv) It follows from (i),(ii),(iii).
2.4.3 Action of Hecke operators on modular forms.
The following two propositions are exercises in group theory.
Proposition 2.4.5. Assume G ⊃ Γ are groups. Then
(i) If [Γ : Γ
t
] < +∞, then Γ
t
contains some Γ
tt
which is normal in Γ, and
[Γ : Γ
tt
] < +∞.
(ii) If [Γ : Γ
1
] < +∞, [Γ : Γ
2
] < +∞, then [Γ : Γ
1
∩ Γ
2
] < +∞.
(iii) If H
t
⊂ H ⊂ G, [H : H
t
] < +∞, then [H ∩ Γ : H
t
∩ Γ] < +∞.
Proposition 2.4.6. (i) Suppose α ∈ GL
2
(Q)
+
, and N ∈ N such that Nα,
Nα
−1
∈ M
2
(Z), then
α
−1
SL
2
(Z)α ∩ SL
2
(Z) ⊃ Γ(N
2
) := SL
2
(Z) ∩ (1 + N
2
M
2
(Z)).
(ii) If [SL
2
(Z) : Γ] < +∞, α ∈ GL
2
(Q)
+
, then
[SL
2
(Z) : SL
2
(Z) ∩ α
−1
Γα] < +∞.
2.4. HECKE OPERATORS 57
Proposition 2.4.7.
´
k
(C) =
¸
[SL
2
(Z):Γ]<+∞
M
k
(Γ, C), o
k
(C) =
¸
[SL
2
(Z):Γ]<+∞
S
k
(Γ, C)
are stable under GL
2
(Q)
+
.
Proof. For any γ ∈ Γ, f
[k
γ = f. For α ∈ GL
2
(Q)
+
, we have
(f
[k
α)
[k
(α
−1
γα) = f
[k
α,
so f
[k
α is invariant for the group α
−1
Γα ∩ SL
2
(Z).
To verify that f
[k
α is slowly increasing at ∞, write α = γ
a b
0 d
for some
γ ∈ SL
2
(Z), then
(f
[k
α)(z) = (ad)
k−1
d
−k
(f
[k
γ)
az + b
d
,
then we get the result.
Let Γ = SL
2
(Z), G = GL
2
(Q)
+
, ϕ =
¸
i∈I
λ
i
1
Γγ
i
∈ Z[Γ`G/Γ], we deﬁne
f
[k
ϕ =
¸
i∈I
λ
i
f
[k
γ
i
, for f ∈ M
k
(1) = ´
k
(C)
Γ
.
The deﬁnition is independent of the choice of γ
i
. From the general theory,
we have
(f
[k
ϕ)
[k
ϕ
t
(z) = f
[k
(ϕ ∗ ϕ
t
)(z).
If f ∈ M
k
(1) (resp. S
k
(1)), then f
[k
ϕ ∈ M
k
(1) (resp. S
k
(1)).
Facts: f
[k
R
n
= n
k−2
f, and f
[k
T
n
= n
k−1
¸
ad=n,a≥1
bmod d
d
−k
f(
az+b
d
).
Proposition 2.4.8. If f =
∞
¸
m=0
a
m
(f)q
m
, then a
m
(f
[k
T
n
) =
¸
a≥1,
a(m,n)
a
k−1
a
mn
a
2
(f).
Proof. For ﬁxed d[n, d ≥ 1,
¸
b mod d
d
−k
f(
az+b
d
) = d
−k
¸
b mod d
∞
¸
m=0
a
m
(f)e
2πim
az+b
d
= d
−k
∞
¸
m=0
a
m
(f)e
2πinaz/d
¸
b mod d
e
2πimb/d
= d
1−k
∞
¸
m=0
d[m
a
m
(f)e
2πimaz/d
= d
1−k
∞
¸
l=0
a
dl
(f)q
al
.
58 CHAPTER 2. MODULAR FORMS
So
f
[k
T
n
= n
k−1
¸
ad=n,a≥1
d
1−k
∞
¸
l=0
a
dl
(f)q
al
,
summing the coeﬃcients of q
m
, this gives:
a
m
(f
[k
T
n
) = n
k−1
¸
a≥1
a[(m,n)
(n/a)
1−k
a
mn
a
2
(f)
=
¸
a≥1
a[(m,n)
a
k−1
a
mn
a
2
(f).
Corollary 2.4.9. (i) M
k
(Γ, Z) and M
k
(Γ, Q) are stable under T
n
and R
n
.
(ii) a
0
(f
[k
T
n
) =
¸
a[n
a
k−1
a
0
(f) = σ
k−1
(n)a
0
(f).
(iii) a
1
(f
[k
T
n
) = a
n
(f), therefore f is determined by
T −→a
1
(f
[k
T).
2.5 Petersson scalar product.
Lemma 2.5.1.
SL
2
(Z)\1
dxdy
y
2
=
1
2
−
1
2
+∞
√
1−x
2
dxdy
y
2
=
π
3
< ∞.
Corollary 2.5.2. (i) If [SL
2
(Z) : Γ] < +∞, then
Γ\1
dxdy
y
2
=
π
3
C(Γ),
where C(Γ) = [PSL
2
(Z) :
¯
Γ],
¯
Γ is the image of Γ in PSL
2
(Z).
(ii) If α ∈ GL
2
(Q)
+
such that α
−1
Γα ⊂ SL
2
(Z), then C(α
−1
Γα) = C(Γ).
Proof. (i) Since
dxdy
y
2
is invariant under the action of Γ, the integral is well
deﬁned. Put ¦γ
i
¦ be a family of representatives of Γ` SL
2
(Z), then Γ`H =
¸
γ
i
(D) up to sets of measure 0 (maybe have overlap in SL
2
(Z)i ∪SL
2
(Z)ρ).
(ii) Since Γ`H = α
α
−1
Γα`H
, the two integrals are the same by the
invariance of
dxdy
y
2
.
2.5. PETERSSON SCALAR PRODUCT. 59
Let f, g ∈ S
k
(C), choose Γ ⊂ SL
2
(Z) of ﬁnite index such that f, g ∈
S
k
(Γ, C).
Proposition 2.5.3.
'f, g` :=
1
C(Γ)
Γ\1
f(z)g(z)y
k
dxdy
y
2
converges and is independent of the choice of Γ.
Proof. For γ ∈ Γ, we have
f(γz) = (cz + d)
k
f(z), g(γz) = (cz + d)
k
g(z),
Im(γz) =
Im z
[cz + d[
2
.
so f(z)g(z)y
k
is invariant under Γ. Now Γ`H =
¸
i∈I
γ
i
D with [I[ = C(Γ). So
if Γ
t
also satisfy that f, g ∈ S
k
(Γ
t
, C), then f, g ∈ S
k
(Γ ∩ Γ
t
, C), and
1
C(Γ)
Γ\1
f(z)g(z)y
k
dxdy
y
2
=
1
C(Γ ∩ Γ
t
)
(Γ∩Γ
)\1
f(z)g(z)y
k
dxdy
y
2
=
1
C(Γ
t
)
Γ
\1
f(z)g(z)y
k
dxdy
y
2
.
Because f
[k
γ
i
and g
[k
γ
i
are exponentially decreasing as y → ∞ on D,
'f, g` converges.
Remark. In fact, we can choose one modular form and one cusp form, and
the integral will still converge.
Proposition 2.5.4. For f ∈ S
k
(1), we have 'G
k
, f` = 0.
Proof. By deﬁnition,
G
k
(z) =
1
2
Γ(k)
(−2πi)
k
¸
m,n
t
1
(mz + n)
k
∈ M
k
(1),
and
¸
m,n
t
1
(mz + n)
k
=
∞
¸
a=1
¸
(m,n)=1
1
(amz + an)
k
=
Γ(k)
(2πi)
k
ζ(k)
¸
γ∈Γ∞\ SL
2
(Z)
1
(cz + d)
k
,
60 CHAPTER 2. MODULAR FORMS
where Γ
∞
denotes the subgroup of SL
2
(Z) consisting of all upper triangular
matrices. So we just compute '
¸
γ∈Γ∞\ SL
2
(Z)
1
(cz+d)
k
, f`. We have
'
¸
γ∈Γ∞\ SL
2
(Z)
1
(cz+d)
k
, f` =
SL
2
(Z)\1
¸
γ∈Γ∞\ SL
2
(Z)
1
(cz+d)
k
f(z)y
k dxdy
y
2
=
SL
2
(Z)\1
¸
γ∈Γ∞\ SL
2
(Z)
f(γz) Im(γz)
k dxdy
y
2
=
Γ∞\1
f(z)y
k dxdy
y
2
=
∞
0
1
0
f(x + iy)y
k−2
dxdy = 0,
where the last equality is because a
0
(f) = 0 and
1
0
e
2πinx
dx = 0 for n ≥
1.
Lemma 2.5.5. (i) For α ∈ GL
2
(Q)
+
, we have
'f
[k
α, g
[k
α` = (det α)
k−2
'f, g`.
(ii) Let α
t
= (det α)α
−1
, then 'f
[k
α, g` = 'f, g
[k
α
t
`.
Proof. (i) Choose Γ such that f, g ∈ S
k
(Γ) and α
−1
Γα ⊂ SL
2
(Z), then
C(α
−1
Γα)'f
[k
α, g
[k
α` = (det α)
2(k−1)
α
−1
Γα\1
f(αz)g(αz)
y
k
[cz + d[
2k
dxdy
y
2
= (det α)
k−2
Γ\1
f(z)g(z)y
k
dxdy
y
2
= (det α)
k−2
C(Γ)'f, g`.
(ii) Replace g by g
[k
α
−1
, then we get
'f
[k
α, g` = (det α)
k−2
'f, g
[k
α
−1
`
= (det α)
k−2
'f, g
[k
1
det α
α
t
`
= 'f, g
[k
α
t
`.
2.6 Primitive forms
Theorem 2.6.1. (i) If n ≥ 1, then R
n
and T
n
are hermitian.
(ii) The eigenvalues of T
n
are integers in a totally real ﬁeld.
(iii) S
k
(1) has a basis of common eigenvectors for all T
n
, n ≥ 1.
2.6. PRIMITIVE FORMS 61
Proof. (i) It is trivial for R
n
. Since T
Z
is generated by R
p
and T
p
for p prime,
it suﬃces to consider T
p
.
Let α ∈ M
2
(Z), det α = p, then there exist γ
1
, γ
2
∈ SL
2
(Z) such that
α = γ
1
p 0
0 1
γ
2
, then
'f
[k
α, g` = 'f
[k
(γ
1
p 0
0 1
γ
2
), g`
= 'f
[k
p 0
0 1
, g
[k
γ
t
2
`
= 'f
[k
p 0
0 1
, g`
= 'f, g
[k
p 0
0 1
`,
thus 'f
[k
T
p
, g` = (p + 1)'f
[k
p 0
0 1
, g` = 'f, g
[k
T
p
`.
(ii) S
k
(SL
2
(Z), Z) is a lattice in S
k
(1) stable under T
n
, so det(XI −T
n
) ∈
Z[X], so the roots are algebraic integers, and real since T
n
is hermitian.
(iii) T
n
t
s are hermitian, hence they are semisimple. Since the T
n
commute
to each other, by linear algebra, there exists a common basis of eigenvectors
for all T
n
.
Theorem 2.6.2. Let f =
+∞
¸
n=0
a
n
(f)q
n
∈ M
k
(1) − ¦0¦. If for all n, f
[k
T
n
=
λ
n
f, then
(i) a
1
(f) = 0;
(ii) if f is normalized, i.e. a
1
(f) = 1, then a
n
(f) = λ
n
, for all n, and
(a) a
mn
(f) = a
m
(f)a
n
(f) when (m, n) = 1.
(b) a
p
(f)a
p
r (f) = a
p
r+1(f) + p
k−1
a
p
r−1(f) for p prime and r ≥ 1.
Proof. (i) Since a
n
(f) = a
1
(f
[k
T
n
) = a
1
(λ
n
f) = λ
n
a
1
(f), if a
1
(f) = 0, then
f = 0.
(ii) The ﬁrst assertion is obvious, and the other two follow by the same
formulae for the R
p
, T
p
.
Deﬁnition 2.6.3. f ∈ S
k
(1) is called primitive if a
1
(f) = 1 and f is an
eigenform for all Hecke operators.
Theorem 2.6.4. (i) If f, g are primitive with the same set of eigenvalues,
then f = g. (called “Multiplicity 1 theorem”).
(ii) The primitive forms are a basis of S
k
(1).
Proof. (i) Apply (i) of the previous theorem to f −g, since a
1
(f −g) = 0, so
f = g.
62 CHAPTER 2. MODULAR FORMS
(ii) By (iii) of Theorem 2.6.1, there exists a basis of primitive forms. For
any two distinct such forms f and f
t
, then there exist n and λ = λ
t
such that
f
[k
T
n
= λf, f
t
[k
T
n
= λ
t
f,
then λ'f, f
t
` = 'f
[k
T
n
, f
t
` = 'f, f
t
[k
T
n
` = λ
t
'f, f
t
`, so 'f, f
t
` = 0. Therefore
one has to take all the primitive forms to get a basis of S
k
(1).
Remark. Since (G
k
)
[k
T
n
= σ
k−1
(n)G
k
, we get a basis of M
k
(1) of eigenforms.
Example 2.6.5. Write
∆ = q
∞
¸
n=1
(1 −q
n
)
24
=
∞
¸
n=1
τ(n)q
n
,
where τ(n) is Ramanujan’s τfunction. Then
τ(mn) = τ(m)τ(n), if (m, n) = 1,
τ(p)τ(p
r
) = τ(p
r+1
) + p
11
τ(p
r−1
), if p is a prime, n ≥ 1.
Proof. Since S
12
(1) = C ∆, and is stable by the T
n
, ∆ is an eigenform of T
n
with eigenvalue τ(n).
Remark. In 1973, Deligne proved Ramanujan’s conjecture that
[τ(p)[ ≤ 2p
11/2
(⇐⇒Re (s) = 11/2, if 1 −τ(p)p
−s
+ p
11−2s
= 0)
as a consequence of the proof of Riemann Hypothesis (Weil Conjecture) for
zeta functions of varieties over ﬁnite ﬁelds.
Chapter 3
padic Lfunctions of modular
forms
3.1 Lfunctions of modular forms.
3.1.1 Estimates for the fourier coeﬃcients
Proposition 3.1.1. Let Γ ⊂ SL
2
(Z) be a subgroup of ﬁnite index, let f =
¸
n∈
1
M
N
a
n
(f)q
n
∈ M
k
(Γ, C). Then
(i)
a
n
(f) =
O(n
k−1
), if k ≥ 3;
O(nlog n), if k = 2;
O(
√
n), if k = 1.
(ii) a
n
(f) = O(n
k/2
), if f ∈ S
k
(Γ).
Proof. We have that
a
n
(f) = e
2πny
y
−
k
2
1
M
M
0
y
k
2
f(x + iy)e
−2πinx
dx, ∀ y.
Deﬁne
ϕ(z) = y
k
2
sup
δ∈Γ\ SL
2
(Z)
[f
[k
δ(z)[.
It is ﬁnite since [SL
2
(Z) : Γ] < +∞, and ϕ(γz) = ϕ(z) for γ ∈ SL
2
(Z).
63
64 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS
Let D be the fundamental domain of SL
2
(Z). For any δ ∈ Γ` SL
2
(Z),
there exists C
δ
such that, for all z ∈ D,
[f
[k
δ(z) −a
0
(f
[k
δ)[ ≤ C
δ
e
−
2πy
M
.
Let C = sup
δ
C
δ
, ψ(z) = sup
(c,d),=(0,0)
y
[cz+d[
2
, then ϕ(z) ≤ Cψ(z)
k/2
+ B for some
B.
a
n
(f) ≤ e
2πny
y
−
k
2
1
M
M
0
ϕ(x + iy)dx
≤ e
2πny
y
−
k
2
1
M
M
0
(Cψ(x + iy)
k/2
+ B)dx.
If C = 0, take y =
1
Mn
, then we get (ii).
We now need to evaluate
M
0
ψ(x + iy)
k
2
.
Let y ≤ 1 (in application, y =
1
Mn
), then ψ(x + iy) ≤
1
y
. Let j ∈ N. If
ψ(x + iy) ≥
1
4
j
y, there exists (c, d) such that c
2
y
2
+ (cx + y)
2
≤ 4
j
y
2
, hence
there exist c, d ∈ Z, such that
1 ≤ [c[ ≤ 2
j
, [cx + d[ ≤ 2
j
y.
Now
Meas(¦x ∈ [0, M] : ∃d, s.t.[cx + d[ ≤ 2
j
y¦) ≤ 2
j+1
yM,
so Meas(¦x ∈ [0, M] : ψ(x + iy) ≥
1
4
j
y
¦) ≤ 4
j
2yM, and
M
0
ψ(x + iy)
k/2
dx
≤
[−log
4
y]
¸
j=1
Meas(¦x ∈ [0, M] :
1
4
j
y
≤ ψ(x + iy) ≤
1
4
j−1
y
¦)(
1
4
j−1
y
)
k/2
+4
k/2
Meas(¦x ∈ [0, M] : ψ(x + iy) ≤ 4¦)
≤ M4
k/2
+
[−log
4
y]
¸
j=1
4
j
2yM(
1
4
j−1
y
)
k/2
= M4
k/2
1 + 2
[−log
4
y]
¸
j=1
y
1−k/2
4
j(1−k/2)
.
When k ≥ 3, let y = 1/Mn. As
[−log
4
y]
¸
j=1
4
j(1−k/2)
converges, we get a
n
(f) =
O(n
k−1
). When k = 2, it is obvious. For k = 1,
[−log
4
y]
¸
j=1
y
1−k/2
4
j(1−k/2)
<
2 −y
1/2
< 2, then we get the result.
3.1. LFUNCTIONS OF MODULAR FORMS. 65
Remark. (i) L(f, s) =
¸
n,=0
a
n
(f)n
−s
converges for Re (s) 0.
(ii) If Γ is a congruence subgroup, f ∈ S
k
(Γ), Deligne showed that
a
n
(f) = O(n
(k−1)/2+ε
), ∀ ε > 0
in the same theorem mentioned above.
Question: What about the noncongruence subgroups?
3.1.2 Dirichlet series and Mellin transform
Deﬁnition 3.1.2. Let ¦a
n
¦
n≥1
be a sequence in C, the Dirichlet series of
(a
n
) is D(s) =
∞
¸
n=1
an
n
s
.
Lemma 3.1.3. If D(s
0
) converges, then D(s) converges uniformly on com
pact subsets of Re (s) > Re (s
0
).
Proof. One can assume s
0
= 0, then use Abel’s summation.
Corollary 3.1.4. There exists a maximal half plane of convergence (resp.
absolute convergence).
Remark. (i) if f(z) =
∞
¸
n=0
a
n
z
n
, then the maximal open disc of convergence
of f is the maximal open disc of absolute converge, and also is the maximal
open disc of center 0 on which f can be extended analytically.
(ii) Let a
n
= (−1)
n−1
, then D(s) = (1 − 2
1−s
)ζ(s), which converges for
Re (s) > 0, absolutely converges Re (s) > 1 and can be extended analytically
to C.
(iii) In general you can’t extend D(s) outside its half plane of absolute
convergence, but for D(s) coming from number theory, it seems that you can
always extend meromorphically to C (Langlands program).
We review some basic facts about Mellin transform:
Proposition 3.1.5. (i) Let ϕ : R
∗
+
→ C be in (
r
, and suppose there exist
A > B satisfying, for 0 ≤ i ≤ n,
ϕ
(i)
(t) =
O(t
A−i
) near 0
O(t
B−i
) near ∞.
66 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS
Let
Mel(ϕ, s) :=
∞
0
ϕ(t)t
s
dt
t
.
Then it is holomorphic on −A < Re (s) < −B, and O([s[
−r
) on −A < a ≤
Re (s) ≤ b < −B.
(ii) If r ≥ 2, ϕ(x) =
1
2πi
C+i∞
C−i∞
Mel(ϕ, s)x
−s
ds, for any C with −a < C <
−B.
Proof. (i) The ﬁrst assertion is clear. For the second, use
Mel(ϕ, s) = (−1)
r
1
s(s + 1) (s + r −1)
Mel(ϕ
(r)
, s + r).
(ii) Mel(ϕ, C + it) =
ˆ
ψ
C
(t), where ψ
C
(x) = ϕ(e
x
)e
Cx
, and
ˆ
ψ
C
is the
Fourier transform of ψ
C
. Then use Fourier inversion formula.
3.1.3 Modular forms and Lfunctions
For f =
∞
¸
n=0
a
n
(f)q
n
∈ M
2k
(1), deﬁne
L(f, s) =
∞
¸
n=1
a
n
(f)
n
s
, Λ(f, s) =
Γ(s)
(2π)
s
L(f, s).
Example 3.1.6. Take f = G
2k
, we get
L(G
2k
, s) =
∞
¸
n=1
σ
2k−1
(n)
n
s
=
∞
¸
n=1
¸
ad=n
d
2k−1
(ad)
−s
=
∞
¸
a=1
a
−s
∞
¸
d=1
d
2k−1−s
= ζ(s)ζ(s −2k + 1).
Theorem 3.1.7. (i) L(f, s) absolutely converges for Re (s) > 2k;
(ii) (a) Λ(f, s) has a meromorphic continuation to C;
(b) Λ(f, s) is holomorphic except for simple poles at s = 0 of residue
a
0
(f) and 2k of residue (−1)
k
a
0
(f);
(c) Λ(f, 2k −s) = (−1)
k
Λ(f, s);
(d) Λ(f, s) goes to zero at ∞ in each vertical strip.
3.1. LFUNCTIONS OF MODULAR FORMS. 67
Proof. (i) The result follows from a
n
(f) = O(n
2k−1
).
(ii) Let ϕ(t) = f(it) − a
0
(f), then ϕ is C
∞
on R
∗
+
, and ϕ(t) = O(e
−2πt
)
at ∞. f ∈ M
2k
(1) implies
ϕ(t
−1
) = (−1)
k
t
2k
ϕ(t) + (−1)
k
a
0
(f)t
2k
−a
0
(f).
For Re (s) > 0, we have
+∞
0
e
−2πnt
t
s dt
t
=
Γ(s)
(2πn)
s
. Then for Re (s) > k,
Λ(f, s) =
∞
¸
n=1
a
n
(f)
Γ(s)
(2πn)
s
=
+∞
0
ϕ(t)t
s dt
t
=
+∞
1
ϕ(t)t
s dt
t
+
+∞
1
ϕ(t
−1
)t
−s dt
t
=
+∞
1
ϕ(t)(t
s
+ (−1)
k
t
2k−s
)
dt
t
−a
0
(f)
(−1)
k
2k−s
+
1
s
, (∗)
since the ﬁrst term is holomorphic for all s ∈ C, this gives (a) and (b).
Replacing s by 2k − s in (∗), we get (c). (d) follows from integration by
part.
Theorem 3.1.8 (Hecke’s converse theorem). Let (c
n
)
n∈N
be a sequence
in C such that L(s) =
∞
¸
n=1
cn
n
s
converges for Re (s) > A, and Λ(s) =
Γ(s)
(2π)
s
L(s)
satisfy (ii)(a) −(d) of previous theorem, then
f(z) :=
∞
¸
n=0
c
n
q
n
∈ M
2k
(1).
Proof. Since f(z) converges if [q[ < 1, it is holomorphic on H. Obviously
f(z + 1) = f(z), we just have to verify
g(z) = f(−
1
z
) −z
2k
f(z) = 0 on H.
It suﬃces to prove that g(it) = 0 for t > 0. Let
ϕ(t) = f(it) −c
0
=
∞
¸
n=1
c
n
e
−2πnt
,
one can check that Λ(s) = Mel(ϕ, s). Take c > A, then
ϕ(t ) −
(−1)
k
t
2k
ϕ(t
−1
)
=
1
2πi
c+i∞
c−i∞
Λ(s)t
−s
ds −(−1)
k
c+i∞
c−i∞
Λ(s)t
s−2k
ds
=
1
2πi
c+i∞
c−i∞
Λ(s)t
−s
ds −
c+i∞
c−i∞
Λ(2k −s)t
s−2k
ds
=
1
2πi
c+i∞
c−i∞
Λ(s)t
−s
ds −
2k−c+i∞
2k−c−i∞
Λ(s)t
−s
ds
.
68 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS

6
?
r r c 2k −c
0 2k
↓
−∞
−R
↑
+∞
R
γ
R
Consider the integral of the function Λ(s)t
−s
around the closed path γ.
Since Λ(s) →0 on vertical strips, by Cauchy formula,
lim
R→+∞
γ
R
Λ(s)t
−s
ds =
c+i∞
c−i∞
Λ(s)t
−s
ds −
2k−c+i∞
2k−c−i∞
Λ(s)t
−s
ds
= 2πi
res
s=0
(Λ(s)t
−s
) + res
s=2k
(Λ(s)t
−s
)
= 2πi(−c
0
+ (−1)
k
c
0
t
−2k
).
So
ϕ(t) −
(−1)
k
t
2k
ϕ(t
−1
) −(−c
0
+ (−1)
k
c
0
t
−2k
) = 0,
by an easy computation, the left hand is just
(−1)
k
t
2k
(−g(it)), then we get
g(it) = 0, which completes the proof.
3.1.4 Euler products
Theorem 3.1.9. If f =
∞
¸
n=0
a
n
(f)q
n
∈ M
2k
(1) is primitive, then
L(f, s) =
¸
p
1
1 −a
p
(f)p
−s
+ p
2k−1−2s
.
Proof. By Theorem 2.6.2, a
nm
(f) = a
n
(f)a
m
(f) whenever (n, m) = 1, so
L(f, s) =
¸
p
∞
¸
r=0
a
p
r (f)p
−rs
.
Since,
a
p
r+1 −a
p
a
p
r + p
2k−1
a
p
r−1 = 0,
3.2. HIGHER LEVEL MODULAR FORMS 69
multiplying by p
−(r+1)s
, and summing over r from 1 to +∞, we get
∞
¸
r=2
a
p
r p
−rs
−a
p
p
−s
∞
¸
r=1
a
p
r p
−rs
+ p
2k−1−2s
∞
¸
r=0
a
p
r p
−rs
= 0.
Using the fact that a
1
= 1, the result follows.
3.2 Higher level modular forms
3.2.1 Summary of the results
For N ≥ 2, deﬁne
Γ
0
(N) =
a b
c d
∈ SL
2
(Z) : c ≡ 0 (mod N)
¸
.
and write S
k
(Γ
0
(N)) = S
k
(N).
Exercise. If DM[N, f ∈ S
k
(M), let f
D
(z) = f(Dz), then f
D
∈ S
k
(N). Such
a form is said to be old if M = N.
Deﬁnition 3.2.1. S
new
k
(N) = ¦f ∈ S
k
(N) : 'f, g` = 0, ∀ g “old”¦.
On S
k
(N), we have the Hecke operators T
n
, (n, N) = 1,
f
[k
T
n
= n
k−1
¸
ad=n,a>1
b mod d
d
−k
f
az + b
d
,
and for p [ n, the operator
f
[k
U
p
=
1
p
p−1
¸
i=0
f
z + i
p
.
We also have a involution w
N
given by
f
[k
w
N
= N
−
k
2
z
−k
f
−
1
Nz
.
Deﬁnition 3.2.2. f ∈ S
k
(N) is called primitive if f ∈ S
new
k
(N), a
1
(f) = 1
and f
[k
T
n
= a
n
(f)f, whenever (n, N) = 1.
70 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS
Theorem 3.2.3. (i) The primitive forms are a basis of S
new
k
(N).
(ii) If f is primitive, then Q(¦a
n
(f)¦, n ∈ N) is a totally real number ﬁeld,
a
n
(f) are integers, and f
σ
is primitive for all σ ∈ Aut(C).
(iii) If f is primitive, then
(a) a
nm
(f) = a
n
(f)a
m
(f) if (n, m) = 1, (nm, N) = 1;
(b) For p N, a
p
r+1 −a
p
(f)a
p
r
(f)
+ p
k−1
a
p
r−1(f) = 0.
(c) f
[k
U
p
= a
p
(f)f, and this implies a
p
r (f) = (a
p
(f))
r
for p[N;
(d) There exists ε
f
= ±1, such that f
[k
w
N
= ε
f
f.
Theorem 3.2.4. Suppose f =
∞
¸
n=1
a
n
q
n
∈ S
k
(N) is primitive. Deﬁne
L(f, s) =
∞
¸
n=1
a
n
n
s
, Λ(f, s) = Γ(s)
√
N
2π
s
L(f, s).
Then
(i) L(f, s) =
¸
p[N
1
1−app
−s
¸
pN
1
1−app
−s
+p
k−1−2s
;
(ii) Λ(s) has an analytic continuation to C. And
Λ(f, s) = i
−k
ε
f
Λ(f, k −s);
(iii) More generally, if (D, N) = 1, χ : (Z/DZ)
∗
→ C is a character of
conductor D. Then
(a) f ⊗χ =
∞
¸
n=1
a
n
χ(n)q
n
∈ S
k
(ND
2
, χ
2
);
(b) L(f ⊗χ, s) =
¸
p[N
1
1−χ(p)app
−s
¸
pN
1
1−χ(p)app
−s
+χ
2
(p)p
k−1−2s
;
(c) Λ(f ⊗ χ, s) = Γ(s)
D
√
N
2π
s
L(f ⊗ χ, s) has a analytic continuation
to C and
χ(−N)
Λ(f ⊗χ, s)
G(x)
= i
−k
ε
f
Λ(f ⊗χ
−1
, s)
G(χ
−1
)
where G(χ) is the Gauss sum
G(χ) =
¸
x∈(Z/DZ)
∗
χ(x)e
2πix
D
.
Theorem 3.2.5 (Weil’s Converse Theorem). Conversely, if (a
m
)
m≥1
satisfy (b) and (c) of condition (iii) of the above theorem for all χ of conductor
D, (D, N) = 1, then
∞
¸
m=1
a
m
q
m
∈ S
k
(N) and is primitive.
3.3. ALGEBRAICITY OF SPECIAL VALUES OF LFUNCTIONS 71
3.2.2 TaniyamaWeil Conjecture
Let Λ be a ﬁnitely generated Zalgebra. Deﬁne its HasseWeil zeta function
ζ
Λ
(s) by
ζ
Λ
(s) =
¸
℘ prime in Λ
1
(1 −[Λ/℘[
−s
)
.
Conjecture 3.2.6 (HasseWeil). ζ
Λ
has a meromorphic continuation to
C.
Let E : y
2
= x
3
+ ax
2
+ bx + c, a, b, c ∈ Q be an elliptic curve, Λ
E
=
Z[x, y]/(y
2
− x
3
− ax
2
− bx − c) be its coordinate ring, which is a ﬁnitely
generated algebra over Z.
Theorem 3.2.7 (Wiles, BreuilConradDiamondTaylor). There exists
a unique N
E
and f
E
∈ S
2
(N
E
) which is primitive, such that
ζ
Λ
E
∼
ζ(s −1)
L(f
E
, s)
while ∼ means up to multiplication by a ﬁnite numbers of Euler factors.
Remark. This proves HasseWeil conjecture in this case thanks to theorem
3.2.4.
Theorem 3.2.8 (MordellWeil). E(Q) ∪ ¦∞¦ · Z
r(E)
⊕ﬁnite group.
Conjecture 3.2.9 (Birch,SwinnertonDyer). ord
s=1
L(f
E
, s) = r(E).
3.3 Algebraicity of special values of Lfunctions
3.3.1 Modular symbols.
Let N ≥ 1, f ∈ S
k
(N), P ∈ A[x]
(k−2)
(polynomials of degree ≤ k − 2) with
A ⊂ C a subring. For r ∈ Q, the integral
i∞
r
f(z)P(z)dz converges because
f is exponentially small around i∞and r. These integrals are called modular
symbols.
For 0 ≤ j ≤ k −2, deﬁne
r
j
(f) =
i∞
0
f(z)z
j
dz =
Γ(j + 1)
(−2πi)
j+1
L(f, j + 1).
Let L
f
be the Zmodule generated by r
j
(f
[k
δ), 1 ≤ j ≤ k − 2 and δ ∈
Γ
0
(N)` SL
2
(Z). Then L
f
is ﬁnitely generated.
72 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS
Theorem 3.3.1. If P ∈ A[x]
(k−2)
, r ∈ Q, then
i∞
r
f(z)P(z)dz ∈ AL
f
⊂ C.
Proof. For γ ∈ SL
2
(Z),
γ(i∞)
γ(0)
f(z)P(z)dz =
i∞
0
f(γz)P(γz)d(γz)
=
i∞
0
f
[k
γ(z)P
[2−k
γ(z)dz,
where P
[2−k
γ(z) = (cz + d)
k−2
P(
az+b
cz+d
) ∈ A[x]
(k−2)
. Take r = a/b, (a, b) =
1, then there exists γ
l
=
a
l−1
a
l
b
l−1
b
l
∈ SL
2
(Z) satisfying (a
0
, b
0
) = (1, 0),
(a
n
, b
n
) = (a, b).
i∞
r
f(z)P(z)dz =
n
¸
l=1
a
l−1
b
l−1
a
l
b
l
f(z)P(z)dz
=
n
¸
l=1
γ
l
(i∞)
γ
l
(0)
f(z)P(z)dz
=
n
¸
l=1
i∞
0
f
[k
γ
l
(z)P
[2−k
γ
l
(z)dz ∈ A L
f
.
Exercise. For N = 1, let L
+
f
(resp. L
−
f
) be the Zmodule generated by r
j
(f)
for all odd (resp. even) j. For P ∈ A[X]
(k−2)
, r ∈ Q, ε = ±, then
i∞
r
f(z)P(z)dz −ε
i∞
−r
f(z)P(−z)dz ∈ A L
ε
f
.
Corollary 3.3.2. (i) Suppose f ∈
∞
¸
n=1
a
n
q
n
, φ : Z →
¯
Q is constant mod
MZ for some M. Then L(f, φ, s) =
∞
¸
n=1
φ(n)
an
n
s
has an analytic continuation
to C and
Λ(f, φ, j) =
Γ(j)
(−2πi)
j
L(f, φ, j) ∈
¯
Q L
f
,
if 1 ≤ j ≤ k −1.
(ii) If N = 1 and φ(−x) = ε(−1)
j
φ(x), then Λ(f, φ, j) ∈
¯
Q L
ε
f
, if
1 ≤ j ≤ k −1.
3.3. ALGEBRAICITY OF SPECIAL VALUES OF LFUNCTIONS 73
Proof. we may assume φ(n) = e
2πi
nu
M
for some 0 ≤ u ≤ M −1 because such
functions form a basis, then
Γ(s)
(2π)
s
L(f, φ, s) =
+∞
0
¸
∞
n=1
a
n
e
2πi
nu
M
e
−2πny
y
s dy
y
=
+∞
0
f(
u
M
+ iy)y
s dy
y
,
this proves the ﬁrst assertion of (i) as f is exponentially small around i∞
and
u
M
.
Λ(f, φ, j) =
+∞
0
f(
u
M
+ iy)(iy)
j
d(iy)
iy
=
i∞
u
M
f(z)(z −
u
M
)
j−1
dz
∈ Q L
f
.
For (ii), we may assume φ(n) = e
2πi
nu
M
+ ε(−1)
j
e
−2πi
nu
M
, and similarly,
Λ(f, φ, j) =
i∞
u
M
f(z)(z −
u
M
)
j−1
dz + ε(−1)
j
i∞
−
u
M
f(z)(z +
u
M
)
j−1
dz
=
i∞
u
M
f(z)(z −
u
M
)
j−1
dz −ε
i∞
−
u
M
f(z)(−z −
u
M
)
j−1
dz,
then one uses the exercise.
3.3.2 The results
Theorem 3.3.3. If f is primitive, then there exist Ω
+
f
and Ω
−
f
∈ C, if
φ : Z →
¯
Q (mod MZ), 1 ≤ j ≤ k −1, φ(x) = ε(−1)
j
φ(−x),
then Λ(f, φ, j) ∈
¯
Q Ω
ε
f
.
Proof. We prove the case N = 1, ε = 1.
We shall prove that
r
k−2
(f)r
l
(f) ∈
¯
Q'f, f`, for l odd. (3.1)
This implies
Ω
+
f
∼
'f, f`
r
k−2
(f)
∼
'f, f`
L(f, k −1)
π
k−2
where ∼stands for equality up to multiplication by an algebraic number. The
method to show (3.1) is the Rankin’s method in the following section.
74 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS
3.3.3 Rankin’s method
Assume k = l + j for k, l, j ∈ N. Suppose χ
1
, χ
2
: (Z/NZ)
→ C
are
multiplicative characters. Let
f =
+∞
¸
n=1
a
n
q
n
∈ S
k
(N, χ
−1
1
), g =
+∞
¸
n=0
b
n
q
n
∈ M
l
(N, χ
2
).
So
f(γz) = χ
−1
1
(d)(cz + d)
k
f(z), g(γz) = χ
2
(d)(cz + d)
l
g(z).
Let
G
j,χ
1
χ
2
,s
(z) =
1
2
Γ(j)
(−2πi)
j
t
¸
N[m
(N,n)=1
χ
1
χ
2
(n)y
s+1−k
(mz + n)
j
[ mz + n [
2(s+1−k)
=
Γ(j)
(−2πi)
j
L(χ
1
χ
2
, j + 2(s + 1 −k))
¸
γ=
a b
c d
∈Γ∞\Γ
0
(N)
χ
1
χ
2
(d)
(cz + d)
j
Im(γz)
s+1−k
.
We have
Proposition 3.3.4.
D(f, g, s) = L(χ
1
χ
2
, j + 2(s + 1 −k))
+∞
¸
n=1
¯ a
n
b
n
n
s
=
(4π)
s
Γ(s)
(−2πi)
j
Γ(j)
'f, gG
j
, χ
1
χ
2
, s` [SL
2
(Z) : Γ
0
(N)].
Proof. Using the Fourier expansion, then
+∞
¸
n=1
¯ a
n
b
n
n
s
=
Γ(s)
(4π)
s
+∞
0
1
0
f(z)g(z) dx y
s
dy
y
=
Γ(s)
(4π)
s
Γ∞\1
f(z)g(z)y
s+1
dxdy
y
2
=
Γ(s)
(4π)
s
Γ
0
(N)\1
¸
γ∈Γ∞\Γ
0
(N)
(f(γz)g(γz) Im(γz)
s+1
)
dxdy
y
2
=
Γ(s)
(4π)
s
Γ
0
(N)\1
f(z)
g(z)
¸
γ∈Γ∞\Γ
0
(N)
χ
1
χ
2
(d)
(cz + d)
j
Im(γz)
s+1−k
y
k
dxdy
y
2
,
this implies the Proposition.
3.3. ALGEBRAICITY OF SPECIAL VALUES OF LFUNCTIONS 75
Theorem 3.3.5. (i) D(f, g, s) admits a meromorphic continuation to C,
which is holomorphic outside a simple pole at s = k if l = k and χ
1
χ
2
= 1
(ii) if f is primitive, g ∈ M
l
(N, χ
2
,
¯
Q), then
D(f, g, k −1) ∈
¯
Q π
j+k−1
'f, f`.
Proof. As D(f, g, s) = 'f, gG
s
`,
(i) we have to prove the same statement for G
s
, which can be done by
computing its Fourier extension. The pole comes from the constant Fourier
coeﬃcients.
(ii) For the case N = 1,χ
1
= χ
2
= 1 and j ≥ 3, then G
j,χ
1
χ
2
,k−1
= G
j
, we
are reduced to prove
'f, gG
j
` ∈
¯
Q'f, f`.
Let f
i
, i ∈ I be a basis of S
k
(1) of primitive forms, with f
1
= f. As
gG
j
∈ M
k
(1,
¯
Q), we can write gG
j
= λ
0
G
k
+
¸
i
λ
i
f
i
, with λ
i
∈
¯
Q. Since
'G
k
, f` = 0, 'f, f
j
` = 0, if j = 1,
Then 'f, gG
j
` = λ
1
'f, f`.
Remark. The general case can be treated in the same way, once we prove
that
G
j,χ
1
χ
2
,k−1
∈ M
j
(N, χ
1
χ
2
,
¯
Q) (if j = 2 or χ
1
χ
2
= 1).
Proposition 3.3.6. If
+∞
¸
n=1
¯ a
n
n
s
=
¸
n∈Z[
1
N
]
×
¯ a
n
n
s
¸
pN
1
(1 −α
p
p
−s
)(1 −β
p
p
−s
)
, α
p
β
p
= χ
1
(p)p
k−1
,
+∞
¸
n=1
b
n
n
s
=
¸
n∈Z[
1
N
]
×
b
n
n
s
¸
pN
1
(1 −γ
p
p
−s
)(1 −δ
p
p
−s
)
, γ
p
δ
p
= χ
2
(p)p
l−1
,
then D(f, g, s) =
¸
n∈Z[
1
N
]
×
¯ a
n
b
n
n
s
¸
pN
1
(1 −α
p
γ
p
p
−s
)(1 −β
p
γ
p
p
−s
)(1 −α
p
δ
p
p
−s
)(1 −β
p
δ
p
p
−s
)
.
76 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS
Proof. Exercice, noting that
¯ a
p
r =
α
r+1
p
−β
r+1
p
α
p
−β
p
, b
p
r =
γ
r+1
p
−δ
r+1
p
γ
p
−δ
p
.
We give one application here:
Corollary 3.3.7. The claim (3.1) holds, i.e.
r
k−2
(f)r
l
(f) ∈
¯
Q'f, f`, for l odd.
Proof. Let f ∈ S
k
(1) be primitive, k given. For l even, let g = G
l
, then
+∞
¸
n=1
b
n
n
s
=
¸
p
1
(1 −p
−s
)(1 −p
l−s−1
)
,
hence D(f, G
l
, s) = L(f, s)L(f, s −l + 1). Therefore
L(f, k −1)L(f, k −l) ∈
¯
Q π
j+k−1
'f, f`
which implies
r
k−2
(f)r
k−l−1
(f) ∈
¯
Q'f, f`.
Remark. In the general case,
L(G
j
, χ
1
χ
2
, k −1, s) ∼ ζ(s)L(χ
1
χ
2
, s −l + 1).
If f
1
, f
2
, , f
n
are primitive forms ∈ S
k
(N
i
) for N
i
[ N. Write
L(f
i
, s) = ∗
¸
pN
1
(1 −α
(i)
p,1
p
−s
)(1 −α
(i)
p,2
p
−s
)
,
then
L(f
1
⊗ ⊗f
n
, s) = ∗
¸
pN
1
¸
j
1
,j
2
, ,jn∈1,2¦
(1 −α
(1)
p,j
1
α
(n)
p,jn
p
−s
)
.
One has the following conjecture:
Conjecture 3.3.8 (Part of Langlands Program). L(f
1
⊗ ⊗f
n
, s) has a
meromorphic continuation to C, and is holomorphic if f
i
=
¯
f
j
, for all i = j.
Remark. Rankin’s method implies the above conjecture is OK for n = 2.
The case for n = 3 is due to Paul Garrett. The case for n ≥ 4 is still open.
3.4. PADIC LFUNCTIONS OF MODULAR FORMS 77
3.4 padic Lfunctions of modular forms
In the following, we assume f ∈ S
k
(N) is primitive.
Deﬁnition 3.4.1. φ
+
(x) =
1
2
(φ(x) + φ(−x)), φ
−
(x) =
1
2
(φ(x) − φ(−x)).
Then
˜
Λ(f, φ, j) =
Λ(f, φ
+
, j)
Ω
(−1)
j
f
+
Λ(f, φ
−
, j)
Ω
(−1)
j+1
f
∈
¯
Q
if φ : Z →
¯
Q and 1 ≤ j ≤ k −1.
Fix an embedding
¯
Q →
¯
Q
p
. The function L(f, s) has an Euler product
L(f, s) =
¸
prime
1
E
(s)
, E
(s) ∈
¯
Q[
−s
], deg E
(s) ≤ 2.
Write E
p
(s) = (1 − αp
−s
)(1 − βp
−s
) and assume α = 0. Then β = 0 if and
only if p [ N. Set
f
α
(z) = f(z) −βf(pz).
Lemma 3.4.2. f
α
[
k
U
p
= αf
α
in all cases.
Proof. It is clear if p [ N as in the case β = 0. If p N, then
α + β = a
p
, αβ = p
k−1
.
and f[
k
T
p
= (α + β)f, thus
f
α
[
k
U
p
−αf
α
=
1
p
p−1
¸
i=0
f
z + i
p
−βf(z + i) −αf(z) + αβf(pz)
=−(α + β)f(z) + f
k
T
p
= 0.
If we write f
α
=
¸
+∞
n=1
b
n
q
n
, the above lemma implies that b
np
= αb
n
for
all n. Deﬁne b
n
for n ∈ Z
1
p
as
b
n
= α
−r
b
p
r
n
, r 0.
78 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS
Take φ ∈ LC
c
(Q
p
,
¯
Q) a locally constant function with compact support and
let
L(f, φ, s) =
¸
n∈Z[
1
p
]
φ(n)
a
n
n
s
.
If φ has support in p
−r
Z
p
, then φ(x) = φ
0
(p
r
x) for φ
0
: Z →
¯
Q constant mod
p
m
Z for some m.Then
L(f, φ, s) = α
−r
p
rs
L(f, φ
0
, s)
which implies
˜
Λ(f, φ, j) ∈
¯
Q ⊂
¯
Q
p
, for all φ ∈ LC
c
(Q
p
,
¯
Q).
Deﬁnition 3.4.3. Assume φ ∈ LC
c
(Q
p
,
¯
Q) and φ is constant modulo p
n
Z.
The discrete Fourier transform of φ is
ˆ
φ(x) = p
−m
¸
y mod p
m
φ(y)e
−2πixy
,
for m ≥ n − v
p
(x), where xy ∈ Q
p
→ Q
p
/Z
p
→ Q/Z. This deﬁnition does
not depend on the choice of m ≥ n −v
p
(x).
Exercise. (i)
ˆ
φ is constant mod p
m
Z
p
if and only if φ has support in p
−m
Z
p
.
(ii)
ˆ
ˆ
φ(x) = φ(−x).
(iii) For a ∈ Q
p
, let φ
a
(x) = φ(ax), then
´
φ
a
(x) = p
vp(a)
ˆ
φ
x
a
.
Theorem 3.4.4. (i) There exists a unique µ
f,α
: LP
[0,k−2]
(Z
p
,
¯
Q
p
) →
¯
Q
p
,
such that for all φ ∈ LC(Z
p
,
¯
Q),
Zp
φ(x)x
j−1
µ
f,α
=
˜
Λ(f
α
,
ˆ
φ, j), 1 ≤ j ≤ k −1.
Moreover, ψ(µ
f,α
) =
1
α
µ
f,α
, or equivalently
pZp
φ
x
p
µ
f,α
=
1
α
Zp
φµ
f,α
.
(ii) if v
p
(α) < k −1, then µ
f,α
extends uniquely as an element of T
vp(α)
.
3.4. PADIC LFUNCTIONS OF MODULAR FORMS 79
Proof. (i) The existence of µ
f,α
: LP
[0,k−2]
(Z
p
,
¯
Q
p
) →
¯
Q
p
is just the linearity
of φ →
ˆ
φ. The uniqueness is trivial. The second claim follows from
pZp
φ
x
p
x
p
j−1
µ
f,α
=
1
p
j−1
˜
Λ(f
α
, p
−1
ˆ
φ(px), j)
=
1
α
˜
Λ(f
α
,
ˆ
φ, j) =
1
α
Zp
φ(x)x
j−1
µ
f,α
.
(ii) One needs to show there exists a constant C, such that
v
p
(
a+p
n
Zp
(x −a)
j
µ
f,α
) ≥ C + (j −v
p
(α))n,
for all a ∈ Z
p
, n ∈ N, j ≤ k −2. Note that
´
1
a+p
n
Zp
(x) =
p
−n
e
−2πiax
, if x ∈ p
−n
Z
p
,
0, if not.
= p
−n
φ
a
(p
n
x)
for
φ
a
(x) =
e
2πi
ax
p
n
x ∈ Z
p
,
0, otherwise.
Then
a+p
n
Zp
(x −a)
j
µ
f,α
=
j
¸
l=0
(−a)
l
j
l
p
−n
˜
Λ(f
α
, φ
a
(p
n
x), l + 1)
=α
−n
j
¸
l=0
(−1)
l
j
l
p
nl
˜
Λ(f
α
, φ
a
, l + 1).
Since
p
nl
˜
Λ(f
α
, φ
a
, l + 1) = p
nl
i∞
0
f
α
(z −
a
p
n
)z
l
dz =
i∞
−
a
p
n
f
α
(z)(p
n
z + a)
l
dz,
we get
j
¸
l=0
(−1)
l
j
l
a+p
n
Zp
(x −a)
j
µ
f,α
= α
−n
p
nj
i∞
−
a
p
n
f
α
(z)z
j
dz ∈ α
−n
p
nj
L
fα
.
We just pick C = min(v
p
(˜ r
j
(f
α
[
k
δ))).
80 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS
Remark. (i) If p [ N, then β = 0, and α = 0 implies v
p(α)
=
k−2
2
< k − 1,
hence µ
f,α
exists by the above Theorem.
(ii) If p N, then v
p
(α), v
p
(β) ≥ 0. Since v
p
(α) + v
p
(β) = k −1, at least
one of µ
f,α
or µ
f,β
always exists.
In the case v
p
(α) = k −1, then α+β = a
p
(f) is a unit. This case is called
the ordinary case. The conditions are not strong enough for the uniqueness
of µ
f,α
, as we can add the (k −1)th derivative of any λ ∈ T
0
.
(iii) In the case α = β = 0, we do not understand what happens.
Deﬁnition 3.4.5. Let χ : Z
∗
p
→C
∗
p
be a continuous character. Set
L
p,α
(f ⊗χ, s) =
Z
∗
p
x
−1
χ(x)µ
f,α
.
In particular, take χ(x) = x
k
2
'x`
s−
k
2
where 'x`
t
= exp(t log x). Set
L
p,α
(f, s) =
Z
∗
p
x
k
2
−1
'x`
s−
k
2
µ
f,α
.
Proposition 3.4.6. For 1 ≤ j ≤ k −1,
L
p,α
(f ⊗χ
j
) = (1 −
p
j−1
α
)(1 −
β
p
j
)
˜
Λ(f, j).
Proof. Follows from
(i)
´
1
Z
∗
p
= 1
Zp
−p
−1
1
p
−1
Zp
,
(ii)
˜
Λ(f
α
, 1
Zp
, j) = (1 −
β
p
j
˜
Λ(f, j)),
(iii) ψ(µ
f,α
) =
1
α
µ
f,α
.
Remark. (i) As Λ(f, s) = Λ(f, k −s) and αβ = p
k−1
if p N, then
(1 −
p
j−1
α
) = 1 −
β
p
k−j
.
Note E
p
(f, s) = (1 −αp
−s
)(1 −βp
−s
). Then the Euler factor of the padic L
function is actually the product of one part of the Euler factor for L(f, s) and
one part of the Euler factor for L(f, k −s). This is a general phenomenon.
(ii) If p [ N, α = 0, the v
p
(α) =
k−2
2
. It can happen that α = p
k−2
2
, which
means L
p,α
(f,
k
2
) = 0. In this case
3.4. PADIC LFUNCTIONS OF MODULAR FORMS 81
Conjecture 3.4.7 (MazurTateTeitelbaum Conjecture).
L
t
p,α
(f,
k
2
) = L
Font.
(f)
˜
Λ(f,
k
2
).
Here the padic Lfunction is related to 2dimensional (ϕ, N)ﬁltered modules
D with N = 0 and Fil
0
D = D, Fil
1
D = D. For the pair (λ, α) as in
Fontaine’s course, where λ is the eigenvalue of ϕ and α is the parameter
associated to the ﬁltration, λ is our α and α is our L
Font.
.
The conjecture is proved by KatoKuriharaTsuji, PerrinRiou, and Stevens,
Orton, Emerton with other deﬁnitions of the Linvariant.
(iii) Mazur, Tate and Teitelbaum have also formulated a padic analog of
the BSD conjecture. For E/Q an elliptic curve, by TaniyamaWeil, it is
associated to a primitive form f ∈ S
2
(N). Set L
p,α
(E, s) = L
p,α
(f, s) if it
exists, which is the case if E has either good reduction (hence p N) or
multiplicative reduction (hence p [ N, p
2
N) mod p.
Conjecture 3.4.8 (padic BSD Conjecture).
ord
s=1
L
p,α
(E, s) =
rank E(Q), if p N or α = 1;
rank E(Q) + 1, if p [ N and α = 1.
Kato showed that
ord
s=1
L
p,α
(E, s) ≥
rank E(Q), if p N or α = 1;
rank E(Q) + 1, if p [ N and α = 1.
(iv) To prove Kato or KatoKuriharaTsuji, we need another construction
of padic Lfunctions via Iwasawa theory and (ϕ, Γ)modules; this construc
tion is the subject of the next part of the course and is based on ideas of
PerrinRiou.
82 CHAPTER 3. PADIC LFUNCTIONS OF MODULAR FORMS
Part II
Fontaine’s rings and Iwasawa
theory
83
Chapter 4
Preliminaries
4.1 Some of Fontaine’s rings
This section is a review of notations and results from Fontaine’s course. For
details, see Fontaine’s notes.
4.1.1 Rings of characteristic p
(1) C
p
is the completion of Q
p
for the valuation v
p
with v
p
(p) = 1.
a = ¦x ∈ C
p
, v
p
(x)
1
p
¦.
(2)
˜
E
+
is the ring R in Fontaine’s course. By deﬁnition
˜
E
+
:= ¦x = (x
n
)
n∈N
[ x
n
∈ C
p
/a, x
p
n+1
= x
n
, ∀n¦
is a ring of characteristic p with an action of G
Qp
. For x = (x
n
) ∈
˜
E
+
, for
every x
n
, pick a lifting ˆ x
n
∈ O
Cp
, then
lim
k→+∞
(ˆ x
n+k
)
p
k
:= x
(n)
∈ O
Cp
is a canonical lifting of x
n
such that
˜
E
+
= ¦x = (x
(n)
)
n∈N
[ x
(n)
∈ O
Cp
, (x
(n+1)
)
p
= x
(n)
, ∀n¦
with the addition and multiplication by
(x + y)
(n)
= lim
k→+∞
(x
(n+k)
+ y
(n+k)
)
p
k
, (xy)
(n)
= x
(n)
y
(n)
.
85
86 CHAPTER 4. PRELIMINARIES
˜
E
+
is a valuation ring with valuation
v
E
(x) = v
p
(x
(0)
)
and maximal ideal
m
˜
E
+ = ¦x ∈
˜
E
+
, v
E
(x) > 0¦.
(3) Choose once for all
ε = (1, ε
(1)
, ) ∈
˜
E
+
, ε
(1)
= 1.
Then ε
(n)
is a primitive p
n
th root of 1 for all n. Set
¯ π = ε −1 ∈
˜
E
+
.
We know that v
E
(¯ π) =
p
p−1
> 0.
From now on, χ : G
Qp
→Z
∗
p
will be the cyclotomic character. The action
of G
Qp
on ε is given by
g(ε) = ε
χ(g)
=
+∞
¸
k=0
χ(g)
k
¯ π
k
.
(4) In the following, without further speciﬁcation, K ⊆ Q
p
will be a ﬁnite
extension of Q
p
. Denote by k = k
K
its residue ﬁeld. Set
K
n
= K(ε
(n)
), K
∞
=
¸
n∈N
K
n
.
Set
F ⊆ K= the maximal unramiﬁed extension of Q
p
inside K,
F
t
⊆ K
∞
= the maximal unramiﬁed extension of Q
p
inside K
∞
.
Set
G
K
= Gal(
¯
Q
p
/K), H
K
= Ker χ = Gal(
¯
Q
p
/K
∞
),
and
Γ
K
= G
K
/H
K
= Gal(K
∞
/K)
χ
→Z
∗
p
.
(5) For every K, let
˜
E
+
K
:= ¦x = (x
n
) ∈
˜
E
+
, x
n
∈ O
K∞
/a, ∀n¦ = (
˜
E
+
)
H
K
(by AxSenTate’s Theorem),
E
+
K
:= ¦x = (x
n
) ∈
˜
E
+
, x
n
∈ O
Kn
/a, ∀n ≥ n(K)¦.
4.1. SOME OF FONTAINE’S RINGS 87
Then
¯ π ∈ E
+
K
⊆
˜
E
+
K
⊆
˜
E
+
, ∀K.
We set
E
K
:= E
+
K
[¯ π
−1
] ⊆
˜
E
K
:=
˜
E
+
K
[¯ π
−1
] ⊆
˜
E =
˜
E
+
[¯ π
−1
] = Fr R
with valuation
v
E
(¯ π
−k
x) = v
E
(x) −kv
E
(¯ π).
The following Theorem is the topics in the last section of Chapter 2 of
Fontaine’s Notes.
Theorem 4.1.1. (i)
˜
E is a ﬁeld complete for v
E
with residue ﬁeld
¯
F
p
, ring
of integers
˜
E
+
and G
Qp
acts continuously with respect to v
E
.
(ii) E
F
= k
F
((¯ π)) if F/Q
p
is unramiﬁed.
In general, E
K
is a totally ramiﬁed extension of E
F
of degree [K
∞
: F
t
∞
],
thus a local ﬁeld of characteristic p, with ring of integers E
+
K
and residue ﬁeld
k
F
.
(iii) E =
¸
[K:Qp]<+∞
E
K
is a separable closure of E
Qp
, is stable under G
Qp
and Gal(E/E
K
) = H
K
. So H
Qp
acts continuously on E for the discrete
topology.
(iv)
˜
E (resp.
˜
E
K
) is the completion of the radical closure of E (resp.
˜
E
K
), i.e.,
¸
n∈N
E
p−n
(resp.
¸
n∈N
E
p−n
K
). In particular,
˜
E is algebraically closed.
4.1.2 Rings of characteristic 0
(6) Set
˜
A
+
:= W(
˜
E
+
) = W(R),
˜
A := W(
˜
E) = W(Fr R).
Every element x ∈
˜
A can be written as
x =
+∞
¸
k=0
p
k
[x
k
]
while x
k
∈
˜
E and [x
k
] is its Teichm¨ uller representative.
As we know from the construction of Witt rings, there are bijections
˜
A
+
∼
= (
˜
E
+
)
N
,
˜
A
∼
= (
˜
E)
N
.
88 CHAPTER 4. PRELIMINARIES
There are two topologies in
˜
A
+
and
˜
A:
(i) Strong topology or padic topology: topology by using the above bijec
tion and the discrete topology on
˜
E
+
or
˜
E. A basis of neighborhoods of 0
are the p
k
˜
A, k ∈ N.
(ii) Weak topology: topology deﬁned by v
E
. A basis of neighborhoods of
0 are the p
k
˜
A + [¯ π
n
]A
+
, k, n ∈ N.
The commuting actions of G
Qp
and ϕ on
˜
A are given by
g(
+∞
¸
k=0
p
k
[x
k
]) =
+∞
¸
k=0
p
k
[g(x
k
)], ϕ(
+∞
¸
k=0
p
k
[x
k
]) =
+∞
¸
k=0
p
k
[x
p
k
].
(7)
˜
B :=
˜
A[
1
p
] is the fraction ﬁeld of
˜
A.
˜
B is complete for the valuation v
p
,
its ring of integers is
˜
A and its residue ﬁeld is
˜
E.
For the G
Qp
and ϕactions,
˜
A
ϕ=1
= Z
p
,
˜
B
ϕ=1
= Q
p
,
˜
A
H
K
= W(
˜
E
K
) :=
˜
A
K
,
˜
B
H
K
=
˜
A
K
[
1
p
] :=
˜
B
K
.
(8) Set
π = [ε] −1, t = log[ε] = log(1 + π).
The element [ε] is the padic analogue of e
2πi
. The G
Qp
 and ϕ actions are
given by
ϕ(π + 1) = (π + 1)
p
, ϕ(π + 1) = (π + 1)
χ(g)
.
(9) Set
A
+
Qp
:= Z
p
[[π]] →
˜
A
+
which is stable under ϕ and G
Qp
. Set
A
Qp
:=
Z
p
[[π]][
1
π
] →
˜
A
while´stands for completion under the strong topology, thus
A
Qp
= ¦
¸
k∈Z
a
k
π
−k
[ a
k
∈ Z
p
, lim
k→−∞
v
p
(a
k
) = +∞¦.
Set B
Qp
:= A
Qp
[
1
p
], then B
Qp
is a ﬁeld complete for the valuation v
p
, with
ring of integers A
Qp
and residue ﬁeld E
Qp
.
4.2. (ϕ, Γ)MODULES AND GALOIS REPRESENTATIONS. 89
Moreover, if [K : Q
p
] < +∞,
˜
B contains a unique extension B
K
of B
Qp
whose residue ﬁeld is E
K
, and A
K
= B
K
∩
˜
A is the ring of integers. By
uniqueness, B
K
is stable under ϕ and G
K
acting through Γ
K
.
The ﬁeld
c
ur
=
¸
[K:Qp]<+∞
B
K
is the maximal unramiﬁed extension of B
Qp
= c. Set
B =
´
c
ur
be the closure of
¸
[K:Qp]<+∞
B
K
in
˜
B for the strong topology. Then A = B∩
˜
A
is the ring of integers O
d
c
ur
and the residue ﬁeld of B is A/pA = E. By
AxSenTate,
B
H
K
= B
K
, A
H
K
= A
K
.
Remark. If ¯ π
K
is a uniformising parameter of E
K
, let π
K
∈ A
K
be any
lifting. Then
A
K
= ¦
¸
k∈Z
a
k
π
k
K
[a
k
∈ O
F
, lim
k→−∞
v
p
(a
k
) = +∞¦.
Remark. In the above construction, the correspondence Λ −→
˜
Λ is obtained
by making ϕ bijective and then complete, where Λ = (E
K
, E, A
K
, A, B
K
, B).
4.2 (ϕ, Γ)modules and Galois representations.
Let K be a ﬁxed ﬁnite extension over Q
p
, let Γ = Γ
K
.
Deﬁnition 4.2.1. (i) A (ϕ, Γ)module over A
K
is a ﬁnitely generated A
K

module with semilinear continuous (for the weak topology) and commuting
actions of ϕ and Γ.
A (ϕ, Γ)module over B
K
is a ﬁnite dimensional B
K
vector space with
semilinear continuous (for the weak topology) and commuting actions of ϕ
and Γ.
(ii) A (ϕ, Γ)module D/A
K
is ´etale (or of slope 0) if ϕ(D) generates D as
an A
K
module.
A (ϕ, Γ)module D/B
K
is ´etale (or of slope 0) if it has an A
K
lattice
which is ´etale, equivalently, there exists a basis ¦e
1
, , e
d
¦ over B
K
, such
that the matrix of ϕ(e
1
), , ϕ(e
d
) in e
1
, , e
d
is inside GL
d
(A
K
).
90 CHAPTER 4. PRELIMINARIES
The following theorem is similar to Theorem 1.5.9 in '1.5.4 of Fontaine’s
Notes.
Theorem 4.2.2. The correspondence
V −→D(V ) := (A ⊗
Zp
V )
H
K
is an equivalence of ⊗ categories from the category of Z
p
representations
(resp. Q
p
resp) of G
K
to the category of ´etale (ϕ, Γ)modules over A
K
(resp.
B
K
), and the Inverse functor is
D −→V (D) = (A ⊗
A
K
D)
ϕ=1
.
Remark. (i) Γ
K
is essentially procyclic, so a (ϕ, Γ)module is given by two
operators and commuting relations between them. For example, if D/A
K
is
free of rank d, let U be the matrix of γ for 'γ` = Γ
K
, let P be the matrix of
ϕ, then
Uγ(P) = Pϕ(U), U, P ∈ GL
d
(A
K
).
(ii) We want to recover from D(V ) the known invariants of V :
 H
i
(G
K
, V ); we shall do so in the coming lectures. We will also recover
the Iwasawa modules attached to V and thus give another construction
of padic Lfunctions.
 D
dR
(V ), D
cris
(V ), D
st
(V ).
Chapter 5
(ϕ, Γ)modules and Galois
cohomology
5.1 Galois Cohomology
Let M be a topological Z
p
module (e.g. a ﬁnite module with discrete topology
or a ﬁnitely generated Z
p
module with padic topology, or a Fontaine’s ring
B
+
dR
), with a continuous action of G
K
.
Let H
i
(G
K
, M) be the ith cohomology groups of M of continuous coho
mology. Then:
H
0
(G
K
, M) = M
G
K
= ¦x ∈ M : (g −1)x = 0 ∀g ∈ G
K
¦;
H
1
(G
K
, M) =
¦c : G
K
→M continuous, g
1
c
g
2
−c
g
1
g
2
+ c
g
1
= 0, ∀g
1
, g
2
∈ G
K
¦
¦c : g →(g −1)x, for some x ∈ M¦
To a 1cocycle c, we associate a G
K
module E
c
such that
0 →M →E
c
→N →0
where E
c
· Z
p
M as a Z
p
module and G
K
acts on E
c
by
g(a, m) = (a, gm + c
g
).
One can check easily
g
1
(g
2
(a, m)) = g
1
(a, g
2
m + c
g
2
) = (a, g
1
g
2
m + g
1
c
g
2
+ c
g
1
) = g
1
g
2
(a, m).
E
c
is trivial if and only if there exists
ˆ
1 ∈ E
c
, such that g
ˆ
1 =
ˆ
1 for all g, i.e.
ˆ
1 = (1, x), g
ˆ
1−
ˆ
1 = (0, gx−x+c
g
) = 0, that is, c
g
= (1−g)x is a coboundary.
91
92 CHAPTER 5. (ϕ, Γ)MODULES AND GALOIS COHOMOLOGY
Theorem 5.1.1 (Tate’s Local Duality Theorem). Suppose K is a ﬁnite
extension of Q
p
. Let M be a Z
p
[G
K
]module of ﬁnite length. Then:
(i) H
i
(G
K
, M) = 0 for i ≥ 3; H
i
(G
K
, M) is ﬁnite if i ≤ 2.
(ii)
2
¸
i=0
[H
i
(G
K
, M)[
(−1)
i
= [M[
−[K:Qp]
;
(iii) H
2−i
(G
K
, Hom(M, µ
p
∞)) · Hom(H
i
(G
K
, M), Q
p
/Z
p
).
We will give a proof using (ϕ, Γ)module (Herr’s thesis).
Remark. (i) If M is a ﬁnitely generated Z
p
module with padic topology,
then M · lim
←−
M/p
n
M, and H
i
(G
K
, M) · lim
←−
H
i
(G
K
, M/p
n
M).
Z! Not tautological, the proof uses ﬁniteness of (i) to ensure MittagLeﬄer
conditions.
(ii) If V is a Q
p
representation of G
K
, let T ⊂ V be a Z
p
lattice stable
by G
K
. Then H
i
(G
K
, V ) · Q
p
⊗H
i
(G
K
, T).
Corollary 5.1.2. If V is a Q
p
representation of G
K
. Then:
(i)
2
¸
i=0
(−1)
i
dim
Qp
H
i
(G
K
, V ) = −[K : Q
p
] dim
Qp
V ;
(ii) H
2
(G
K
, V ) = H
0
(G
K
, V
∗
(1))
∗
.
5.2 The complex C
ϕ,γ
(K, V )
Assume that Γ
K
is procyclic (Γ
Qp
· Z
∗
p
), γ is a topological generator of Γ
K
.
This assumption is automatic if p ≥ 3, or if K ⊃ Q(µ
4
) when p = 2. Let V
be a Z
p
 or Q
p
representation of G
K
. Set
D(V ) = (A ⊗
Zp
V )
H
K
.
Deﬁnition 5.2.1. The complex C
•
ϕ,γ
(K, V ) = C
ϕ,γ
(K, V ) is
0 →D(V )
(ϕ−1, γ−1)
−−−−−−→D(V ) ⊕D(V )
(γ−1) pr
1
−(ϕ−1) pr
2
−−−−−−−−−−−−→D(V ) →0.
It is easy to see C
ϕ,γ
(K, V ) is really a complex (as ϕ, γ commute to each
other). We shall denote the complex by C
•
(V ) if no confusion is caused. We
5.2. THE COMPLEX C
ϕ,γ
(K, V ) 93
have
H
0
(C
•
(V )) = ¦x ∈ D(V ), γ(x) = x, ϕ(x) = x¦,
H
1
(C
•
(V )) =
¦(x, y) : (γ −1)x = (ϕ −1)y¦
¦((ϕ −1)z, (γ −1)z) : z ∈ D(V )¦
,
H
2
(C
•
(V )) =
D(V )
(γ −1, ϕ −1)
,
H
i
(C
•
(V )) = 0, for i ≥ 3.
Theorem 5.2.2. H
i
(C
ϕ,γ
(K, V )) · H
i
(G
K
, V ) for all i in N.
Proof. We have the following exact sequence (which can be proved by reduc
ing mod p):
0 →Z
p
→A
ϕ−1
−−→A →0,
here A = O
d
c
ur
in Fontaine’s course.
(1) i = 0: For x ∈ D(V )
ϕ=1
, since D(V ) = (A ⊗
Zp
V )
H
K
, we have
D(V )
ϕ=1
= (A
ϕ=1
⊗
Zp
V )
H
K
= V
H
K
, and (V
H
K
)
γ=1
= V
G
K
.
(2) i = 1: Let (x, y) satisfy the condition (γ − 1)x = (ϕ − 1)y. Choose
b ∈ (A ⊗
Zp
V )
H
K
, (ϕ −1)b = x. We deﬁne the map:
g ∈ G
K
→c
x,y
(g) =
g −1
γ −1
y −(g −1)b.
while the meaning of
g−1
γ−1
y is: as χ(g) = lim
i→+∞
χ(γ)
n
i
, y is ﬁxed by H
K
, we
let
g −1
γ −1
y = lim
i→+∞
(1 + γ + + γ
n
i
−1
)y.
This is a cocycle with values in V , because g →(g −1)(
y
γ−1
−b) is a cocycle,
and (ϕ − 1)c
xy
(g) = (g − 1)x − (ϕ − 1)(g − 1)b = 0, which implies that
c
xy
(g) ∈ D(V )
ϕ=1
= V .
Injectivity: If c
xy
= 0 in H
1
(G
K
, V ), then there exists z ∈ V , c
xy
(g) =
(g − 1)z for all g ∈ G
K
, that is,
g−1
γ−1
y = (g − 1)(b − z) for all g. Now
b−z ∈ D(V ), because it is ﬁxed by g ∈ H
K
. Then we have: y = (γ−1)(b−z)
and x = (ϕ −1)(b −z), hence (x, y) equal to 0 in H
1
(C
•
(V )).
Surjectivity: If c ∈ H
1
(G
K
, V ), we have:
0 →V −→E
c
−→Z
p
→0,
94 CHAPTER 5. (ϕ, Γ)MODULES AND GALOIS COHOMOLOGY
here E
c
= Z
p
V , e ∈ E
c
→1 ∈ Z
p
and ge = e + c
g
for g →c
g
representing
c. We have:
0 →D(V ) −→D(E
c
) −→A
K
→0,
here D(E
c
) ⊂ A ⊗E
c
and ˜ e ∈ D(E
c
) →1 ∈ A
K
. Let
x = (ϕ −1)˜ e, y = (γ −1)˜ e,
they are both in D(V ) and satisfy (γ − 1)x = (ϕ − 1)y. Let b = ˜ e − e ∈
A ⊗
Zp
E
c
. Then c
x,y
(g) =
g−1
γ−1
y −(g −1)b = c
g
and (ϕ −1)(b) = x.
(3) i general: from the exact sequence:
0 →Z
p
→A
ϕ−1
−−→A →0,
tensoring with V and taking the cohomology H
i
(H
K
, −), we get
0 →V
H
K
→D(V )
ϕ−1
−−→D(V ) →H
1
(H
K
, V ) →0,
because A ⊗ V · ⊕(A/p
i
) as H
K
modules and H
i
(H
K
, E) = 0, if i ≥ 1, so
H
i
(H
K
, A ⊗V ) = 0 for all i ≥ 1. Hence H
i
(H
K
, V ) = 0 for all i ≥ 1.
By the HochschildSerre Spectral Sequence for
1 →H
K
→G
K
→Γ
K
→1,
we have H
i
(Γ
K
, H
j
(H
K
, V )) ⇒ H
i+j
(G
K
, V ). When j or i ≥ 2, the coho
mology vanishes. So we have:
H
q
(G
K
, V ) = 0, if q ≥ 3
H
2
(G
K
, V ) · H
1
(Γ
K
, H
1
(H
K
, V )).
Since H
1
(H
K
, V ) =
D(V )
ϕ−1
, we get
H
2
(G
K
, V ) ·
D(V )
ϕ −1
(γ −1)
D(V )
ϕ −1
=
D(V )
(ϕ −1, γ −1)
.
Remark. (1) The inﬂationrestriction exact sequence becomes the commu
tative diagram
0
H
1
(Γ
K
, V
H
K
)
H
1
(G
K
, V )
ιϕ,γ
H
1
(H
K
, V
Γ
K
)
0
0
D(V )
ϕ=1
γ−1
H
1
(C
ϕ,γ
(K, V ))
(
D(V )
ϕ−1
)
Γ
K
0
5.3. TATE’S EULERPOINCAR
´
E FORMULA. 95
where the map H
1
(C
ϕ,γ
(K, V )) →(
D(V )
ϕ−1
)
Γ
K
is given by sending (x, y) to the
image of x.
(2) Let γ
t
be another generator of Γ
K
, we have
γ−1
γ
−1
∈ (Z
p
[[Γ
K
]])
∗
and a
commutative diagram:
C
ϕ,γ
: 0
D(V )
γ−1
γ
−1
D(V )
γ−1
γ
−1
⊕ D(V )
Id
D(V )
Id
0
C
ϕ,γ
: 0
D(V )
D(V ) ⊕ D(V )
D(V )
0
It induces a commutative diagram
H
1
(C
•
ϕ,γ
)
ι
γ,γ
l
K
(γ)ιϕ,γ
N
N
N
N
N
N
N
N
N
N
N
H
1
(C
•
ϕ,γ
)
l
K
(γ
)ι
ϕ,γ
p
p
p
p
p
p
p
p
p
p
p
H
1
(G
K
, V )
where l
K
(γ) =
log χ(γ)
p
r(K)
for log χ(Γ
K
) · p
r(K)
Z
p
. So l
K
(γ)c
,ϕγ
“does not depend
on the choice of γ”.
5.3 Tate’s EulerPoincar´e formula.
5.3.1 The operator ψ.
Lemma 5.3.1. (i) ¦1, ε, , ε
p−1
¦ is a basis of E
Qp
over ϕ(E
Qp
);
(ii) ¦1, ε, , ε
p−1
¦ is a basis of E
K
over ϕ(E
K
), for all [K : Q
p
] < +∞;
(iii) ¦1, ε, , ε
p−1
¦ is a basis of E over ϕ(E);
(iv) ¦1, [ε], , [ε]
p−1
¦ is a basis of A over ϕ(A).
Proof. (i) Since E
Qp
= F
p
((¯ π)) with ¯ π = ε −1, we have ϕ(E
Qp
) = F
p
((¯ π
p
));
(ii) Use the following diagram of ﬁelds, note that E
Qp
/ϕ(E
Qp
) is purely
inseparable and ϕ(E
K
)/ϕ(E
Qp
) is separable:
E
K
ϕ(E
K
)
E
Qp
ϕ(E
Qp
)
(iii) Because E = ∪E
K
.
96 CHAPTER 5. (ϕ, Γ)MODULES AND GALOIS COHOMOLOGY
(iv) To show that
(x
0
, x
1
, , x
p−1
) ∈ A
p
∼
−→
p−1
¸
i=0
[ε]
i
ϕ(x
i
) ∈ A
is a bijection, it suﬃces to check it mod p and use (iii).
Deﬁnition 5.3.2. The operator ψ : A →A is deﬁned by
ψ(
p−1
¸
i=0
[ε]
i
ϕ(x
i
)) = x
0
.
Proposition 5.3.3. (i) ψϕ = Id;
(ii) ψ commutes with G
Qp
.
Proof. (i) The ﬁrst statement is obvious.
(ii) Note that
g(
p−1
¸
i=0
[ε]
i
ϕ(x
i
)) =
p−1
¸
i=0
[ε]
iχ(g)
ϕ(g(x
i
)).
If for 1 ≤ i ≤ p −1, write iχ(g) = i
g
+ pj
g
with 1 ≤ i
g
≤ p −1, then
ψ(
p−1
¸
i=0
[ε]
iχ(g)
ϕ(g(x
i
))) = ψ(ϕ(g(x
0
)) +
p−1
¸
i=1
[ε]
ig
ϕ([ε]
jg
g(x
i
))) = g(x
0
).
Corollary 5.3.4. (i) If V is a Z
p
representation of G
K
, there exists a unique
operator ψ : D(V ) →D(V ) with
ψ(ϕ(a)x) = aψ(x), ψ(aϕ(x)) = ψ(a)x
if a ∈ A
K
, x ∈ D(V ) and moreover ψ commute with Γ
K
.
(ii) If D is an ´etale (ϕ, Γ)module over A
K
or B
K
, there exists a unique
operator ψ : D →D with as in (i). Moreover, for any x ∈ D,
x =
p
n
−1
¸
i=0
[ε]
i
ϕ
n
(x
i
)
where x
i
= ψ
n
([ε]
−i
x).
5.3. TATE’S EULERPOINCAR
´
E FORMULA. 97
Proof. (i) The uniqueness follows from A
K
⊗
ϕ(A
K
)
ϕ(D) = D. For the exis
tence, use ψ on A⊗V ⊃ D(V ). D(V ) is stable under ψ because ψ commutes
with H
K
, ψ commutes with Γ
K
since ψ commutes with G
K
.
(ii) D = D(V (D)), thus we have existence and uniqueness of ψ. The rest
is by induction on n.
Example 5.3.5. Let D = A
Qp
⊃ A
+
Qp
= Z
p
[[π]] be the trivial (ϕ, Γ)module,
here [ε] = (1 +π). Then for x = F(π) ∈ A
+
Qp
, ϕ(x) = F((1 +π)
p
−1). Write
F(π) =
p−1
¸
i=0
(1 +π)
i
F
i
((1 + π)
p
−1),
then ψ(F(π)) = F
0
(π). It is easy to see if F(π) belongs to Z
p
[[π]], F
i
(π)
belongs to Z
p
[[π]] for all i. Then ψ(E
+
Qp
) ⊂ E
+
Qp
= F
p
[[π]]. Hence ψ(A
+
Qp
) ⊂
A
+
Qp
. Consequently, ψ is continuous for the weak topology.
Moreover, we have:
ϕ(ψ(F)) = F
0
((1 + π)
p
−1) =
1
p
¸
z
p
=1
p−1
¸
i=0
(z(1 +π))
i
F
i
((z(1 +π))
p
−1)
=
1
p
¸
z
p
=1
F(z(1 +π) −1).
Recall T
0
(Z
p
, Q
p
) · B
+
Qp
= Q
p
⊗
Zp
A
+
Qp
by µ → A
µ
(π) =
Zp
[ε]
x
µ. Recall
that ψ(µ) is deﬁned by
Zp
φ(x)ψ(µ) =
pZp
φ(
x
p
)µ.
From the above formula, we get, using formulas for Amice transforms,
A
ψ(µ)
(π) = ψ(A
µ
)(π).
Proposition 5.3.6. If D is an ´etale ϕmodule over A
K
, then ψ is continuous
for the weak topology.
Proof. As A
K
is a free A
Qp
module of rank [K
∞
: Q
p
(µ
p
∞)], we can assume
K = Q
p
. Choose e
1
, e
2
, , e
d
in D, such that
D = ⊕(A
Qp
/p
n
i
)e
i
, n
i
∈ N ∪ ¦∞¦.
98 CHAPTER 5. (ϕ, Γ)MODULES AND GALOIS COHOMOLOGY
Since D is ´etale, we have D = ⊕(A
Qp
/p
n
i
)ϕ(e
i
). Then we have the following
diagram:
D
ψ
t
D
t
⊕(A
Qp
/p
n
i
)ϕ(e
i
)
⊕(A
Qp
/p
n
i
)e
i
¸
x
i
ϕ(e
i
)
¸
ψ(x
i
)e
i
Now x →ψ(x) is continuous in A
Qp
, hence ψ is continuous in D.
5.3.2 D
ψ=1
and D/(ψ −1)
Lemma 5.3.7. If D is an ´etale ϕmodule over E
Qp
, then:
(i) D
ψ=1
is compact;
(ii) dim
Fp
(D/(ψ −1)) < +∞.
Proof. (i) choose a basis ¦e
1
, , e
d
¦, then ¦ϕ(e
1
), , ϕ(e
d
)¦ is still a basis.
Set v
E
(x) = inf
i
v
E
(x
i
) if x =
¸
i
x
i
ϕ(e
i
), x
i
∈ E
+
Qp
. We have
ψ(x) =
¸
i
ψ(x
i
)e
i
and e
i
=
d
¸
i=1
a
i,j
ϕ(e
j
).
Let c = inf
i,j
v
E
(a
i,j
), then we have
v
E
(ψ(x)) ≥ c + inf
i
v
E
(ψ(x
i
)). (5.1)
From ψ(E
+
Qp
) ⊂ E
+
Qp
and ψ(¯ π
p
k
x) = ¯ π
k
ψ(x), we get v
E
(ψ(x)) ≥
v
E
(x)
p
. So
v
E
(ψ(x)) ≥ c + inf
i
[
v
E
(x
i
)
p
] ≥ c + [
v
E
(x)
p
].
If v
E
(x) <
p(c−1)
p−1
, then v
E
(ψ(x)) > v
E
(x). Now D
ψ=1
is closed since ψ is
continuous, and is a subset of the compact set
M := ¦x : v
E
(x) ≥
p(c −1)
p −1
¦ ⊆
d
¸
i=1
¯ π
k
F
p
[[¯ π]] ϕ(e
i
).
5.3. TATE’S EULERPOINCAR
´
E FORMULA. 99
Hence D
ψ=1
is also compact.
(ii) ψ − 1 is bijective on D/M from the proof of (i). We only need to
prove that M/((ψ −1)D∩M) is ﬁnite, equivalently, that (ψ −1)D contains
¦x : v
E
(x) ≥ c
t
¦ for some c
t
.
ϕ(x
i
) can be written uniquely as ϕ(x
i
) =
d
¸
j=1
b
i,j
e
j
. Let c
0
= inf
i,j
v
E
(b
i,j
),
then
x =
d
¸
i=1
x
i
ϕ(e
i
) =
d
¸
i=1
x
i
d
¸
j=1
b
i,j
e
j
=
d
¸
j=1
(
d
¸
i=1
x
i
b
i,j
)e
j
.
Let y
j
=
d
¸
i=1
x
i
b
i,j
, then x =
d
¸
j=1
y
j
e
j
, and
v
E
(y
j
) ≥ c
0
+ v
E
(x).
From ϕ(x) =
d
¸
j=1
ϕ(y
j
)ϕ(e
j
), we get
v
E
(ϕ(x)) = inf v
E
(ϕ(y
j
)) = p inf v
E
(y
j
) ≥ pv
E
(x) + pc
0
.
So, if v
E
(x) ≥
−pc
0
p−1
+ 1, then v
E
(ϕ
n
(x)) ≥ p
n
. It implies y =
+∞
¸
i=1
ϕ
i
(x)
converges in D. Now
(ψ −1)y =
+∞
¸
i=0
ϕ
i
(x) −
+∞
¸
i=1
ϕ
i
(x) = x
implies that (ψ −1)D contains ¦x[v
E
(x) ≥
−pc
0
p−1
+ 1¦.
Proposition 5.3.8. If D is an ´etale ϕmodule over A
K
(resp. over B
K
),
then:
(i) D
ψ=1
is compact (resp. locally compact);
(ii) D/(ψ −1) is ﬁnitely generated over Z
p
(resp. over Q
p
).
Proof. We can reduce to K = Q
p
. B
K
follows from A
K
by Q
p
⊗
Zp
−. So we
consider D over A
Qp
.
(i) Note that D
ψ=1
= lim
←−
(D/p
n
D)
ψ=1
. From the previous lemma we have
D/p
n
D is compact by easy induction on n. So D
ψ=1
is compact.
(ii) The quotient (D/(ψ−1))/p · (D/p)/ψ −1 is ﬁnite dimensional over
F
p
. We have to check that
100 CHAPTER 5. (ϕ, Γ)MODULES AND GALOIS COHOMOLOGY
if x = (ψ −1)y
n
+ p
n
Z
p
for all n, then x ∈ (ψ −1)D.
If m ≥ n, y
m
−y
n
∈ (D/p
n
)
ψ=1
, which is compact, we can extract a sequence
converging mod p
n
. Thus we can diagonally extract a sequence converging
mod p
n
for all n. Then y
n
converges to y in D and x = (ψ −1)y.
5.3.3 The Γmodule D
ψ=0
.
If p = 2, we let Γ
0
= Γ
Qp
· Z
∗
p
. Let Γ
n
⊆ Γ
0
and Γ
n
· 1 + p
n
Z
p
if n ≥ 1.
Then Γ
0
= ´Γ
1
where ´ = µ
p−1
, and Γ
n
= lim
←−
m
Γ
n
/Γ
n+m
. We deﬁne
Z
p
[[Γ
n
]] = lim
←−
Z
p
[Γ
n
/Γ
n+m
] = T(Γ
n
, Z
p
).
If n ≥ 1, let γ
n
be a topological generator of Γ
n
. So Γ
n
= γ
Zp
n
. The corre
spondence
Z
p
[[Γ
n
]] Z
p
[[T]]
∼
∼
A
+
Qp
γ
n
−1
T
π
is just the Amice transform. Then
Z
p
[[Γ
0
]] = Z
p
[´] ⊗Z
p
[[Γ
1
]],
Z
p
¦¦Γ
n
¦¦ := (Z
p
[[Γ
n
]][(γ
n
−1)
−1
])
∧
· A
Qp
(as a ring),
Z
p
¦¦Γ
0
¦¦ = Z
p
[´] ⊗Z
p
¦¦Γ
1
¦¦.
Modulo p, we get F
p
¦¦Γ
n
¦¦ · E
Qp
as a ring.
Remark. Z
p
[[Γ
0
]] · T
0
(Γ
0
, Z
p
) · T
0
(Z
∗
p
, Z
p
) · (A
+
Qp
)
ψ=0
. So (A
+
Qp
)
ψ=0
is
a free Z
p
[[Γ
0
]]module of rank 1. This a special case of a general theorem
which will come up later on.
Lemma 5.3.9. (i) If M is a topological Z
p
module (M = lim
←−
M/M
i
) with
a continuous action of Γ
n
(i.e. for all i, there exists k, such that Γ
n+k
acts
trivially on M/M
i
), then Z
p
[[Γ
n
]] acts continuously on M;
(ii) If γ
n
−1 has a continuous inverse, then Z
p
¦¦Γ
n
¦¦ also acts continu
ously on M.
5.3. TATE’S EULERPOINCAR
´
E FORMULA. 101
Lemma 5.3.10. (i) If n ≥ 1, v
E
(γ
n
(¯ π) − ¯ π) = p
n
v
E
(¯ π);
(ii) For all x in E
Qp
, we have v
E
(γ
n
(x) −x) ≥ v
E
(x) + (p
n
−1)v
E
(¯ π).
Proof. Since χ(γ) = 1 + p
n
u, u ∈ Z
∗
p
, we have
γ
n
(¯ π) − ¯ π =γ
n
(1 + ¯ π) −(1 + ¯ π) = (1 + ¯ π)((1 + ¯ π)
p
n
u
−1)
=(1 + ¯ π)((1 + ¯ π)
u
−1)
p
n
.
Then we get (i).
In general, for x =
+∞
¸
k=k
0
a
k
¯ π
k
, then v
E
(x) = k
0
v
E
(¯ π). Now
γ
n
(x) −x
γ
n
(¯ π) − ¯ π
=
+∞
¸
k=k
0
a
k
γ
n
(¯ π)
k
− ¯ π
k
γ
n
(¯ π) − ¯ π
,
and
v
E
(
γ
n
(¯ π)
k
− ¯ π
k
γ
n
(¯ π) − ¯ π
) ≥ (k −1)v
E
(¯ π).
Proposition 5.3.11. Let D be an ´etale (ϕ, Γ)module of dimension d over
E
Qp
. Assume n ≥ 1, (i, p) = 1. Then
(i) γ ∈ Γ induces ε
i
ϕ
n
(D) · ε
χ(γ)i
ϕ
n
(D);
(ii) γ
n
−1 admits a continuous inverse on ε
i
ϕ
n
(D). Moreover if ¦e
1
, , e
d
¦
is a basis of D, then:
F
p
¦¦Γ
n
¦¦
d
∼
−→ϕ
n
(D)
(λ
1
, , λ
d
) −→λ
1
∗ ε
i
ϕ
n
(e
1
) + + λ
d
∗ ε
i
ϕ
n
(e
d
)
is a topological isomorphism.
Proof. (i) is obvious. Now, remark that (ii) is true for n + 1 implies (ii) is
true for n, since
ε
i
ϕ
n
(D) = ε
i
ϕ
n
(⊕
p−1
j=0
ε
j
ϕ(D)) = ⊕
p−1
j=0
ε
i+p
n
j
ϕ
n+1
(D),
and for n > 1, γ
n+1
= γ
p
n
, so
1
γn−1
=
1
γ
n+1
−1
(1 +γ
n
+ + γ
p−1
n
), and
F
p
¦¦Γ
n
¦¦ = F
p
¦¦Γ
n+1
¦¦ + + γ
p−1
n
F
p
¦¦Γ
n+1
¦¦.
102 CHAPTER 5. (ϕ, Γ)MODULES AND GALOIS COHOMOLOGY
So we can assume n big enough.
Recall v
E
(x) = inf
i
v
E
(x
i
) if x =
¸
i
x
i
e
i
. We can, in particular, assume
v
E
(γ
n
(e
i
) − e
i
) ≥ 2v
E
(¯ π), it implies v
E
(γ
n
(x) − x) ≥ v
E
(x) + 2v
E
(¯ π) for all
x ∈ D (as v
E
(γ
n
(x) −x) ≥ v
E
(x) + (p
n
−1)v
E
(¯ π) for all x ∈ E
Qp
). Now
χ(γ
n
) = 1 + p
n
u, u ∈ Z
∗
p
,
so
γ
n
(ε
i
ϕ
n
(x)) −ε
i
ϕ
n
(x) = ε
i
(ε
ip
n
u
ϕ
n
(γ
n
(x)) −ϕ
n
(x)) = ε
i
ϕ
n
(ε
iu
γ
n
(x) −x).
So we have to prove x →f(x) = ε
iu
γ
n
(x) −x has a continuous inverse on D,
and D is a F
p
¦¦f¦¦module with basis ¦e
1
, , e
d
¦. Let α = ε
iu
−1; iu ∈ Z
∗
p
,
so v
E
(α) = v
E
(¯ π). Then v
E
(
f
α
(x) −x) ≥ v
E
(x) + v
E
(¯ π). It implies
f
α
has an
inverse
g =
+∞
¸
n=0
(1 −
f
α
)
n
and v
E
(g(x) −x) ≥ v
E
(x) + v
E
(¯ π).
So f has an inverse f
−1
(x) = g(
x
α
) and v
E
(f
−1
(x) −
x
α
) ≥ v
E
(x).
By induction, for all k in Z, we have
v
E
(f
k
(x) −α
k
x) ≥ v
E
(x) + (k + 1)v
E
(¯ π).
Let M = E
+
Qp
e
1
⊕ ⊕E
+
Qp
e
d
, then f
k
induces
M/¯ πM · α
k
M/α
k+1
M · ¯ π
k
M/¯ π
k+1
M.
So f
k
F
p
[[f]]e
1
⊕ ⊕f
k
F
p
[[f]]e
d
is dense in ¯ π
k
M and is equal by compactness.
Corollary 5.3.12. γ −1 has a continuous inverse on D
ψ=0
, and D
ψ=0
is a
free F
p
¦¦Γ
0
¦¦module with basis ¦εϕ(e
1
), , εϕ(e
d
)¦.
Proof. Copy the proof that (ii) for n + 1 implies (ii) for n in the previous
proposition, using γ
1
= γ
p−1
0
.
Proposition 5.3.13. If D is an ´etale (ϕ, Γ)module over A
K
or B
K
, then
γ −1 has a continuous inverse on D
ψ=0
.
5.3. TATE’S EULERPOINCAR
´
E FORMULA. 103
Proof. B
K
follows from A
K
by Q
p
⊗
Zp
; and we can reduce A
K
to A
Qp
.
Since D
ψ=0
→(D/p)
ψ=0
is surjective, (
p−1
¸
i=1
ε
i
ϕ(x
i
) can be lifted to
p−1
¸
i=1
[ε]
i
ϕ(ˆ x
i
)),
so we have the following exact sequence:
0 −→(pD)
ψ=0
−→D
ψ=0
−→(D/p)
ψ=0
−→0.
Everything is complete for the padic topology, so we just have to verify the
result mod p, which is in corollary 5.3.12.
5.3.4 Computation of Galois chomology groups
Proposition 5.3.14. Let C
ψ,γ
be the complex
0 →D(V )
(ψ−1, γ−1)
−−−−−−−→D(V ) ⊕D(V )
(γ−1) pr
1
−(ψ−1) pr
2
−−−−−−−−−−−−→D(V ) →0.
Then we have a commutative diagram of complexes
C
ϕ,γ
: 0
D(V )
Id
D(V )
−ψ
⊕ D(V )
Id
D(V )
−ψ
0
C
ψ,γ
: 0
D(V )
D(V ) ⊕ D(V )
D(V )
0
which induces an isomorphism on cohomology.
Proof. Since (−ψ)(ϕ − 1) = ψ − 1 and ψ commutes with γ (i.e. ψγ = γψ),
the diagram commutes. ψ is surjective, hence the cokernel complex is 0. The
kernel is nothing but
0 −→0 −→D(V )
ψ=0
γ−1
−−→D(V )
ψ=0
−→0,
it has no cohomology by Proposition 5.3.13.
Theorem 5.3.15. If V is a Z
p
or a Q
p
representation of G
K
, then C
ψ,γ
(K, V )
computes the Galois cohomology of V :
(i) H
0
(G
K
, V ) = D(V )
ψ=1,γ=1
= D(V )
ϕ=1,γ=1
.
(ii)H
2
(G
K
, V ) ·
D(V )
(ψ−1,γ−1)
.
(iii) One has an exact sequence
0 −→
D(V )
ψ=1
γ −1
−→H
1
(G
K
, V ) −→
D(V )
ψ −1
γ=1
−→0
(x, y) −→ x
104 CHAPTER 5. (ϕ, Γ)MODULES AND GALOIS COHOMOLOGY
Let ((V ) = (ϕ −1)D
ψ−1
⊂ D
ψ=0
, the exact sequence
0 −→D(V )
ϕ=1
−→D(V )
ψ=1
−→((V ) −→0
induces an exact sequence
0 −→
D(V )
ϕ=1
γ −1
−→
D(V )
ψ=1
γ −1
−→
((V )
γ −1
−→0
since ((V )
γ=1
⊂ (D
ψ=0
)
γ=1
= 0.
Proposition 5.3.16. If D is an ´etale (ϕ, Γ)module of dimension d over
E
Qp
, then ( = (ϕ −1)D
ψ=1
is a free F
p
[[Γ
0
]]module of rank d.
Proof. We know:
• ( ⊂ D
ψ=0
, it implies ( is a F
p
[[Γ
0
]]module of rank less than d;
• ( is compact, because D
ψ=1
is compact;
• So we just have to prove (see proposition 5.3.11 and corollary 5.3.12)
that ( contains ¦εϕ(e
1
), , εϕ(e
d
)¦, where ¦e
1
, , e
d
¦ is any basis
of D over E
Qp
.
Let ¦f
1
, , f
d
¦ be any basis. Then ϕ
n
(¯ π
k
f
i
) goes to 0 when n goes to +∞
if k 0. Let g
i
=
+∞
¸
n=0
ϕ
n
(εϕ(¯ π
k
f
i
)). Then we have:
• ψ(g
i
) = g
i
, because ψ(εϕ(¯ π
k
f
i
)) = 0;
• (ϕ −1)g
i
= −εϕ(¯ π
k
f
i
) ∈ (.
We can take e
i
= ¯ π
k
f
i
.
5.3.5 The EulerPoincar´e formula.
Theorem 5.3.17. If V is a ﬁnite Z
p
representation of G
K
, then
χ(V ) =
2
¸
i=0
[H
i
(G
K
, V )[
(−1)
i
= [V [
−[K:Q
P
]
.
5.4. TATE’S DUALITY AND RESIDUES 105
Proof. From Shapiro’s lemma, we have
H
i
(G
K
, V ) · H
i
(G
Qp
, Ind
G
Qp
G
K
V ).
Since [ Ind
G
Qp
G
K
V [ = [V [
[K:Qp]
, we can assume K = Q
p
. Given an exact se
quence
0 →V
1
→V →V
2
→0,
then χ(V ) = χ(V
1
)χ(V
2
) and [V [ = [V
1
[[V
2
[ from the long exact sequence
in Galois Cohomology, thus we can reduce to the case that V is a F
p

representation of G
K
. Then we have:
[H
0
[ = [D(V )
ϕ=1,γ=1
[;
[H
1
[ = [
D(V )
ϕ=1
γ −1
[ [
((V )
γ −1
[ [
D(V )
ψ −1
γ=1
[;
[H
2
[ = [
D(V )
(ψ −1, γ −1)
[.
So [H
0
[[H
2
[[H
1
[
−1
= [
((V )
γ−1
[
−1
, because D(V )
ϕ=1
and
D(V )
ψ−1
are ﬁnite groups,
and for a ﬁnite group M, the exact sequence:
0 −→M
γ=1
−→M
γ−1
−→M −→
M
γ −1
−→0
implies that [M
γ=1
[ = [
M
γ−1
[. Now
((V )
γ−1
is a (F
p
[[Γ
0
]]/(γ − 1)) = F
p
module
of rank dim
E
Qp
D(V ) = dim
Fp
V . Hence [
((V )
γ−1
[ = [V [.
5.4 Tate’s duality and residues
Let M be a ﬁnite Z
p
module. We want to construct a perfect pairing
H
i
(G
K
, M) H
2−i
(G
K
, M
∧
(1)) −→Q
p
/Z
p
.
By using Shapiro’s lemma, we may assume K = Q
p
.
Deﬁnition 5.4.1. Let x =
¸
k∈Z
a
k
π
k
∈ B
Qp
, deﬁne
res(xdπ) = a
−1
.
The residue of x, denoted by Res(x) is deﬁned as
Res(x) = res(x
dπ
1 + π
).
106 CHAPTER 5. (ϕ, Γ)MODULES AND GALOIS COHOMOLOGY
The map Res : B
Qp
→ Q
p
maps A
Qp
to Z
p
, thus it induced a natural
map B
Qp
/A
Qp
→Q
p
/Z
p
.
Proposition 5.4.2.
Res(ψ(x)) = Res(x);
Res(γ(x)) = χ(γ)
−1
Res(x)
Proof. Exercise.
Let D be an ´etale (ϕ, Γ)module over A
Qp
, denote D
∨
= Hom
A
Qp
(D, B
Qp
/A
Qp
),
let x ∈ D
∨
, y ∈ D, denote
'x, y` = x(y) ∈ B
Qp
/A
Qp
.
Then
'γ(x), γ(y)` = γ('x, y`),
'ϕ(x), ϕ(y)` = ϕ('x, y`)
determines the (ϕ, Γ)module structure on D
∨
. Set
[x, y] := Res('x, y`) ∈ Q
p
/Z
p
.
The main step is following proposition.
Proposition 5.4.3. (i) The map x → (y → [x, y]) gives an isomorphism
from D
∨
to D
∧
(V ) = Hom
cont
(D, Q
p
/Z
p
).
(ii) The following formulas hold:
[x, ϕ(y)] = [ψ(x), y]
[γ(x), y] = χ(γ)
−1
[x, γ
−1
(y)].
Corollary 5.4.4. Let V
∧
(1) = Hom
Zp
(V, (Q
p
/Z
p
)(1)), then D(V
∧
(1)) =
D
∨
(1).
Now the two complexes
C
ϕ,γ
(Q
p
, V ) : D(V )
d
1
D(V ) ⊕D(V )
d
2
D(V )
D
∨
(V ) D
∨
(1) ⊕D
∨
(1)
d
2
D
∨
(1)
d
1
: C
ψ,γ
−1(Q
p
, V
∧
(1))
5.4. TATE’S DUALITY AND RESIDUES 107
are in duality, where d
1
z = ((ϕ−1)z, (γ −1)z), d
2
(x, y) = (γ −1)x−(ϕ−1)y,
d
t
1
z
t
= ((ψ −1)z
t
, (γ
−1
−1)z
t
), d
2
t
(x
t
, y
t
) = (γ
−1
−1)x
t
−(ψ −1)y
t
, and the
duality map in the middle given by [(x, y), (x
t
, y
t
)] = [x
t
, x] −[y
t
, y].
One can check that the images are closed. Therefore their cohomology
are in duality. For details, see Herr’s paper in Math Annalen (2001?).
108 CHAPTER 5. (ϕ, Γ)MODULES AND GALOIS COHOMOLOGY
Chapter 6
(ϕ, Γ)modules and Iwasawa
theory
6.1 Iwasawa modules H
i
Iw
(K, V )
6.1.1 Projective limits of cohomology groups
In this chapter we assume that K is a ﬁnite extension of Q
p
and G
K
is the
Galois group of
¯
K/K. Then K
n
= K(µ
p
n) and Γ
n
= Gal(K
∞
/K
n
) = γ
Zp
n
if n ≥ 1 (n ≥ 2 if p = 2) where γ
n
is a topological generator of Γ
n
. We
choose γ
n
such that γ
n
= γ
p
n−1
1
. The Iwasawa algebra Z
p
[[Γ
K
]] is isomorphic
to Z
p
[[T]] with the (p, T)adic topology by sending T to γ −1. We have
Z
p
[[Γ
K
]]/(γ
n
−1) = Z
p
[Gal(K
n
/K)].
Furthermore Z
p
[[Γ
K
]] is a G
K
module: let g ∈ G
K
and x ∈ Z
p
[[Γ
K
]], then
gx = ¯ gx, where ¯ g is the image of g in Γ
K
. By the same way, G
K
acts on
Z
p
[Gal(K
n
/K)].
Using Shapiro’s Lemma, we get, for M a Z
p
[G
K
]module,
H
i
(G
Kn
, M)
∼
−→H
i
(G
K
, Z
p
[Gal(K
n
/K)] ⊗M),
with the inverse map given by
(σ
1
, ..., σ
i
) →
¸
g∈Gal(Kn/K)
g⊗C
g
(σ
1
, ..., σ
i
)
−→
(σ
1
, ..., σ
i
) →C
id
(σ
1
, ..., σ
i
)
.
109
110 CHAPTER 6. (ϕ, Γ)MODULES AND IWASAWA THEORY
Thus we have a commutative diagram:
H
i
(G
K
n+1
, M)
∼
−−−→ H
i
(G
K
, Z
p
[Gal(K
n+1
/K)] ⊗M)
cor
H
i
(G
Kn
, M)
∼
−−−→ H
i
(G
K
, Z
p
[Gal(K
n
/K)] ⊗M)
One can check that the second vertical arrow is just induced by the natural
map Gal(K
n+1
/K) →Gal(K
n
/K).
Deﬁnition 6.1.1. (i) If V is a Z
p
representation of G
K
, deﬁne
H
i
Iw
(K, V ) = lim
←−
n
H
i
(G
Kn
, V )
while the transition maps are the corestriction maps.
(ii) If V is a Q
p
representation, choose T a stable Z
p
lattice in V , then
deﬁne
H
i
Iw
(K, V ) = Q
p
⊗
Zp
H
i
Iw
(K, T).
6.1.2 Reinterpretation in terms of measures
Proposition 6.1.2. H
i
(G
K
, Z
p
[[Γ
K
]] ⊗V )
∼
= H
i
Iw
(K, V ).
Proof. The case of Q
p
follows from the case of Z
p
by using Q
p
⊗
Zp
. Now
assume that V is a Z
p
representation of G
K
. By deﬁnition,
Λ = Z
p
[[Γ
K
]] = lim
←−
Z
p
[[Γ
K
]]/(γ
n
−1),
it induces the map θ:
H
i
(G
K
, Λ ⊗V )
θ
α
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
W
lim
←−
H
i
(G
K
, Λ/(γ
n
−1) ⊗V ) = H
i
Iw
(K, V )
lim
←−
H
i
(G
K
, Λ/(p
n
, γ
n
−1) ⊗V )
The surjectivity is general abstract nonsense.
The injectivity of α implies the injectivity of θ; to prove that of α, it is
enough to verify the MittagLeﬄer conditions of H
i−1
, which are automatic,
because of the Finiteness Theorem: Λ/(p
n
, γ
n
−1) ⊗V is a ﬁnite module, so
H
i−1
(G
K
, Λ/(p
n
, γ
n
−1) ⊗V ) is a ﬁnite group.
6.1. IWASAWA MODULES H
I
IW
(K, V ) 111
Remark. (i) Recall that T
0
(Γ
K
, V ) is the set of padic measures from Γ
K
to V :
Z
p
[[Γ
K
]] ⊗V
∼
= T
0
(Γ
K
, V ), γ ⊗v →δ
γ
⊗v,
where δ
γ
is the Dirac measure at γ. Let g ∈ G
K
, µ ∈ T
0
(Γ
K
, V ); the action
of G
K
on T
0
(Γ
K
, V ) is as follow:
Γ
K
φ(x)(gµ) = g(
Γ
K
φ(¯ gx)µ).
Hence, for any n ∈ N, the map H
i
(G
K
, Z
p
[[Γ
K
]] ⊗V ) →H
i
(G
Kn
, V ) (trans
lation of Shapiro’s lemma) can be written in the following concrete way:
(σ
1
, ..., σ
i
) →µ(σ
1
, ..., σ
i
)
−→
(σ
1
, ..., σ
i
) →
Γ
K
1
Γ
Kn
µ(σ
1
, ..., σ
i
) ∈ V
n∈N
.
(ii) Let g ∈ G
K
, λ, µ ∈ Z
p
[[Γ
K
]], x ∈ V , then
g(λµ ⊗v) = ¯ gλµ ⊗gv = λ¯ gµ ⊗gv = λg(µ ⊗µ).
So λ and g commutes, it implies that H
i
Iw
(K, V ) are Z
p
[[Γ
K
]]modules.
6.1.3 Twist by a character (`a la Soul´e)
Let η : Γ
K
→Q
∗
p
be a continuous character. It induces a transform
T
0
(Γ
K
, V ) →T
0
(Γ
K
, V ), µ →η µ.
For λ ∈ Z
p
[[Γ
K
]], we have
η (λµ) = (η λ)(η µ).
Indeed, it is enough to check it on Dirac measures. In this case
η (δ
λ
1
δ
λ
2
⊗v) = η(λ
1
λ
2
)δ
λ
1
δ
λ
2
⊗v = (η δ
λ
1
)(η δ
λ
2
) ⊗v.
Recall that Z
p
(η) = Z
p
e
η
, where, if g ∈ G
K
, then ge
η
= η(¯ g)e
η
. Deﬁne
V (η) = V ⊗Z
p
(η).
Exercise. The map µ ∈ T
0
(Γ
K
, V ) →(η µ) ⊗e
η
∈ T
0
(Γ
K
, V ) is an isomor
phism of Z
p
[G
K
]modules.
112 CHAPTER 6. (ϕ, Γ)MODULES AND IWASAWA THEORY
By the above exercise, we have a commutative diagram:
H
i
Iw
(K, V )
iη
H
i
Iw
(K, V (η))
H
i
(G
K
, T
0
(Γ
K
, V ))
∼
H
i
(G
K
, T
0
(Γ
K
, V (η)))
So i
η
is an isomorphism of cohomology groups. It can be written in a concrete
way
i
η
:
(σ
1
, ..., σ
i
) →µ(σ
1
, ..., σ
i
)
−→
(σ
1
, ..., σ
i
) →
Γ
K
1
Γ
Kn
ηµ(σ
1
, ..., σ
i
)⊗e
η
n∈N
.
It is an isomorphism of Z
p
modules.
Warning: i
η
is not an isomorphism of Z
p
[[Γ
K
]]modules, because i
η
(λx) =
(η λ)i
η
(x): there is a twist.
6.2 Description of H
i
Iw
in terms of D(V )
Remark. H
i
Iw
(K, V ) = lim
←−n≥n
0
H
i
(G
Kn
, V ), so we can always assume n 0.
Lemma 6.2.1. Let τ
n
=
γn−1
γ
n−1
−1
= 1 + γ
n−1
+, ..., +γ
p−1
n−1
∈ Z
p
[[Γ
K
]], the
diagram
C
ψ,γn
(K
n
, V ) :
0
D(V )
τn
D(V )
τn
⊕ D(V )
Id
D(V )
Id
0
C
ψ,γ
n−1
(K
n−1
, V ) :
0
D(V )
D(V ) ⊕ D(V )
D(V )
0
is commutative and induces corestrictions on cohomology via
H
i
(C
ψ,γn
(K
n
, V ))
∼
−→H
i
(G
Kn
, V ).
Proof. τ
n
is a cohomological functor and induces Tr
Kn/K
n−1
on H
0
, so it
induces corestrictions on H
i
.
Theorem 6.2.2. If V is a Z
p
or Q
p
representation of G
K
, then we have:
(i) H
i
Iw
(K, V ) = 0, if i = 1, 2.
(ii) H
1
Iw
(K, V )
∼
= D(V )
ψ=1
, H
2
Iw
(K, V )
∼
=
D(V )
ψ−1
, and the isomorphisms are
canonical.
6.2. DESCRIPTION OF H
I
IW
IN TERMS OF D(V ) 113
Remark. (i) The isomorphism
Exp
∗
: H
1
Iw
(K, V ) →D(V )
ψ=1
is the map that will produce padic Lfunctions. Let’s describe (Exp
∗
)
−1
. Let
y ∈ D(V )
ψ−1
, then (ϕ −1)y ∈ D(V )
ψ=0
. There exists unique x
n
∈ D(V )
ψ=0
satisfying that (γ
n
− 1)x
n
= y
n
, then we can ﬁnd b
n
∈ A ⊗ V such that
(ϕ −1)b
n
= x
n
. Then
g →
log χ(γ
n
)
p
n
(g −1)
(γ
n
−1)
y −(g −1)b
n
gives a cocycle on G
Kn
with values in V , and
log χ(γn)
p
n
does not depend on n.
Denote by ι
ψ,n
(y) ∈ H
1
(G
Kn
, V ) the image of this cocycle, then
(Exp
∗
)
−1
: y −→( , ι
ψ,n
(y), )
n∈N
∈ H
1
Iw
(K, V )
doesn’t depend on the choice of γ
n
.
(ii) We see that
D(V )
ψ−1
is dual to D(V
∧
(1))
ψ=1
= V
∧
(1)
H
K
, so H
2
Iw
(K, V ) =
D(V )
ψ−1
= (V
∧
(1)
H
K
)
∧
.
Before proving the theorem, we introduce a lemma.
Lemma 6.2.3. If M is compact with continuous action of Γ
K
, then
M · lim
←−
n
(M/γ
n
−1).
Proof. We have a natural map from M to lim
←−n
(M/γ
n
−1).
Injectivity: let V be an open neighborhood of 0. For all x ∈ M, there
exists n
x
∈ N and U
x
÷ x, an open neighborhood of x such that (γ−1)x
t
∈ V
for γ ∈ Γ
Knx
and x
t
∈ U
x
. By compactness, M =
¸
i∈I
U
x
i
, where I is a ﬁnite
set. Let n = max
i∈I
n
x
i
. It implies that (γ − 1)M ⊂ V , if γ ∈ Γ
n
, then
¸
n∈N
(γ
n
−1)M = 0, this shows the injectivity.
Surjectivity: Let (x
n
)
n∈N
∈ lim
←−n
(M/γ
n
−1). From the proof of injectivity,
we know that x
n
is a Cauchysequence. Because M is compact, there exists
x = limx
n
. We have x
n+k
−x
n
= (γ
n
−1)y
k
for all k ≥ 0, as M is compact,
there exists a subsequence of y
k
converging to y, passing to the limit, we get
x −x
n
= (γ
n
−1)y. This shows the surjectivity.
114 CHAPTER 6. (ϕ, Γ)MODULES AND IWASAWA THEORY
Proof of Theorem 6.2.2. H
i
Iw
(K, V ) is trivial if i ≥ 3 and the case of Q
p
follows from Z
p
by Q
p
⊗
Zp
.
For i = 0,
H
0
Iw
(K, V ) = lim
←−
Tr
V
G
Kn
.
V
G
Kn
is increasing and ⊂ V , as V is a ﬁnite dimensional Z
p
module, the
sequence is stationary for n ≥ n
0
. Then Tr
K
n+1
/Kn
is just multiplication by
p for n ≥ n
0
, but V does not contain pdivisible elements. This shows that
lim
←−Tr
V
G
Kn
= 0.
For i = 2: H
2
(G
Kn
, V ) =
D(V )
(ψ−1,γn−1)
. The corestriction map is induced by
Id on D(V ), thus
H
2
Iw
(K, V ) = lim
←−
D(V )
ψ −1
(γ
n
−1) =
D(V )
ψ −1
by Lemma 6.2.3, as
D(V )
ψ−1
is compact (and even ﬁnitely generated over Z
p
).
For i = 1: we have commutative diagrams:
0
D(V )
ψ=1
γn−1
p
1
H
1
(G
Kn
, V )
p
2
cor
(
D(V )
ψ−1
)
γn=1
τn
0
0
D(V )
ψ=1
γ
n−1
−1
H
1
(G
K
n−1
, V )
(
D(V )
ψ−1
)
γ
n−1
=1
0
where p
1
(¯ y) = ¯ y, p
2
((¯ x, ¯ y)) = ¯ x, for any x, y ∈ D(V ). Using the functor lim
←−
,
we get:
0
lim
←−
D(V )
ψ=1
γn−1
lim
←−
H
1
(G
Kn
, V )
lim
←−
(
D(V )
ψ−1
)
γn=1
.
Because D(V )
ψ=1
is compact, by Lemma 6.2.3 we have D(V )
ψ=1
· lim
←−
D(V )
ψ=1
γn−1
.
By deﬁnition, H
1
Iw
(K, V ) = lim
←−
H
1
(G
Kn
, V ). The same argument for showing
H
0
Iw
(K, V ) = 0 shows that lim
←−
(
D(V )
ψ−1
)
γn=1
= 0. So we get
D(V )
ψ=1
∼
−→H
1
Iw
(K, V ).
6.3. STRUCTURE OF H
1
IW
(K, V ) 115
6.3 Structure of H
1
Iw
(K, V )
Recall that we proved that if D is an ´etale (ϕ, Γ)module of dim d over E
Qp
,
then ( = (ϕ − 1)D
ψ=1
is a free F
p
[[Γ
Qp
]]module of rank d. The same proof
shows that if n ≥ 1, i ∈ Z
∗
p
, ( ∩ εϕ
n
(D) is free of rank d over F
p
[[Γ
n
]].
Corollary 6.3.1. If D is an ´etale (ϕ, Γ)module of dimension d over E
K
,
then ( is a free F
p
[[Γ
K
]]module of rank d [K : Q
p
].
Proof. Exercise. Hint: D is of dimension d [H
Qp
: H
K
] over E
Qp
and [K :
Q
p
] = [G
Qp
: G
K
] = [Γ
Qp
: Γ
K
][H
Qp
: H
K
].
Proposition 6.3.2. If V is a free Z
p
or Q
p
representation of rank d of G
K
,
then
(i) D(V )
ϕ=1
is the torsion subZ
p
[[Γ
K
∩ Γ
1
]]module of D(V )
ψ=1
.
(ii) We have exact sequences:
0 −→D(V )
ϕ=1
−→D(V )
ψ=1
ϕ−1
−−→((V ) −→0.
and ((V ) is free of rank d [K : Q
p
] over Z
p
[[Γ
K
]] (or over Q
p
⊗
Zp
Z
p
[[Γ
K
]]).
Corollary 6.3.3. If V is a free Z
p
representation of rank d of G
K
, then
the torsion Z
p
[[Γ
K
∩ Γ
1
]]module of H
1
Iw
(K, V ) is D(V )
ϕ=1
= V
H
K
, and
H
1
Iw
(K, V )/V
H
K
is free of rank d [K : Q
p
] over Z
p
[[Γ
K
]].
Proof of Proposition 6.3.2. D(V )
ϕ=1
= V
H
K
is torsion because it is ﬁnitely
generated over Z
p
, so (ii) implies (i). To prove (ii), we have to prove
((V )/p((V ) is free of rank d [K : Q
p
] over F
p
[[Γ
K
]].
Consider the following commutative diagram with exact rows
0
D(V )
ϕ=1
D(V )
ψ=1
ϕ−1
((V )
0
0
(D(V )/p)
ϕ=1
(D(V )/p)
ψ=1
ϕ−1
((V/p)
0
Using the exact sequence
0 →pV →V →V/p →0
116 CHAPTER 6. (ϕ, Γ)MODULES AND IWASAWA THEORY
and apply the snake lemma to the vertical rows of the diagram above, we
have the cokernel complex is
p −torsion of
D(V )
(ϕ −1)
→p −torsion of
D(V )
(ψ −1)
→
((V/p)
((V )/p((V )
→0.
Note that the ptorsion of
D(V )
(ψ−1)
is a ﬁnite dimensional F
p
vector space, thus
((V/p)
((V )/p((V )
is also a ﬁnite dimensional F
p
vector space, hence ((V )/p((V ) is
a F
p
[[Γ
K
]]lattice of ((V/p), but ((V/p) is a free F
p
[[Γ
K
]]module of rank
d [K : Q
p
] by Corollary 6.3.1.
Remark. (i) The sequence
0 →D(V )
ϕ=1
→D(V )
ψ=1
→((V ) →0
is just the inﬂationrestriction exact sequence
0 →H
1
(Γ
K
, Λ ⊗V
H
K
) →H
1
(G
K
, Λ ⊗V ) →H
1
(H
K
, Λ ⊗V )
Γ
K
→0.
(ii) Let 0 → V
1
→ V → V
2
→ 0 be an exact sequence, then the exact
sequence
0 →D(V
1
) →D(V ) →D(V
2
) →0
and the snake lemma induces
0 →D(V
1
)
ψ=1
→D(V )
ψ=1
→D(V
2
)
ψ=1
→
D(V
1
)
ψ −1
→
D(V )
ψ −1
→
D(V
2
)
ψ −1
→0.
By Theorem 6.2.2, this is just
0 →H
1
Iw
(K, V
1
) →H
1
Iw
(K, V ) →H
1
Iw
(K, V
2
)
→H
2
Iw
(K, V
1
) →H
2
Iw
(K, V ) →H
2
Iw
(K, V
2
) →0.
It can also be obtained from the longer exact sequence in continuous coho
mology from the exact sequence
0 →Z
p
[[Γ
K
]] ⊗V
1
→Z
p
[[Γ
K
]] ⊗V →Z
p
[[Γ
K
]] ⊗V
2
→0.
Chapter 7
Z
p
(1) and KubotaLeopoldt zeta
function
7.1 The module D(Z
p
(1))
ψ=1
The module Z
p
(1) is just Z
p
with the action of G
Qp
by g ∈ G
Qp
, x ∈ Z
p
(1),
g(x) = χ(g)x. We shall study the exponential map
Exp
∗
: H
1
Iw
(Q
p
, Z
p
(1)) →D(Z
p
(1))
ψ=1
.
Note that D(Z
p
(1)) = (A⊗Z
p
(1))
H
Qp
= A
Qp
(1), with usual actions of ϕ and
ψ, and for γ ∈ Γ, γ(f(π)) = χ(γ)f((1 + π)
χ(γ)
−1), for all f(π) ∈ A
Qp
(1).
Proposition 7.1.1. (i) A
ψ=1
Qp
= Z
p
1
π
⊕(A
+
Qp
)
ψ=1
.
(ii) We have an exact sequence:
0 −→Z
p
−→(A
+
Qp
)
ψ=1
ϕ−1
−−→(πA
+
Qp
)
ψ=0
−→0.
Remark. Under the map µ →
Zp
[ε]
x
µ, (πA
+
Qp
)
ψ=0
is the image of measures
with support in Z
∗
p
(ψ = 0) and
Z
∗
p
µ = 0
(πA
+
Qp
)
ψ=0
= ((Z
p
) = (γ −1)Z
p
[[Γ
Qp
]].
Z
p
[[Γ
Qp
]] can be viewed as measures on Γ
Qp
∼
= Z
∗
p
, and µ ∈ (γ − 1)Z
p
[[Γ
Qp
]]
means
Zp
µ = 0. It implies that ((Z
p
) is free of rank 1 over Z
p
[[Γ
Qp
]] which
is a special case of what we have proved.
117
118CHAPTER 7. Z
P
(1) AND KUBOTALEOPOLDT ZETA FUNCTION
Proof. (i) We have proved
ψ(A
+
Qp
) ⊂ A
+
Qp
, ψ(
1
π
) =
1
π
,
v
E
(ψ(x)) ≥ [
v
E
x
p
], if x ∈ E
Qp
.
These facts imply that ψ −1 is bijective on E
Qp
/¯ π
−1
E
+
Qp
and hence it is also
bijective on A
Qp
/π
−1
A
+
Qp
. So
ψ(x) = x ⇒ x ∈ π
−1
A
+
Qp
.
(ii) We know that (ϕ −1)A
+
Qp
⊂ πA
+
Qp
For x ∈ (πA
+
Qp
)
ψ=0
, then
ϕ
n
(x) ∈ ϕ
n
(π)A
+
Qp
→0 if n →∞.
Hence y =
+∞
¸
n=0
ϕ
n
(x) converges, and one check that ψ(y) = y, (ϕ−1)y = −x.
This implies the surjectivity of ϕ −1.
7.2 Kummer theory
Recall that
ε = (1, ε
(1)
, ε
(2)
, ..., ε
(n)
, ...) ∈ E
+
Qp
⊂
˜
E
+
= R, ε
(1)
= 1.
Let π
n
= ε
(n)
−1, F
n
= Q
p
(π
n
) for n ≥ 1. Then π
n
is a uniforming parameter
of F
n
, and
N
F
n+1
/Fn
(π
n+1
) = π
n
, O
F
n+1
= O
Fn
[π
n+1
]/((1 + π
n+1
)
p
= 1 + π
n
).
For an element a ∈ F
∗
n
, choose x = (a, x
(1)
, ...) ∈
˜
E. This x is unique up to
ε
u
with u ∈ Z
p
. So if g ∈ G
Fn
, then
g(x)
x
= ε
c(g)
, c(g) ∈ Z
p
gives a 1cocycle c on G
Fn
with values in Z
p
(1). This deﬁnes the Kummer
map:
κ : F
∗
n
−→H
1
(G
Fn
, Z
p
(1))
a −→κ(a).
7.3. COLEMAN’S POWER SERIES 119
By Kummer theory, we have H
1
(G
Fn
, Z
p
(1)) = Z
p
κ(π
n
) ⊕ κ(O
∗
Fn
). The
diagram
F
∗
n+1
κ
−−−→ H
1
(G
F
n+1
, Z
p
(1))
N
F
n+1
/Fn
cor
F
∗
n
κ
−−−→ H
1
(G
Fn
, Z
p
(1))
is commutative, we have a map:
κ : lim
←−
F
∗
n
→H
1
Iw
(Q
p
, Z
p
(1))
and
H
1
Iw
(Q
p
, Z
p
(1)) = Z
p
κ(π
n
) ⊕κ(lim
←−
O
∗
Fn
).
7.3 Coleman’s power series
Theorem 7.3.1 (Coleman’s power series). Let u = (u
n
)
n≥1
∈ lim
←−
(O
Fn
)−
¦0¦ (pour les applications N
F
n+1
/Fn
), then there exists a unique power series
f
u
∈ Z
p
[[T]] such that f
u
(π
n
) = u
n
for all n ≥ 1.
Lemma 7.3.2. (i) If x ∈ O
Fn
, γ ∈ Gal(F
n+1
/F
n
), then γ(x) −x ∈ π
1
O
F
n+1
.
(ii) N
F
n+1
/Fn
x −x
p
∈ π
1
O
F
n+1
.
Proof. It is easy to see that (i) implies (ii) since [F
n+1
: F
n
] = p. Write
χ(γ) = 1 +p
n
u for u ∈ Z
p
. Let x =
p−1
¸
i=0
x
i
(1 +π
n+1
)
i
, where x
i
∈ O
Fn
. Then
γ(x) −x =
p−1
¸
i=0
x
i
(1 +π
n+1
)
i
((1 + π
1
)
iu
−1) ∈ π
1
O
F
n+1
.
Corollary 7.3.3. ¯ u = (¯ u
p
1
, ¯ u
1
, ..., ¯ u
n
, ...) ∈ E
+
Qp
, where ¯ u
n
is the image of
u
n
mod π
1
.
Deﬁnition 7.3.4. Let N : O
F
1
[[T]] →O
F
1
[[T]] such that
N(f)((1 + T)
p
−1) =
¸
z
p
=1
f((1 + T)z −1).
120CHAPTER 7. Z
P
(1) AND KUBOTALEOPOLDT ZETA FUNCTION
Lemma 7.3.5. (i) N(f)(π
n
) = N
F
n+1
/Fn
(f(π
n+1
)),
(ii) N(Z
p
[[T]]) ⊂ Z
p
[[T]],
(iii) N(f) −f ∈ π
1
O
F
1
[[T]],
(iv) If f ∈ O
F
1
[[T]]
∗
, k ≥ 1, if (f −g) ∈ π
k
1
O
F
1
[[T]], then
N(f) −N(g) ∈ π
k+1
1
O
F
1
[[T]].
Proof. (i) The conjugates of π
n+1
under Gal(F
n+1
/F
n
) are those (1+π
n
)z −1
for z
p
= 1, this implies (i).
(ii) Obvious, is just Galois theory.
(iii) Look mod π
1
, because z = 1 mod π
1
, we have N(f)(T
p
) = f(T)
p
.
(iv) We have N(
f
g
) =
N(f)
N(g)
, so we can reduce to f = 1 and g = 1 + π
k
1
h.
Then
N(g)((1 + T)
p
−1) = 1 + π
k
1
¸
z
p
=1
h((1 + T)z −1) mod π
k+1
1
,
and
¸
z
p
=1
h((1 + T)z −1) is divisible by p.
Corollary 7.3.6. (i) If ¯ u ∈ E
+
Qp
and v
E
(¯ u) = 0, then there exists a unique
g
u
∈ Z
p
[[T]] such that N(g
u
) = g
u
and g
u
(¯ π) = ¯ u.
(ii) If x ∈ 1 + π
k
1
O
F
n+1
, then N
F
n+1
/Fn
(x) ∈ 1 + π
k+1
1
O
Fn
.
Proof. (i) Take any g ∈ Z
p
[[T]] such that g(¯ π) = ¯ u, then g ∈ Z
p
[[T]]
∗
, by (iv)
of Lemma 7.3.5, N
k
(g) converges in g + π
1
Z
p
[[T]] and g
u
is the limit.
(ii) There exists f ∈ 1 + π
k
1
O
F
1
[T] such that x = f(π
n+1
). Then use (i)
and (iv) of Lemma 7.3.5.
Proof of Theorem 7.3.1. The uniqueness follows from the fact that 0 = f ∈
Z
p
[[T]] has only many ﬁnitely zeros in m
Cp
(Newton polygons).
Existence: let u = (u
n
), write u
n
= π
k
n
αu
t
n
, where k ∈ Z and α ∈ µ
p−1
do not depend on n, and u
t
n
∈ 1 + m
Fn
. Then N
F
n+1
/Fn
u
t
n+1
= u
t
n
. If for all
n, f
u
(π
n
) = u
t
n
, let f
u
= T
k
αf
u
, then f
u
(π
n
) = u
n
. Thus we are reduced to
the case that u
n
∈ 1 + m
Fn
for all n.
By (i) of Corollary 7.3.6, we can ﬁnd g
u
∈ Z
p
[[T]] for ¯ u. We have to check
that g
u
(π
n
) = u
n
for all n = 1. Write v
n
= g
u
(π
n
). Then N(g
u
) = g
u
, by
(i) of Lemma 7.3.5, implies that N
F
n+1
/Fn
(v
n+1
) = v
n
; and g
u
(¯ π) = ¯ u implies
that v
n
= u
n
mod π
1
for all n. Let w
n
=
vn
un
, then we have
N
F
n+1
/Fn
(w
n+1
) = w
n
and w
n
∈ 1 + π
1
O
Fn
.
7.3. COLEMAN’S POWER SERIES 121
By (ii) of Corollary 7.3.6, we have
w
n
= N
F
n+k
/Fn
(w
n+k
) ∈ 1 + π
k
1
O
Fn
for all k,
then w
n
= 1. This completes the proof.
Corollary 7.3.7.
N(f
u
) = f
u
, ψ(
∂f
u
f
u
) =
∂f
u
f
u
where ∂ = (1 + T)
d
dT
.
Proof. By (i) of Lemma 7.3.5, we have N(f
u
)(π
n
) = N
F
n+1
/Fn
(f
u
(π
n+1
)) =
f
u
(π
n
), for all n, thus N(f
u
) = f
u
.
Using the formula ψ(∂ log f) = ∂(log N(f)), we immediately get the result
for ψ. As for the proof of this last formula, we know that
ϕ(N(f)(T)) = N(f)((1 + T)
p
−1) =
¸
z
p
=1
f((1 + T)z −1)
ψ(f)((1 + T)
p
−1) =
1
p
¸
z
p
=1
f((1 + T)z −1)
Then we have two ways to write ∂(log ϕ(N(f)))
∂(log ϕ(N(f))) = pϕ (∂ log N(f))(∂ ◦ ϕ = pϕ ◦ ∂)
= pϕ(
∂N(f)
N(f)
) = p(
∂N(f)
N(f)
)((1 + T)
p
−1)
= p(∂ log N(f))((1 + T)
p
−1),
∂(log ϕ(N(f))) = ∂(log
¸
z
p
=1
f((1 + T)z −1))
=
¸
z
p
=1
(1 +T)zf
t
((1 + T)z −1)
f((1 + T)z −1)
=
¸
z
p
=1
∂f
f
((1 + T)z −1) = pψ(
∂f
f
)((1 + T)
p
−1)
= p(ψ(∂ log f))((1 + T)
p
−1),
hence the formula.
122CHAPTER 7. Z
P
(1) AND KUBOTALEOPOLDT ZETA FUNCTION
7.4 An explicit reciprocity law
Theorem 7.4.1. The diagram
lim
←−
(O
Fn
−¦0¦)
u→
∂fu
fu
(π)
Q
Q
Q
Q
Q
Q
Q
Q
Q
Q
Q
Q
Q
κ
H
1
Iw
(Q
p
, Z
p
(1))
Exp
∗
∼
m
m
m
m
m
m
m
m
m
m
m
m
m
D(Z
p
(1))
ψ=1
is commutative.
Remark. (i) The proof is typical of invariants deﬁned via Fontaine’s rings:
easy to deﬁne and hard to compute.
(ii) For another example, let X/K be a smooth and projective variety,
then
D
dR
(H
i
´ et
(X K, Q
p
)) = H
i
dR
(X/K).
The proof is very hard and is due to Faltings and Tsuji.
(iii) Let a ∈ Z such that a = 1, (a, p) = 1. The element
u
n
=
e
−a
2πi
p
n
−1
e
−
2πi
p
n
−1
∈ Q(µ
p
n)
is a cyclotomic unit in O
Q(µ
p
n)
(whose units are called global units). Then
u
n
∈ F
n
= Q
p
(µ
p
n), u
n
=
γ
−a
(π
n
)
γ
−1
(π
n
)
,
where γ
b
∈ Γ
Qp
such that χ(γ
b
) = b. From N
F
n+1
/Fn
(π
n+1
) = π
n
, one gets
N
F
n+1
/Fn
(u
n+1
) = u
n
(γ commutes with norm), thus
u = (u
n
) ∈ lim
←−
O
Fn
.
Obviously the Coleman power series
f
u
=
(1 +T)
−a
−1
(1 + T)
−1
−1
,
∂f
u
f
u
=
a
(1 +T)
a
−1
−
1
T
.
So
∂fu
fu
is nothing but the Amice transform of µ
a
that was used to construct
padic zeta function. So Exp
∗
produces KubotaLeopoldt zeta function from
the system of cyclotomic units.
7.5. PROOF OF THE EXPLICIT RECIPROCITY LAW 123
(iv) The example in (iii) is part of a big conjectural picture. For V a ﬁxed
representation of G
Q
, then conjecturally
¦compatible system of global elements of V ¦ −→H
1
Iw
(Q, V )
−→H
1
Iw
(Q
p
, V )
Exp
∗
−→D(V )
ψ=1
Amice
−−−−−−→
transform
padic Lfunctions.
At present there are very few examples representation of G
Q
for which this
picture is known to work. The Amice transform works well for Z
p
(1), because
ψ improves denominators in π, and A
ψ=1
Qp
⊂
1
π
A
+
Qp
can be viewed as measures.
In general, to use the properties of ψ, we will have to introduce overconvergent
(ϕ, Γ)modules.
7.5 Proof of the explicit reciprocity law
7.5.1 Strategy of proof of Theorem 7.4.1
Let u ∈ lim
←−
O
Fn
, and g → C
n
(g) be the cocycle on G
Fn
by Kummer theory,
i.e the image of u under the composition of
lim
←−
(O
Fn
−¦0¦)
κ
−→H
1
Iw
(Q
p
, Z
p
(1)) −→H
1
(G
Fn
, Z
p
(1)).
Let y ∈ D(Z
p
(1))
ψ=1
= A
ψ=1
Qp
(1), let g →C
t
n
(g) be the image of y under the
composition of
D(Z
p
(1))
ψ=1
= A
ψ=1
Qp
(1)
(Exp
∗
)
−1
−−−−−→H
1
Iw
(Q
p
, Z
p
(1)) −→H
1
(G
Fn
, Z
p
(1)).
We need to prove that C
n
= C
t
n
for all n implies y =
∂fu
fu
(π).
For C
t
n
, we have
C
t
n
(g) =
log χ(γ
n
)
p
n
χ(g) −1
χ(γ
n
) −1
y −(χ(g)g −1)b
n
,
where b
n
∈ A is a solution of (ϕ −1)b
n
= (χ(γ
n
)γ
n
−1)
−1
(ϕ −1)y, we know
that (ϕ −1)y ∈ A
ψ=0
Qp
. The exact value of b
n
is not important.
For C
n
, choose x
n
= (x
(0)
n
, ..., x
(k)
n
, ...) ∈
˜
E
+
such that x
(0)
n
= u
n
. Let
˜ u
n
= [x
n
], then
g(˜ u
n
)
˜ u
n
= [ε]
Cn(g)
.
124CHAPTER 7. Z
P
(1) AND KUBOTALEOPOLDT ZETA FUNCTION
Proposition 7.5.1. Assume n ≥ 1.
(i) There exists k ∈ Z, b
t
n
∈ O
Cp
/p
n
such that
p
2
C
t
n
(g) = p
2
log χ(γ
n
)
p
n
g −1
γ
n
−1
y(π
n+k
) + (g −1)b
t
n
in O
Cp
/p
n
.
(ii) There exists k ∈ Z, b
tt
n
∈ O
Cp
/p
n
such that
p
2
C
n
(g) = p
2
log χ(g)
p
n
∂f
u
f
u
y(π
n
) + (g −1)b
tt
n
in O
Cp
/p
n
.
Proposition 7.5.2. There exists a constant c ∈ N, such that for all n and
for all k, if x ∈ O
F
n+k
, b ∈ O
Cp
satisfy
v
p
g −1
γ
n
−1
x + (g −1)b
≥ n, ∀g ∈ G
Fn
then
p
−k
Tr
F
n+k
/Fn
x ∈ p
n−c
O
Fn
.
We shall prove Proposition 7.5.1 in the next n
o
, and Proposition 7.5.2
in the third n
o
. We ﬁrst explain how the above two propositions imply the
theorem:
If h(π) = ψ(h(π)), then h(π
n
) = p
−1
Tr
F
n+1
/Fn
(h(π
n+1
)). By hypothesis,
ψ(y) = y, we get
p
−k
Tr
F
n+k
/Fn
(y(π
n+k
)) = y(π
n
), ∀n, ∀k (∗).
Let
x = p
2
log χ(γ
n
)
p
n
y(π
n+k
) −
∂f
u
f
u
(π
n
)
, b = b
t
n
−b
tt
n
.
By Proposition 7.5.1, and the hypothesis C
n
(g) = C
t
n
(g), we get
g −1
γ
n
−1
x + (g −1)b = p
2
(C
t
n
(g) −C
n
(g)) = 0.
The ﬁrst equality is because for every x ∈ F
n
,
g−1
γn−1
x =
log χ(g)
log χ(γn)
x. Using
Proposition 7.5.2, we get
p
2
log(χ(γ
n
))
p
n
y(π
n
) −
∂f
u
f
u
(π
n
)
∈ p
n−c
O
Fn
,
7.5. PROOF OF THE EXPLICIT RECIPROCITY LAW 125
then for every n,
y(π
n
) −
∂f
u
f
u
(π
n
) ∈ p
n−c−2
O
Fn
.
Let h = y −
∂fu
fu
, then ψ(h) = h and h(π
n
) ∈ p
n−c−2
O
Fn
. Using the fact
p
−k
Tr
F
n+k
/Fn
O
F
n+k
⊂ O
Fn
and the formula (*), then for every i ∈ N, n ≥ i,
h(π
i
) = p
−(n−i)
Tr
Fn/F
i
(h(π
n
)) ∈ p
n−c−2
O
F
i
,
thus h(π
i
) = 0 for every i ∈ N, hence h = 0.
7.5.2 Explicit formulas for cocyles
This n
o
is devoted to the proof of Proposition 7.5.1
(i) Recall that π = [ε] −1, θ(
¸
p
m
[x
m
]) =
¸
p
m
x
(0)
m
and θ(π) = 1 −1 =
0. Let ˜ π
n
= ϕ
−n
(π) ∈
˜
A
+
, then ˜ π
n
= [ε
1/p
n
] − 1, θ(˜ π
n
) = π
n
. Write
b
n
=
+∞
¸
l=0
p
l
[z
l
], where z
l
∈
˜
E. As C
t
n
(g) ∈ Z
p
, we have
ϕ
−(n+k)
C
t
n
(g) = C
t
n
(g), for all n and k.
As
v
E
(ϕ
−k
(z
l
)) =
1
p
k
v
E
(z
l
),
we can ﬁnd k such that
v
E
(ϕ
−(n+k)
(z
l
)) ≥ −1, for all l n −1.
Let ˜ p = (p, ...) ∈
˜
E
+
, then for every l n−1, ˜ p ϕ
−(n+k)
(z
l
) ∈
˜
E
+
. We have
[ ˜ p]C
t
n
(g) =
log χ(γ
n
)
p
n
[ ˜ p]
χ(g)g −1
χ(γ
n
)γ
n
−1
y(˜ π
n+k
) + [ ˜ p](χ(g)g −1)ϕ
−(n+k)
(b
n
).
Both sides live in
˜
A
+
+ p
n
˜
A, reduce mod p
n
and use θ :
˜
A
+
/p
n
→ O
Cp
/p
n
,
then [ ˜ p] →p and
pC
t
n
(g) = p
log χ(γ
n
)
p
n
p
g −1
γ
n
−1
y(π
n+k
) + (g −1)b
t
n
where b
t
n
= θ([ ˜ p]ϕ
−(n+k)
(b
n
)).
126CHAPTER 7. Z
P
(1) AND KUBOTALEOPOLDT ZETA FUNCTION
(ii) Write u = (π
k
n
)(v
n
), where v
n
are units ∈ O
∗
Fn
. So we just have to
prove the formula for (π
n
) and (v
n
). Thus we can assume v
p
(u
n
) 1.
Let
H : 1 +tB
+
dR
→C
p
, x →θ(
x −1
π
) = θ(
x −1
t
),
recall that t = log(1 + π). We have
H((1+πx
t
)(1+πy
t
)) = H(1+π(x
t
+y
t
)+π
2
x
t
y
t
) = θ(x
t
+y
t
) = H(1+πx
t
)+H(1+πy
t
),
thus H(xy) = H(x) + H(y).
Write ˜ u
n
= [(u
n
, u
1
p
n
, ...)], we have
g(˜ un)
˜ un
= [ε]
Cn(g)
= 1+C
n
(g)π+ , thus
C
n
(g) = H(
g(˜ u
n
)
˜ u
n
).
We know u
n
= f
u
(π
n
) and θ(˜ u
n
) = u
n
, then
θ(f
u
(˜ π
n
)) = f
u
(θ(˜ π
n
)) = f
u
(π
n
) = u
n
= θ(˜ u
n
).
So, if we set a
n
=
fu(˜ πn)
˜ un
, then θ(a
n
) = 1.
We know that [ ˜ p]a
n
∈
˜
A
+
since v
p
(u
n
) ≤ 1. Then we get H(a
n
) ∈
1
pπ
1
O
Cp
because of the following identity
H(a
n
) = θ
[ ˜ p]a
n
−[ ˜ p]
[ ˜ p]π
= θ
[ ˜ p]a
n
−[ ˜ p]
ω
θ
1
[ ˜ p]˜ π
1
,
and because ω =
π
˜ π
1
is a generator of Ker θ in
˜
A
+
as ω ∈ Ker θ, and
¯ ω =
ε −1
ε
1/p
−1
, so v
E
(¯ ω) = (1 −
1
p
)v
E
(ε −1) = 1.
Then we have
g(f
u
(˜ π
n
))
f
u
(˜ π
n
)
=
f
u
((1 + ˜ π
n
)
χ(g)
−1)
f
u
(˜ π
n
)
=
f
u
((1 + ˜ π
n
)(1 + π)
χ(g)−1
p
n
−1)
f
u
(˜ π
n
)
= 1 +
∂f
u
f
u
(˜ π
n
)
χ(g) −1
p
n
π + terms of higher degree in π,
7.5. PROOF OF THE EXPLICIT RECIPROCITY LAW 127
hence
H
g(f
u
(˜ π
n
))
f
u
(˜ π
n
)
=
χ(g) −1
p
n
∂f
u
f
u
(π
n
).
Using formula f
u
(˜ π
n
) = ˜ u
n
a
n
, we get
C
n
(g) = H
g(˜ u
n
)
˜ u
n
= H
g(f
u
(˜ π
n
))
f
u
(˜ π
n
)
−H
g(a
n
)
a
n
=
χ(g) −1
p
n
∂f
u
f
u
(π
n
) −(χ(g)g −1)H(a
n
).
We conclude the proof by multiplying p
2
, noticing that χ(g) = 1 mod p
n
, so
χ(g) −1
p
n
=
exp(log χ(g)) −1
p
n
=
log χ(g)
p
n
mod p
n
;
set b
tt
n
= −p
2
H(a
n
), we get the result.
7.5.3 Tate’s normalized trace maps
Let π
n
= ε
(n)
−1, F
n
= Q
p
(π
n
), F
∞
=
¸
n
F
n
.
Lemma 7.5.3. If n ≥ 1, x ∈ F
∞
, then p
−k
Tr
F
n+k
/Fn
x does not depend on
k such that x ∈ F
n+k
.
Proof. Use the transitive properties of the trace map and the fact [F
n+k
:
F
n
] = p
k
.
Let R
n
: F
∞
−→F
n
be the above map. Denote
Y
i
= ¦x ∈ F
i
, Tr
F
i
/F
i−1
x = 0¦.
Lemma 7.5.4. (i) R
n
commutes with Γ
Qp
, is F
n
linear and R
n
◦R
n+k
= R
n
.
(ii) Let x ∈ F
∞
, then x = R
n
(x)+
+∞
¸
i=1
R
∗
n+i
(x), where R
∗
n+i
(x) = R
n+i
(x)−
R
n+i−1
(x) ∈ Y
n+i
and is 0 if i 0.
(iii) Let k ∈ Z, then v
p
(x) kv
p
(π
n
) if and only if v
p
(R
n
(x)) kv
p
(π
n
)
and v
p
(R
∗
n+i
(x)) kv
p
(π
n
) for every i ∈ N.
128CHAPTER 7. Z
P
(1) AND KUBOTALEOPOLDT ZETA FUNCTION
Proof. (i) is obvious.
(ii) is also obvious, since R
n+i−1
(R
∗
n+i
(x)) = 0 ⇒R
∗
n+i
(x) ∈ Y
n+i
.
(iii) ⇐ is obvious. For ⇒, let x ∈ O
F
n+k
, then
x =
p
k
−1
¸
j=0
a
j
(1 + π
n+k
)
j
, a
j
∈ O
Fn
.
Write j = p
k−i
j
t
with (j
t
, p) = 1, then
R
n
(x) = a
0
, R
∗
n+i
(x) =
¸
(j
,p)=1
a
p
n−i
j
(1 +π
n+i
)
j
since
p
−1
Tr
F
n+i
/F
n+i−1
(1 +π
n+i
)
j
=
(1 +π
n+i
)
j
, if p [ j
0, if (j, p) = 1.
Thus
v
p
(x) 0 ⇒v
p
(R
n
(x)) 0 and v
p
(R
∗
n+i
(x)) 0.
By F
n
linearity we get the result.
Remark. In the whole theory, the following objects play similar roles:
ψ ←→p
−1
Tr
F
n+1
/Fn
ψ = 0 ←→ Y
i
.
Lemma 7.5.5. Assume that j i − 1 and j ≥ 2. and assume γ
j
is a
generator of Γ
j
. Let u ∈ Q
∗
p
. If v
p
(u −1) > v
p
(π
1
), then uγ
j
−1 is invertible
on Y
i
. Moreover if x ∈ Y
i
, v
p
(x) kv
p
(π
n
), then v
p
((uγ
j
−1)
−1
x) kv
p
(π
n
)−
v
p
(π
1
).
Proof. If γ
i−1
= γ
p
i−j−1
j
, then
(uγ
j
−1)
−1
= (u
p
i−j−1
γ
i−1
−1)
−1
(1 +uγ
j
+ + (uγ
j
)
p
i−j−1
−1
),
so it is enough to treat the case j = i −1.
Let x ∈ O
F
i
∩ Y
i
, write
x =
p−1
¸
a=1
x
a
(1 + π
i
)
a
, x
a
∈ O
F
i−1
,
7.5. PROOF OF THE EXPLICIT RECIPROCITY LAW 129
write χ(γ
i−1
) = 1 + p
i−1
v with v ∈ Z
∗
p
, then
(uγ
i−1
−1)x =
p−1
¸
a=1
x
a
(1 + π
i
)
a
(u(1 + π
1
)
av
−1).
We can check directly
(uγ
i−1
−1)
−1
x =
p−1
¸
a=1
x
a
(u(1 + π
1
)
av
−1)
(1 +π
i
)
a
.
Moreover, if v
p
(x) ≥ 0, then v
p
((uγ
j
−1)
−1
x) −v
p
(π
1
).
Proposition 7.5.6. Assume n 1, u ∈ Q
∗
p
and v
p
(u −1) > v
p
(π
1
), then
(i) x ∈ F
∞
can be written uniquely as x = R
n
(x) + (uγ
n
− 1)y with
R
n
(y) = 0, and we have
v
p
(R
n
(x)) > v
p
(x) −v
p
(π
n
), v
p
(y) > v
p
(x) −v
p
(π
n
) −v
p
(π
1
).
(ii) R
n
extends by continuity to
ˆ
F
∞
, and let X
n
= ¦x ∈
ˆ
F
∞
, R
n
(x) = 0¦.
Then every x ∈
ˆ
F
∞
can be written uniquely as x = R
n
(x) + (uγ
n
−1)y with
y ∈ X
n
and R
n
(x) ∈ F
n
, and with the same inequalities
v
p
(R
n
(x)) v
p
(x) −v
p
(π
n
), v
p
(y) v
p
(x) −v
p
(π
n
) −v
p
(π
1
).
Proof. (i) As
x = R
n
(x) +
+∞
¸
i=1
(uγ
n
−1)((uγ
n
−1)
−1
R
∗
n+i
(x)).
we just let y =
+∞
¸
i=1
(uγ
n
−1)
−1
R
∗
n+i
(x).
(ii) By (i), we have v
p
(R
n
(x)) v
p
(x) − C, so R
n
extends by continuity
to
ˆ
F
∞
; the rest follows by continuity.
Remark. (i) The maps R
n
:
ˆ
F
∞
−→F
n
are Tate’s normalized trace maps.
(ii) they commutes with Γ
Qp
(or G
Qp
).
(iii) R
n
(x) = x if x ∈ F
∞
and n 0, hence R
n
(x) → x if x ∈
ˆ
F
∞
and
n →∞.
130CHAPTER 7. Z
P
(1) AND KUBOTALEOPOLDT ZETA FUNCTION
7.5.4 Applications to Galois cohomology
Proposition 7.5.7. (i) The map
x ∈ F
n
−→(γ →x log χ(γ)) ∈ H
1
(Γ
Fn
, F
n
)
induces isomorphism
F
n
∼
→H
1
(Γ
Fn
, F
n
)
∼
→H
1
(Γ
Fn
,
ˆ
F
∞
).
(ii) If η : Γ
Fn
−→Q
∗
p
is of inﬁnite order, then H
1
(Γ
Fn
,
ˆ
F
∞
(η)) = 0.
Proof. If n 0 so that v
p
(η(γ
n
) −1) > v
p
(π
1
). Using the above proposition
(let u = η(γ
n
)), we get
H
1
(Γ
Fn
,
ˆ
F
∞
(η)) =
ˆ
F
∞
(uγ
n
−1)
=
F
n
¸
X
n
(uγ
n
−1)
=
F
n
uγ
n
−1
.
If u = 1, we get (γ
n
−1)F
n
= 0. If u = 1, we get F
n
/(u −1)F
n
= 0.
For n small, using inﬂation and restriction sequence, as Gal(F
n+k
/F
n
) is
ﬁnite, and
ˆ
F
∞
(η) is a Q
p
vector space, we have
H
1
(Gal(F
n+k
/F
n
),
ˆ
F
∞
(η)
Γ
F
n+k
) = 0, H
2
= 0,
then we get an isomorphism
H
1
(Γ
Fn
,
ˆ
F
∞
(η))
∼
−→H
1
(Γ
F
n+k
,
ˆ
F
∞
(η))
Gal(F
n+k
/Fn)
.
From the case of n 0, we immediately get the result.
Recall that the following result is the main step in Ax’s proof of the
AxSenTate theorem (cf. Fontaine’s course).
Proposition 7.5.8. There exists a constant C ∈ N, such that if x ∈ C
p
, if
H ⊂ G
Qp
is a closed subgroup, if for all g ∈ H, v
p
((g −1)x) a for some a,
then there exists y ∈ C
H
p
such that v
p
(x −y) a −C.
The following corollary is Proposition 7.5.2 in the previous section.
Corollary 7.5.9. For x ∈ O
ˆ
F∞
, if there exists c ∈ O
Cp
such that
v
p
g −1
γ
n
−1
x −(g −1)c
n, for all g ∈ G
Fn
.
Then we have
v
p
(R
n
(x)) n −C −1(or 2).
7.5. PROOF OF THE EXPLICIT RECIPROCITY LAW 131
Proof. By assumption, we get
v
p
((g −1)c) n, ∀ g ∈ H
Qp
= Ker χ,
then by Ax, there exists c
t
∈
ˆ
F
∞
such that
v
p
(c −c
t
) n −C.
Take g = γ
n
, then v
p
(x−(γ
n
−1)c
t
) n−C. As R
n
γ
n
= γ
n
R
n
= R
n
, we get
v
p
(R
n
(x)) = v
p
(R
n
(x −(γ
n
−1)c
t
)) n −C −v
p
(π
1
) −v
p
(π
n
),
hence the result.
7.5.5 No 2πi in C
p
!
Theorem 7.5.10. (i) C
p
does not contain log 2πi, i.e. there exists no x ∈ C
p
satisﬁes that g(x) = x+log χ(g) for all g ∈ G
K
, where K is a ﬁnite extension
of Q
p
.
(ii) C
p
(k) = 0, if k = 0.
Proof. (i) If K = Q
p
, if there exists such an x, by AxSenTate, we get
x ∈
ˆ
F
∞
= C
H
Qp
p
. Then we have:
R
n
(g(x)) = g(R
n
(x)) = R
n
(x) + log χ(g).
Because R
n
(x) ∈ F
n
, it has only ﬁnite number of conjugates but log χ(g) has
inﬁnitely many values, contradiction!
Now for K general, we can assume K/Q
p
is Galois, let
y =
1
[K : Q
p
]
¸
σ∈S
σ(x)
where S are representatives of G
Qp
/G
K
. For g ∈ G
Qp
, we can write gσ = σ
t
σ
h
σ
for h
σ
∈ G
K
and σ
t
σ
∈ S. From this we get
¸
σ∈S
log χ(h
σ
) = [K : Q
p
] log χ(g).
132CHAPTER 7. Z
P
(1) AND KUBOTALEOPOLDT ZETA FUNCTION
Then we have
g(y) =
1
[K : Q
p
]
¸
σ∈S
gσ(x) =
1
[K : Q
p
]
¸
σ∈S
σ
t
σ
h
σ
x
=
1
[K : Q
p
]
¸
σ∈S
σ
t
σ
(x + log χ(h
σ
))
=
1
[K : Q
p
]
¸
σ∈S
σ(x) +
1
[K : Q
p
]
¸
σ∈S
log χ(h
σ
)
= y + log χ(g).
Then by the case K = Q
p
, we get the result.
(ii) If 0 = x ∈ C
p
(k), then g(x) = χ(g)
−k
x. Let y =
log x
−k
, then we have
g(y) = y + log χ(g), which is a contradiction by (i).
Chapter 8
(ϕ, Γ)modules and padic
Lfunctions
8.1 TateSen’s conditions
8.1.1 The conditions (TS1), (TS2) and (TS3)
Let G
0
be a proﬁnite group and χ : G
0
→Z
∗
p
be a continuous group homo
morphism with open image. Set v(g) = v
p
(log χ(g)) and H
0
= Ker χ.
Suppose
˜
Λ is a Z
p
algebra and
v :
˜
Λ −→R ∪ ¦+∞¦
satisﬁes the following conditions:
(i) v(x) = +∞ if and only if x = 0;
(ii) v(xy) v(x) + v(y);
(iii) v(x + y) inf(v(x), v(y));
(iv) v(p) > 0, v(px) = v(p) + v(x).
Assume
˜
Λ is complete for v, and G
0
acts continuously on
˜
Λ such that
v(g(x)) = v(x) for all g ∈ G
0
and x ∈
˜
Λ.
Deﬁnition 8.1.1. The TateSen’s conditions for the quadruple (G
0
, χ,
˜
Λ, v)
are the following three conditions TS1–TS3.
(TS1). For all C
1
> 0, for all H
1
⊂ H
2
⊂ H
0
open subgroups, there exists an
α ∈
˜
Λ
H
1
with
v(α) > −C
1
and
¸
τ∈H
2
/H
1
τ(α) = 1.
133
134 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
(In Faltings’ terminology,
˜
Λ/
˜
Λ
H
0
is called almost ´etale.)
(TS2). Tate’s normalized trace maps: there exists C
2
> 0 such that for
all open subgroups H ⊂ H
0
, there exist n(H) ∈ N and (Λ
H,n
)
n≥n(H)
, an
increasing sequence of sub Z
p
algebras of
˜
Λ
H
and maps
R
H,n
:
˜
Λ
H
−→Λ
H,n
satisfying the following conditions:
(a) if H
1
⊂ H
2
, then Λ
H
2
,n
= (Λ
H
1
,n
)
H
2
, and R
H
1
,n
= R
H
2
,n
on
˜
Λ
H
2
;
(b) for all g ∈ G
0
,
g(Λ
H,n
) = Λ
gHg
−1
,n
g ◦ R
H,n
= R
gHg
−1
,n
◦ g;
(c) R
H,n
is Λ
H,n
linear and is equal to Id on Λ
H,n
;
(d) v(R
H,n
(x)) v(x) −C
2
if n n(H) and x ∈
˜
Λ
H
;
(e) lim
n→+∞
R
H,n
(x) = x.
(TS3). There exists C
3
, such that for all open subgroups G ⊂ G
0
, H =
G ∩ H
0
, there exists n(G) n(H) such that if n n(G), γ ∈ G/H and
v(γ) = v
p
(log χ(γ)) n, then γ −1 is invertible on X
H,n
= (R
H,n
−1)
˜
Λ and
v((γ −1)
−1
x) v(x) −C
3
for x ∈ X
H,n
.
Remark. R
H,n
◦ R
H,n
= R
H,n
, so
˜
Λ
H
= Λ
H,n
⊕X
H,n
.
8.1.2 Example : the ﬁeld C
p
Theorem 8.1.2. The quadruple (
˜
Λ = C
p
, v = v
p
, G
0
= G
Qp
and χ=the
cyclotomic character) satisﬁes (TS1), (TS2), (TS3).
Proof. (TS1): In Fontaine’s course, we know that for any Q
p
⊂ K ⊂ L such
that [L : Q
p
] < +∞, then
v
p
(d
Ln/Kn
) →0 as n →+∞.
The proof showed that v
p
(γ(π
n
) − π
n
) → 0 as n → +∞, where π
n
is a
uniformizer of L
n
and γ ∈ Gal(L
n
/K
n
) = Gal(L
∞
/K
∞
) when n 0. We
also have
Tr
L∞/K∞
= Tr
Ln/Kn
8.1. TATESEN’S CONDITIONS 135
on L
n
if n 0 and
Tr
Ln/Kn
(O
Ln
) ⊃ d
Ln/Kn
¸
O
Kn
,
thus Tr
L∞/K∞
(O
L∞
) contains elements with v
p
as small as we want. Take
x ∈ O
L∞
and let α =
x
Tr
L∞/K∞
(x)
, then
¸
τ∈H
K
/H
L
τ(α) = Tr
L∞/K∞
(α) = 1.
Then for all C
1
> 0, we can ﬁnd x ∈ O
L∞
such that v
p
(Tr
L∞/K∞
(x)) is small
enough, thus v
p
(α) > −C
1
.
(TS2) and (TS3): By AxSenTate, C
H
K
p
=
ˆ
K
∞
, let Λ
H
K
,n
= K
n
, and
R
H
K
,n
= p
−k
Tr
K
n+k
/Kn
on K
n+k
.
If K = Q
p
, R
H
K
,n
= R
n
, that’s what we did in last chapter. We are going
to use what we know about R
n
.
For G = G
K
, then H = H
K
, choose m big enough such that for any
n m, v
p
(d
Kn/Fn
) is small and [K
∞
: F
∞
] = [K
n
: F
n
] = d. Let ¦e
1
, ..., e
d
¦
be a basis of O
Kn
over O
Fn
and ¦e
∗
1
, ..., e
∗
d
¦ be the dual basis of K
n
over F
n
for the trace map (x, y) → Tr
Kn/Fn
(xy). This implies that ¦e
∗
1
, ..., e
∗
d
¦ is a
basis of d
−1
Kn/Fn
and v
p
(e
∗
i
) −v
p
(d
Kn/Fn
) are small. Any x ∈ K
∞
can be
written as
x =
d
¸
i=1
Tr
K∞/K
(xe
i
)e
∗
i
,
then
inf
i
v
p
(Tr
K∞/F∞
(xe
i
)) ≥ v
p
(x) ≥ inf
i
v
p
(Tr
K∞/F∞
(xe
i
)) −v
p
(d
Kn/Fn
),
and
R
H
K
,n
(x) =
d
¸
i=1
R
n
(Tr
K∞/F∞
(xe
i
))e
∗
i
, n m.
So use what we know about R
n
to conclude.
Remark. By the same method as Corollary 7.5.7, we get
(i) H
1
(Γ,
ˆ
K
∞
)
∼
= K, where the isomorphism is given by x ∈ K −→(γ →
x log χ(γ)).
(ii) H
1
(Γ,
ˆ
K
∞
(η)) = 0 if η is of inﬁnite order.
136 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
8.2 Sen’s method
Proposition 8.2.1. Assume
˜
Λ verifying (TS1), (TS2) and (TS3). Let σ →
U
σ
be a continuous cocycle from G
0
to GL
d
(
˜
Λ). If G ⊂ G
0
is an open normal
subgroup of G
0
such that v(U
σ
− 1) > 2C
2
+ 2C
3
for any σ ∈ G. Set
H = G ∩ H
0
, then there exists M ∈ GL
d
(
˜
Λ) with v(M − 1) > C
2
+ C
3
such
that
σ −→V
σ
= M
−1
U
σ
σ(M)
satisﬁes V
σ
∈ GL
d
(Λ
H,n(G)
) and V
σ
= 1 if σ ∈ H.
Example 8.2.2. Example of Sen: For the case
˜
Λ = C
p
, for U
σ
a 1cocycle
on G
K
with values in GL
d
(C
p
), there exists [L : K] < ∞, such that U
σ
is
cohomologous to a cocycle that which is trivial on H
L
and with values in
GL
d
(L
n
) for some n.
The proof of Proposition 8.2.1 needs three Lemmas below. It is technical,
but the techniques come over again and again.
8.2.1 Almost ´etale descent
Lemma 8.2.3. If
˜
Λ satisﬁes (TS1), a > 0, and σ →U
σ
is a 1cocycle on H
open in H
0
and
v(U
σ
−1) a for any σ ∈ H,
then there exists M ∈ GL
d
(
˜
Λ) such that
v(M −1)
a
2
, v(M
−1
U
σ
σ(M) −1) a + 1.
Proof. The proof is approximating Hilbert’s Theorem 90.
Fix H
1
⊂ H open and normal such that v(U
σ
− 1) a + 1 + a/2 for
σ ∈ H
1
, which is possible by continuity. Because
˜
Λ satisﬁes (TS1), we can
ﬁnd α ∈
˜
Λ
H
1
such that
v(α) −a/2,
¸
τ∈H/H
1
τ(α) = 1.
Let S ⊂ H be a set of representatives of H/H
1
, denote M
S
=
¸
σ∈S
σ(α)U
σ
, we
have M
S
−1 =
¸
σ∈S
σ(α)(U
σ
−1), this implies v(M
S
−1) a/2 and moreover
M
−1
S
=
+∞
¸
n=0
(1 −M
S
)
n
,
8.2. SEN’S METHOD 137
so we have v(M
−1
S
) 0 and M
S
∈ GL
d
(
˜
Λ).
If τ ∈ H
1
, then U
στ
−U
σ
= U
σ
(σ(U
τ
) −1). Let S
t
⊂ H be another set of
representatives of H/H
1
, so for any σ
t
∈ S
t
, there exists τ ∈ H
1
and σ ∈ S
such that σ
t
= στ, so we get
M
S
−M
S
=
¸
σ∈S
σ(α)(U
σ
−U
στ
) =
¸
σ∈S
σ(α)U
σ
(1 −σ(U
τ
)),
thus
v(M
S
−M
S
) a + 1 +a/2 −a/2 = a + 1.
For any τ ∈ H,
U
τ
τ(M
S
) =
¸
σ∈S
τσ(α)U
τ
τ(U
σ
) = M
τS
.
Then
M
−1
S
U
τ
τ(M
S
) = 1 + M
−1
S
(M
τS
−M
S
),
with v(M
−1
S
(M
τS
− M
S
)) ≥ a + 1. Take M = M
S
for any S, we get the
result.
Corollary 8.2.4. Under the same hypotheses as the above lemma, there
exists M ∈ GL
d
(
˜
Λ) such that
v(M −1) a/2, M
−1
U
σ
σ(M) = 1, ∀ σ ∈ H.
Proof. Repeat the lemma (a →a+1 →a+2 → ), and take the limits.
Exercise. Assume
˜
Λ satisﬁes (TS1), denote by
˜
Λ
+
= ¦x ∈
˜
Λ[v(x) 0¦. Let
M be a ﬁnitely generated
˜
Λ
+
module with semilinear action of H, an open
subgroup of H
0
. Then H
i
(H, M) is killed by any x ∈
˜
Λ
H
with v(x) > 0.
Hint: Adapt the proof that if L/K is ﬁnite Galois and M is a Lmodule with
semilinear action of Gal(L/K), then H
i
(Gal(L/K), L) = 0 for all i 1. Let
α ∈ L such that Tr
L/K
(α) = 1. For any c(g
1
, , g
n
) an ncocycle, let
c
t
(g
1
, , g
n−1
) =
¸
h∈Gal(L/K)
g
1
g
n−1
h(α)c(g
1
, , g
n−1
, h),
then dc
t
= c.
138 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
Theorem 8.2.5. (i) The map x −→(g →x log χ(g)) gives an isomorphism
K
∼
→H
1
(G
K
, C
p
).
(ii) If η : G
K
→Γ
K
→Q
∗
p
is of inﬁnite order, then H
1
(G
K
, C
p
(η)) = 0.
Proof. Using the inﬂation and restriction exact sequence
0 −→H
1
(Γ
K
, C
p
(η)
H
K
)
inf
−→H
1
(G
K
, C
p
(η))
res
−→H
1
(H
K
, C
p
(η))
Γ
K
.
by the above exercise, H
1
(H
K
, C
p
(η))
Γ
K
= 0, then the inﬂation map is actu
ally an isomorphism. We have C
p
(η)
H
K
=
ˆ
K
∞
(η), and use Corollary 7.5.7.
In fact
K = H
1
(Γ
K
,
ˆ
K
∞
) = H
1
(Γ
K
, K) = Hom(Γ, K) = K log χ,
the last equality is because Γ
K
is procyclic.
8.2.2 Decompletion
Now recall that we have the continuous character: G
0
χ
−→Z
∗
p
, H
0
= Ker χ.
¯
Λ
is complete for v, with continuous action of G
0
. H is an open subgroup of H
0
,
and we have the maps:R
H,n
:
¯
Λ
H
→Λ
H,n
. By (TS2), v(R
H,n
(x)) ≥ v(x)−C
2
;
and by (TS3), v((γ−1)
−1
x) ≥ v(x)−C
3
, if R
H,n
(x) = 0 and v
p
(log χ(γ)) ≤ n.
We can use these properties to reduce to something reasonable.
Lemma 8.2.6. Assume given δ > 0, b ≥ 2C
2
+ 2C
3
+ δ, and H ⊂ H
0
is
open. Suppose n ≥ n(H), γ ∈ G/H with n(γ) ≤ n, U = 1 + U
1
+ U
2
with
U
1
∈ M
d
(Λ
H,n
), v(U
1
) ≥ b −C
2
−C
3
U
2
∈ M
d
(
¯
Λ
H
), v(U
2
) ≥ b.
Then, there exists M ∈ GL
d
(
¯
Λ
H
), v(M −1) ≥ b −C
2
−C
3
such that
M
−1
Uγ(M) = 1 + V
1
+ V
2
,
with
V
1
∈ M
d
(Λ
H,n
), v(V
1
) ≥ b −C
2
−C
3
),
V
2
∈ M
d
(
¯
Λ
H
), v(V
2
) ≥ b + δ.
8.2. SEN’S METHOD 139
Proof. Using (TS2) and (TS3), one gets U
2
= R
H,n
(U
2
) + (1 −γ)V , with
v(R
H,n
(U
2
)) ≥ v(U
2
) −C
2
, v(V ) ≥ v(U
2
) −C
2
−C
3
.
Thus,
(1 +V )
−1
Uγ(1 + V ) = (1 −V + V
2
− )(1 + U
1
+ U
2
)(1 + γ(V ))
= 1 + U
1
+ (γ −1)V + U
2
+ (terms of degree ≥ 2)
Let V
1
= U
1
+R
H,n
(U
2
) ∈ M
d
(Λ
H,n
) and W be the terms of degree ≥ 2. Thus
v(W) ≥ 2(b −C
2
−C
3
) ≥ b + δ. So we can take M = 1 + V, V
2
= W.
Corollary 8.2.7. Keep the same hypotheses as in Lemma 8.2.6. Then there
exists M ∈ GL
d
(
¯
Λ
H
), v(M − 1) ≥ b − C
2
− C
3
such that M
−1
Uγ(M) ∈
GL
d
(Λ
H,n
).
Proof. Repeat the lemma (b →b+δ →b+2δ → ), and take the limit.
Lemma 8.2.8. Suppose H ⊂ H
0
is an open subgroup, i ≥ n(H), γ ∈ G/H,
n(γ) ≥ i and B ∈ GL
d
(
¯
Λ
H
). If there exist V
1
, V
2
∈ GL
d
(Λ
H,i
) such that
v(V
1
−1) > C
3
, v(V
2
−1) > C
3
, γ(B) = V
1
BV
2
,
then B ∈ GL
d
(Λ
H,i
).
Proof. Take C = B − R
H,i
(B). We have to prove C = 0. Note that C has
coeﬃcients in X
H,i
= (1 − R
H,i
)
¯
Λ
H
, and R
H,i
is Λ
H,i
linear and commutes
with γ. Thus,
γ(C) −C = V
1
CV
2
−C = (V
1
−1)CV
2
+ V
1
C(V
2
−1) −(V
1
−1)C(V
2
−1)
Hence, v(γ(C) − C) > v(C) + C
3
. By (TS3), this implies v(C) = +∞, i.e.
C = 0.
Proof of Proposition 8.2.1. Let σ →U
σ
be a continuous 1cocycle on G
0
with
values in GL
d
(
¯
Λ). Choose an open normal subgroup G of G
0
such that
inf
σ∈G
v(U
σ
−1) > 2(C
2
+ C
3
).
By Lemma 8.2.3, there exists M
1
∈ GL
d
(
¯
Λ), v(M
1
− 1) > 2(C
2
+ C
3
) such
that σ →U
t
σ
= M
−1
1
U
σ
σ(M
1
) is trivial in H = G∩ H
0
(In particular, it has
values in GL
d
(
¯
Λ
H
)).
140 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
Now we pick γ ∈ G/H with n(γ) = n(G). In particular, we want n(G)
big enough so that γ is in the center of G
0
/H. Indeed, the center is open,
since in the exact sequence:
1 →H
0
/H →G
0
/H →G/H →1,
G/H · Z
p
(ﬁnite), and H
0
/H is ﬁnite. So we are able to choose such a
n(G).
Then we have v(U
t
γ
) > 2(C
2
+ C
3
), and by Corollary 8.2.7, there exists
M
2
∈ GL
d
(
¯
Λ
H
) satisfying
v(M
2
−1) > C
2
+ C
3
and M
−1
2
U
t
γ
γ(M
2
) ∈ GL
d
(Λ
H,n(G)
).
Take M = M
1
M
2
, then the cocycle
σ →V
σ
= M
−1
U
σ
σ(M)
a cocycle trivial on H with values in GL
d
(
¯
Λ
H
), and we have
v(V
γ
−1) > C
2
+ C
3
and V
γ
∈ GL
d
(Λ
H,n(G)
).
This implies V
σ
comes by inﬂation from a cocycle on G
0
/H.
The last thing we want to prove is V
τ
∈ GL
d
(Λ
H,n(G)
) for any τ ∈ G
0
/H.
Note that γτ = τγ as γ is in the center, so
V
τ
τ(V
γ
) = V
τγ
= V
γτ
= V
γ
γ(V
τ
)
which implies γ(V
τ
) = V
−1
γ
V
τ
τ(V
γ
). Apply Lemma 8.2.8 with V
1
= V
−1
γ
, V
2
=
τ(V
γ
), then we obtain what we want.
8.2.3 Applications to padic representations
Proposition 8.2.9. Let T be a free Z
p
representation of G
0
, k ∈ N, v(p
k
) >
2C
2
+ 2C
3
, and suppose G ⊂ G
0
is an open normal subgroup acting trivially
on T/p
k
T, and H = G ∩ H
0
. Let n ∈ N, n ≥ n(G). Then there exists a
unique D
H,n
(T) ⊂
¯
Λ ⊗T, a free Λ
H,n
module of rank d, such that:
(i) D
H,n
(T) is ﬁxed by H, and stable by G;
(ii)
¯
Λ ⊗
Λ
H,n
D
H,n
(T)
∼
−→
¯
Λ ⊗T;
(iii) there exists a basis ¦e
1
, . . . , e
d
¦ of D
H,n
over Λ
H,n
such that if γ ∈
G/H, then v(V
γ
−1) > C
3
, V
γ
being the matrix of γ.
8.3. OVERCONVERGENT (ϕ, Γ)MODULES 141
Proof. Translation of Proposition 8.2.1, by the correspondence
¯
Λrepresentations of G
0
←→H
1
(G
0
, GL
d
(
¯
Λ)).
For the uniqueness, one uses Lemma 8.2.8.
Remark. H
0
acts through H
0
/H (which is ﬁnite) on D
H,n
(T). If Λ
H,n
is ´etale
over Λ
H
0
,n
(the case in applications), and then D
H
0
,n
(T) = D
H,n
(T)
(H
0
/H)
, is
locally free over Λ
H
0
,n
(in most cases it is free), and
Λ
H,n
Λ
H
0
,n
D
H
0
,n
(T)
∼
−→D
H,n
(T).
Example 8.2.10. For
¯
Λ = C
p
, let V be a Q
p
representation of G
K
for
[K : Q
p
] < +∞, T ⊂ V be a stable lattice. Then
D
Sen,n
(V ) := D
H
K
,n
(T)
is a K
n
vector space of dimension d = dim
Qp
V with a linear action of Γ
Kn
.
Sen’s operator is deﬁned as follows:
Θ
Sen
=
log γ
log χ(γ)
, where γ ∈ Γ
Kn
, log χ(γ) = 0.
It is easy to see:
Proposition 8.2.11. V is HodgeTate if and only if Θ
Sen
is semisimple,
and the eigenvalues lie in Z. These eigenvalues are the HodgeTate weights
of V .
Remark. For general V , the eigenvalues of Θ
Sen
are the generalized Hodge
Tate weights of V .
8.3 Overconvergent (ϕ, Γ)modules
8.3.1 Overconvergent elements
Deﬁnition 8.3.1. (i) For x =
+∞
¸
i=0
p
i
[x
i
] ∈
¯
A, x
i
∈
¯
E = Fr R, k ∈ N, deﬁne
w
k
(x) := inf
i≤k
v
E
(x
i
) (One checks easily that w
k
(x) ≥ v
E
(α), α ∈
¯
E, if and
only if [α]x ∈
¯
A
+
+ p
k+1
¯
A).
142 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
(ii) For a real number r > 0, deﬁne
v
(0, r]
(x) := inf
k∈N
w
k
(x) +
k
r
= inf
k∈N
v
E
(x
k
) +
k
r
∈ R ∪ ¦±∞¦.
(iii)
¯
A
(0, r]
:= ¦x ∈
¯
A : lim
k→+∞
v
E
(x
k
) +
k
r
= lim
k→+∞
w
E
(x
k
) +
k
r
= +∞¦.
Proposition 8.3.2.
¯
A
(0, r]
is a ring and v = v
(0, r]
satisﬁes the following
properties:
(i) v(x) = +∞⇔x = 0;
(ii) v(xy) ≥ v(x) + v(y);
(iii) v(x + y) ≥ inf(v(x), v(y));
(iv) v(px) = v(x) +
1
r
;
(v) v([α]x) = v
E
(α) + v(x) if α ∈
¯
E;
(vi) v(g(x)) = v(x) if g ∈ G
Qp
;
(vii) v
(0, p
−1
r]
(ϕ(x)) = pv
(0, r]
(x).
Proof. Exercise.
Lemma 8.3.3. Given x ∈
+∞
¸
k=0
p
k
[x
k
] ∈
¯
A, the following conditions are equiv
alent:
(i)
+∞
¸
k=0
p
k
[x
k
] converges in B
+
dR
;
(ii)
+∞
¸
k=0
p
k
x
(0)
k
converges in C
p
;
(iii) lim
k→+∞
(k + v
E
(x
k
)) = +∞;
(iv) x ∈
¯
A
(0, 1]
.
Proof. (iii) ⇔ (iv) is by deﬁnition of
¯
A
(0, r]
. (ii) ⇔ (iii) is by deﬁnition of
v
E
. (i) ⇒ (ii) is by the continuity of θ : B
+
dR
→ C
p
. So it remains to show
(ii) ⇒(i).
Write ¯ p = (p, p
1/p
, ) ∈
¯
E
+
, then ξ = [¯ p] −p is a generator of Ker θ∩
˜
A
+
.
We know
a
k
= k + [v
E
(x
k
)] →+∞ if k →+∞.
Write x
k
= ¯ p
k−a
k
y
k
, then y
k
∈
¯
E
+
. We have
p
k
[x
k
] =
p
¯ p
k
[¯ p]
a
k
[y
k
] = p
a
k
(1 +
ξ
p
)
a
k
−k
[y
k
].
8.3. OVERCONVERGENT (ϕ, Γ)MODULES 143
Note that p
k
(1 +
ξ
p
)
a
k
−k
∈ p
a
k
−m
¯
A
+
+(Ker θ)
m+1
for all m. Thus, a
k
→+∞
implies that p
k
[x
k
] →0 ∈ B
+
dR
/(Ker θ)
m+1
for every m, and therefore also in
B
+
dR
by the deﬁnition of the topology of B
+
dR
.
Remark. We just proved
¯
A
(0,1]
:= B
+
dR
∩
¯
A, and we can use
ϕ
−n
:
¯
A
(0,p
−n
]
∼
−→
¯
A
(0, 1]
to embed
¯
A
(0,r]
in B
+
dR
, for r ≥ p
−n
.
Deﬁne
¯
A
†
:=
¸
r>0
¯
A
(0, r]
= ¦x ∈
¯
A : ϕ
−n
(x) converges in B
+
dR
for n 0¦.
Lemma 8.3.4. x ∈
+∞
¸
k=0
p
k
[x
k
] is a unit in
¯
A
(0,r]
if and only if x
0
= 0 and
v
E
(
x
k
x
0
) > −
k
r
for all k ≥ 1.
Proof. Exercise. Just adapt the proof of Gauss Lemma.
Set
¯
B
(0, r]
=
¯
A
(0,r]
[
1
p
] =
¸
n∈N
p
−n
¯
A
(0,r]
,
endowed with the topology of inductive limit, and
¯
B
†
=
¸
r>0
¯
B
(0, r]
,
again with the topology of inductive limit.
Theorem 8.3.5.
¯
B
†
is a subﬁeld of
¯
B, stable by ϕ and G
Qp
, both acting
continuously.
¯
B
†
is called the ﬁeld of overconvergent elements. We are going to prove
elements of D(V )
ψ=1
are overconvergent.
Deﬁnition 8.3.6. (i) B
†
=
¯
B
†
∩ B, A
†
=
¯
A
†
∩ B (so B
†
is a subﬁeld of B
stable by ϕ and G
Qp
), A
(0, r]
=
¯
A
(0, r]
∩ B.
(ii) If K/Q
p
is a ﬁnite extension and Λ ∈ ¦
¯
A
†
,
¯
B
†
, A
†
, B
†
, A
(0, r]
, B
(0, r]
¦,
deﬁne Λ
K
= Λ
H
K
. For example A
(0, r]
K
=
¯
A
(0, r]
∩ A
K
.
(iii) If Λ ∈ ¦A, B, A
†
, B
†
, A
(0, r]
, B
(0, r]
¦, and n ∈ N, deﬁne Λ
K,n
= ϕ
−n
(Λ
K
) ⊂
¯
B.
144 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
We now want to make A
(0, r]
K
more concrete. Let F
t
⊂ K
∞
be the maximal
unramiﬁed extension of Q
p
, ¯ π
K
be a uniformizer of E
K
= k
F
((¯ π
K
)),
¯
P
K
∈
E
F
[X] be a minimal polynomial of ¯ π
K
. Let P
K
∈ A
+
F
[X] (note that A
+
F
=
O
F
[[π]]) be a lifting of
¯
P
K
. By Hensel’s lemma, there exists a unique π
K
∈
A
K
such that P
K
(π
K
) = 0 and ¯ π
K
= π
K
mod p. If K = F
t
, we take π
K
= π.
Lemma 8.3.7. If we deﬁne
r
K
=
1, if E
K
/E
Qp
is unramiﬁed,
(2v
E
(d
E
K
/E
Qp
))
−1
, otherwise .
then π
K
and P
t
K
(π
K
) are units in A
(0, r]
K
for all 0 < r < r
K
.
Proof. The proof is technical but not diﬃcult and is left to the readers.
Proposition 8.3.8. (i) A
K
= ¦
¸
n∈N
a
n
π
n
K
: a
n
∈ O
F
, lim
n→−∞
v
p
(a
n
) = +∞¦;
(ii) A
(0, r]
K
= ¦
¸
n∈N
a
n
π
n
K
: a
n
∈ O
F
, lim
n→−∞
(v
p
(a
n
) + rnv
E
(¯ π
K
)) = +∞¦.
So f → f(π
K
) is an isomorphism from bounded analytic functions on the
annulus 0 < v
p
(T) ≤ rv
E
(¯ π
K
) to the ring B
(0, r]
K
.
Proof. The technical but not diﬃcult proof is again left as an exercise. See
CherbonnierColmez Invent. Math. 1998.
Corollary 8.3.9. (i) A
(0, r]
K
is a principal ideal domain;
(ii) If L/K is a ﬁnite Galois extension, then A
(0, r]
L
is an ´etale extension
of A
(0, r]
K
if r < r
L
, and the Galois group is nothing but H
K
/H
L
.
Deﬁne ¯ π
n
= ϕ
−n
(π), ¯ π
K,n
= ϕ
−n
(π
K,n
).
Proposition 8.3.10. (i) If p
n
r
K
≥ 1, θ(¯ π
K,n
) is a uniformizer of K
n
;
(ii) ¯ π
K,n
∈ K
n
[[t]] ⊂ B
+
dR
.
Proof. First by deﬁnition
¯ π
n
= [ε
1/p
n
] −1 = ε
(n)
e
t/p
−1 ∈ F
n
[[t]] ⊂ B
+
dR
(for [ε
1/p
n
] = ε
(n)
e
t/p
n
: the θ value of both sides is ε
(n)
, and the p
n
th power
of both side is [ε] = e
t
(recall t = log[ε])). This implies the proposition in
the unramiﬁed case.
8.3. OVERCONVERGENT (ϕ, Γ)MODULES 145
For the ramiﬁed case, we proceed as follows.
By the deﬁnition of E
K
, π
K,n
= θ(¯ π
K,n
) is a uniformizer of K
n
mod a =
¦x : v
p
(x) ≥
1
p
¦. Write ω
n
be the image of π
K,n
in K
n
mod a. So we just have
to prove π
K,n
∈ K
n
.
Write
P
K
(x) =
d
¸
i=0
a
i
(π)x
i
, a
i
(π) ∈ O
F
[[π]].
Deﬁne
P
K,n
(x) =
d
¸
i=0
a
ϕ
−n
i
(π
n
)x
i
,
then P
K,n
(π
K,n
) = θ(ϕ(P
K
(π
K
))) = 0. Then we have v
p
(P
K,n
(ω
n
)) ≥
1
p
and
v
p
(P
t
K,n
(ω
n
)) =
1
p
n
v
E
(P
t
K
(¯ π
K
)) =
1
p
n
v
E
(d
E
K
/E
Qp
) <
1
2p
if p
n
r
K
> 1.
Then one concludes by Hensel’s Lemma that π
K,n
∈ K
n
.
For (ii) , one uses Hensel’s Lemma in K
n
[[t]] to conclude ¯ π
K,n
∈ K
n
[[t]].
Corollary 8.3.11. If 0 < r < r
K
and p
n
r ≥ 1, ϕ
−n
(A
(0, r]
K
) ⊂ K
n
[[t]] ⊂ B
+
dR
.
8.3.2 Overconvergent representations
Suppose V is a free Z
p
representation of rank d of G
K
. Let
D
(0, r]
:= (A
(0, r]
⊗
Zp
V )
H
K
⊂ D(V ).
This is a A
(0, r]
K
module stable by Γ
K
. As for ϕ, we have
ϕ : D
(0, r]
(V ) −→D
(0, p
−1
r]
(V ).
Deﬁnition 8.3.12. V is overconvergent if there exists an r
V
> 0, r
V
≤ r
K
such that
A
K
A
(0, r
V
]
K
D
(0, r
V
]
(V )
∼
−→D(V ).
By deﬁnition, it is easy to see for all 0 < r < r
V
,
D
(0, r]
(V ) = A
(0, r]
K
A
(0, r
V
]
K
D
(0, r
V
]
(V ).
146 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
Proposition 8.3.13. If V is overconvergent, then there exists a C
V
such
that if γ ∈ Γ
K
, n(γ) = v
p
(log(χ(γ))) and r < inf¦p
−1
r
V
, p
−n(γ)
¦, then γ − 1
is invertible in D
(0, r]
(V )
ψ=0
and
v
(0, r]
((γ −1)
−1
x) ≥ v
(0, r]
(x) −C
V
−p
n(γ)
v
E
(¯ π).
Proof. Write x =
p−1
¸
i=1
[ε]
i
ϕ(x
i
) and adapt the proof of the same statement as
in the characteristic p case. One has to use the fact that [ε]
ip
n
−1
is a unit in
A
(0, r]
K
if r < p
−n
and i ∈ Z
∗
p
.
Remark. This applies to (A
(0, r]
K
)
ψ=0
.
Theorem 8.3.14 (Main Theorem). (i) All (free Z
p
or Q
p
) representations
of G
K
are overconvergent.
(ii) D(V )
ψ=1
⊂ D
(0, r
V
]
(V ).
Sketch of Proof. (ii) is just because ψ improves convergence.
(i) follows from Sen’s method applied to
¯
Λ =
¯
A
(0, 1]
, v = v
(0, 1]
, G
0
= G
K
, Λ
H
K,n
= ϕ
−n
(A
(0, 1]
K
).
Now we show how to check the three conditions.
(TS1). Let L ⊃ K ⊃ Q
p
be ﬁnite extensions, for α = [¯ π
L
](
¸
τ∈H
K
/H
L
τ([¯ π
L
]))
−1
,
then for all n,
¸
τ∈H
K
/H
L
τ(ϕ
−n
(α)) = 1,
and
lim
n→+∞
v
(0, 1]
(ϕ
−n
(α)) = 0.
(TS2). First Λ
H
K
,n
= A
(0,1]
K,n
. Suppose p
n
r
K
≥ 1. We can deﬁne R
K, n
by
the following commutative diagram:
R
K,n
:
¯
A
(0,1]
K
A
(0,1]
K,n
A
(0,1]
K,n+k
ϕ
n
◦ψ
n+k
◦ϕ
n+k
x
x
x
x
x
x
x
x
?
8.3. OVERCONVERGENT (ϕ, Γ)MODULES 147
One veriﬁes that ϕ
−n
◦ψ
n+k
◦ϕ
n+k
does not depend on the choice of k, using
the fact ψϕ = Id. Then the proof is entirely parallel to that for C
p
with ψ
in the role of p
−1
Tr
F
n+1
/Fn
and ¯ π
n+k
in the role of π
n+k
.
(TS3). For an element x such that R
K,n
(x) = 0, write
x =
+∞
¸
i=0
R
∗
K,n
(x), where R
∗
K,n
(x) ∈ ϕ
−(n+i+1)
((A
(0,p
−(n+i+1)
]
K
)
ψ=0
).
Then just apply Proposition 8.3.13 on (A
(0,p
−(n+i+1)
]
K
)
ψ=0
.
Now Sen’s method implies that there exists an n and a A
(0,1]
K,n
module
D
(0, 1]
K,n
⊂
¯
A
(0, 1]
¸
V such that
¯
A
(0, 1]
⊗
A
(0, 1]
K,n
D
(0, 1]
K,n
∼
−→
¯
A
(0, 1]
⊗V.
Play with (TS3) just like Lemma 8.2.8, one concludes that D
(0, 1]
K,n
⊂ ϕ
−n
(D(V ))
and ϕ
n
(D
(0, 1]
K,n
) ⊂ D
(0, p
−n
]
(V ). We can just take r
V
= n.
8.3.3 padic Hodge theory and (ϕ, Γ)modules
Suppose we are given a representation V , 0 < r < r
V
and n such that
p
n
r > 1. Then we have
ϕ
−n
(D
(0, r]
(V )) →B
+
dR
⊗V
θ
−→C
p
⊗V
and
ϕ
−n
(A
(0, r]
K
) →K
n
[[t]]
θ
−→K
n
.
So we get the maps
θ ◦ ϕ
−n
: K
n
⊗
A
(0, 1]
K
D
(0, r]
(V ) −→C
p
⊗V (8.1)
and
ϕ
−n
: t
i
K
n
[[t]] ⊗
A
(0, r]
K
D
(0, r]
(V ) −→t
i
B
+
dR
⊗V, ∀ i ∈ Z. (8.2)
Theorem 8.3.15. There exists an n(V ) ∈ N such that if n ≥ n(V ), then we
have
(i) the image of θ ◦ ϕ
−n
in (8.1) is exactly D
Sen, n
(V );
(ii) Fil
i
D
dR
(V ) = (Im ϕ
−n
)
Γ
K
in (8.2) for all i;
(iii) D
dR
(V ) =
K
n
((t))
¸
A
(0, r]
K
D
(0, r]
(V )
Γ
K
.
148 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
Let K/Q
p
be a ﬁnite extension, and deﬁne
B
†
K
= ¦F(π
K
) : F is a bounded analytic function on 0 < v
p
(t) ≤ r(F), r(F) > 0¦,
B
†
rig, K
= ¦F(π
K
) : F is an analytic function on 0 < v
p
(t) ≤ r(F), r(F) > 0¦
(this last ring is the Robba ring in the variable π
K
), and
B
†
log, K
= B
†
rig, K
[log π
K
].
Extend ϕ, Γ
K
by continuity on B
†
rig, K
, and set
ϕ(log π
K
) = p log π
K
+ log
ϕ(π
K
)
π
p
K
,
γ(log π
K
) = log π
K
+ log
γ(π
K
)
π
K
where log
ϕ(π
K
)
π
p
K
∈ B
†
K
and log
γ(π
K
)
π
K
∈ B
†
rig, K
. Let
N = −
1
v
E
(¯ π
K
)
d
d log π
K
.
Theorem 8.3.16 (Berger). For
D
†
(V ) = (B
†
⊗V )
H
K
=
¸
r>0
D
(0, r]
(V ),
if V is semistable, then
B
†
log, K
1
t
⊗
K
0
D
st
(V ) = B
†
log, K
1
t
⊗
B
†
K
D
†
(V )
is an isomorphism of (ϕ, N, Γ
K
)modules. This implies that D
st
(V ) is the
invariant under Γ
K
.
8.3.4 A map of the land of the rings
The following nice picture outlines most of the objects that we have discussed
till now and that we shall have to discover more about in the future.
8.4. EXPLICIT RECIPROCITY LAWS AND PADIC LFUNCTIONS149
padic
Hodge
Theory
B
+
dR
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
˜
B
†
log
ϕ
−n
˜
B
+
log
B
+
st
(ϕ, Γ)
modules
˜
B
†
rig
˜
B
+
rig
B
+
cris
B
˜
B
˜
B
†
˜
B
+
θ
C
p
A
1
p
mod p
˜
A
˜
A
†
˜
A
+
O
Cp
E
˜
E
˜
E
†
˜
E
+
O
Cp
/p
_ _ _
_ _ _
_ _
_ _
where
˜
B
+
rig
=
¸
n
ϕ
n
(B
+
cris
),
˜
B
+
log
=
¸
n
ϕ
n
(B
+
st
).
Note that most arrows from (ϕ, Γ)modules to padic Hodge theory are in
the wrong direction, but overconvergence and Berger’s theorem allow us to
go backwards.
8.4 Explicit reciprocity laws and padic Lfunctions
8.4.1 Galois cohomology of B
dR
Suppose K is a ﬁnite extension of Q
p
. Recall that we have the following:
Proposition 8.4.1. For k ∈ Z, then
(i) if k = 0, then H
i
(G
K
, C
p
(k)) = 0 for all i
(ii) if k = 0, then H
i
(G
K
, C
p
) = 0 for i ≥ 2, H
0
(G
K
, C
p
) = K,
and H
1
(G
K
, C
p
) is a 1dimensional Kvector space generated by log χ ∈
H
1
(G
K
, Q
p
). (i.e, the cup product x → x ∪ log χ gives an isomorphism
H
0
(G
K
, C
p
) · H
1
(G
K
, C
p
)).
150 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
Remark. This has been proved for i ≤ 1. For i ≥ 2, H
i
(H
K
, C
p
(k)) = 0 by
using the same method as for H
1
. Then just use the exact sequence
1 −→H
K
−→G
K
−→Γ
K
−→1
and HochschildSerre spectral sequence to conclude.
Proposition 8.4.2. Suppose i < j ∈ Z ∪ ¦±∞¦, then if i ≥ 1 or j ≤ 0,
H
1
(G
K
, t
i
B
+
dR
/t
j
B
+
dR
) = 0;
if i ≤ 0 and j > 0, then x →x ∪ log χ gives an isomorphism
H
0
(G
K
, t
i
B
+
dR
/t
j
B
+
dR
)(· K)
∼
−→H
1
(G
K
, t
i
B
+
dR
/t
j
B
+
dR
).
Proof. Use the long exact sequence in continuous cohomology attached to
the exact sequence
0 −→t
i+n
C
p
(· C
p
(i + n)) −→t
i
B
+
dR
/t
n+i+1
B
+
dR
−→t
i
B
+
dR
/t
i+n
B
+
dR
−→0,
and use induction on j−i (note that in the base step j = i+1, t
i
B
+
dR
/t
j
B
+
dR
∼
=
C
p
(i)), and Proposition 8.4.1 to do the computation. This concludes for
the case where i, j are ﬁnite. For the general case, one proves it by taking
limits.
8.4.2 BlochKato’s dual exponential maps
Let V be a de Rham representation of G
K
, we have
B
dR
⊗
Qp
V
∼
= B
dR
⊗
K
D
dR
(V ) = H
0
(G
K
, B
dR
⊗V )
and
H
1
(G
K
, B
dR
⊗V ) = H
1
(G
K
, B
dR
⊗
K
D
dR
(V )) = H
1
(G
K
, B
dR
) ⊗
K
D
dR
(V ).
So we get an isomorphism
D
dR
(V )
∼
−→H
1
(G
K
, B
dR
⊗V ); x →x ∪ log χ.
8.4. EXPLICIT RECIPROCITY LAWS AND PADIC LFUNCTIONS151
Deﬁnition 8.4.3. The exponential map exp
∗
is deﬁned through the com
mutative diagram:
exp
∗
: H
1
(G
K
, V )
R
R
R
R
R
R
R
R
R
R
R
R
R
D
dR
(V )
∼
m
m
m
m
m
m
m
m
m
m
m
m
H
1
(G
K
, B
dR
⊗V )
m
m
m
m
m
m
m
m
m
m
m
m
Proposition 8.4.4. (i) The image of exp
∗
lies in Fil
0
D
dR
(V ).
(ii) For c ∈ H
1
(G
K
, V ), exp
∗
(c) = 0 if and only if the extension E
c
0 −→V −→E
c
−→Q
p
−→0,
is de Rham as a representation of G
K
.
Proof. (ii) is just by the deﬁnition of de Rham. For (i), c ∈ H
1
(G
K
, V )
implies c = 0 ∈ H
1
(G
K
, (B
dR
/B
+
dR
) ⊗ V ). But x → x ∪ log χ gives an
isomorphism
D
dR
(V )/ Fil
0
(D
dR
(V ))(= H
0
(G
K
, (B
dR
/B
+
dR
)⊗V )) −→H
1
(G
K
, (B
dR
/B
+
dR
)⊗V )).
So exp
∗
(c) = 0 (mod Fil
0
)
Remark. exp
∗
is a very useful tool to prove the nontriviality of cohomology
classes.
Now suppose k ∈ Z, L is a ﬁnite extension of K. Then V (k) is still de
Rham as a representation of G
L
. Deﬁne
D
dR, L
(V (k)) := H
0
(G
L
, B
dR
⊗V (k)) = t
−k
L ⊗
K
D
dR
(V )
by an easy computation. Thus,
Fil
0
(D
dR, L
(V (k))) = t
−k
⊗
K
Fil
k
D
dR
(V )
and this is 0 if k 0. So for every k ∈ Z, for L/K ﬁnite,
exp
∗
: H
1
(G
L
, V (k)) −→t
−k
L ⊗
K
D
dR
(V )
is identically 0 for k 0.
152 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
8.4.3 The explicit reciprocity law
Recall that
H
1
Iw
(K, V )
∼
−→H
1
(G
K
, Z
p
[[Γ
K
]] ⊗V ) = H
1
(G
K
, T
0
(Γ
K
, V )).
If η : Γ
K
→Q
∗
p
is a continuous character, for n ∈ N,
µ ∈ H
1
(G
K
, T
0
(Γ
K
, V )) −→
Γ
Kn
ηµ ∈ H
1
(G
Kn
, V ⊗η).
where we write V ⊗η, not as V (η) to distinguish from V (k) = V ⊗χ
k
. Then
exp
∗
(
Γ
Kn
χ
k
µ) ∈ t
−k
K
n
⊗
K
D
dR
(V )
and is 0 if k 0.
Recall also that we have the isomorphism Exp
∗
: H
1
(K, V )
∼
−→D(V )
ψ=1
,
that D(V )
ψ=1
⊂ D
(0, r
V
]
(V ) and that there exists n(V ) such that
ϕ
−n
(D
(0, r
V
]
(V )) ⊂ K
n
((t)) ⊗
K
D
dR
(V ), for all n ≥ n(V ).
Now denote by
Tr
K
n+k
/Kn
= Tr
K
n+k
((t))/Kn((t))
⊗Id : K
n+k
((t))⊗D
dR
(V ) →K
n
((t))⊗D
dR
(V ).
Theorem 8.4.5 (Explicit Reciprocity Law). Let V be a de Rham repre
sentation of G
K
and µ ∈ H
1
Iw
(K, V ).
(i) If n ≥ n(V ), then
p
−n
ϕ
−n
(Exp
∗
(µ)) =
¸
k∈Z
exp
∗
(
Γ
Kn
χ
k
µ).
(ii) For n ∈ N, n + i ≥ n(V ), then
Exp
∗
Kn
(µ) := Tr
K
n+i
/Kn
p
−(n+i)
ϕ
−(n+i)
(Exp
∗
(µ))
does not depend on i, and Exp
∗
Kn
(µ) =
¸
k∈Z
exp
∗
(
Γ
Kn
χ
k
µ).
8.4. EXPLICIT RECIPROCITY LAWS AND PADIC LFUNCTIONS153
Proof. (ii) follows from (i) and from the commutative diagram:
H
1
(G
L
2
, V
)
exp
∗
−−−→ L
2
⊗
K
D
dR
(V )
cor
Tr
L
2
/L
1
⊗
K
Id
H
1
(G
L
1
, V
)
exp
∗
−−−→ L
1
⊗
K
D
dR
(V )
where L
1
⊂ L
2
are two ﬁnite extensions of K.
For (i), suppose y = Exp
∗
(µ), x ∈ D(V ), and x(k) is the image of x in
D(V (k)) = D(V )(k) (Thus, ϕ(x(k)) = ϕ(x)(k) and γ(x(k)) = χ(γ)
k
γ(x)(k)).
The integral
Γ
Kn
χ
k
µ is represented by the cocycle:
g →c
g
=
log χ(γ
n
)
p
n
g −1
γ
n
−1
y(k) −(g −1)b
where b ∈ A ⊗V is the solution of
(ϕ −1)b = (γ
n
−1)
−1
(ϕ −1)(y)(k)
.
From y ∈ D
(0, r
V
]
(V )
ψ=1
one gets
(ϕ −1)y ∈ D
(0, p
−1
r
V
]
(V )
ψ=0
and then
(γ
n
−1)
−1
(ϕ −1)y ∈ D
(0, p
−n
]
(V )
ψ=0
.
Thus b ∈ A
(0, p
1−n
]
⊗V . This implies that ϕ
−n
(b) and ϕ
−n
(y) both converge
in B
+
dR
⊗V . Then c
g
= ϕ
−n
(c
g
) diﬀers from
c
t
g
=
log χ(γ
n
)
p
n
g −1
γ
n
−1
ϕ
−n
(y)(k)
by the coboundary (g −1)(ϕ
−n
(b)). Therefore, they have the same image in
H
1
(G
Kn
B
+
dR
⊗V (k)). Write
p
−n
ϕ
−n
(y) =
¸
i≥i
0
y
i
t
i
, y
i
∈ K
n
⊗
K
D
dR
(V ),
then
c
t
g
= log χ(g)y
−k
t
−k
+
¸
i,=−k
χ(g)
i+k
−1
χ(γ
n
)
i+k
−1
y
i
t
i
= log χ(g)y
−k
t
−k
+ (g −1)
¸
i,=−k
y
i
t
i
(χ(γ
n
)
i+k
−1)
.
So we get exp
∗
Γ
Kn
χ
k
µ
= y
−k
t
−k
.
154 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
8.4.4 Cyclotomic elements and CoatesWiles morphisms.
Let K = Q
p
, V = Q
p
(1), u =
πn
1+πn
n≥1
∈ lim
←−
O
Fn
, κ(u) ∈ H
1
Iw
(Q
p
, Q
p
(1)),
the Coleman power series f
u
=
T
1+T
. Then we have
Exp
∗
(κ(u)) = (1 + T)
df
u
dT
(π) =
1
π
.
Note that
ϕ
−1
(π)
−1
= ([ε
1/p
] −1)
−1
=
1
(1 +π
1
)e
t/p
−1
,
then
Exp
∗
Qp
(κ(u)) =
1
p
Tr
Qp(π
1
)/Qp
ϕ
−1
(π)
−1
=
1
p
¸
z
p
=1, z,=1
1
e
t/p
−1
=
1
e
t
−1
−
1
p
1
e
t/p
−1
=
1
t
t
e
t
−1
−
t/p
e
t/p
−1
=
+∞
¸
n=1
(1 −p
−n
)ζ(1 −n)
(−t)
n−1
(n −1)!
.
So
exp
∗
Γ
Qp
χ
k
κ(µ)
=
0, if k ≥ 0;
(1 −p
k
)ζ(1 +k)
(−t)
−k−1
(−k−1)!
, if k ≤ −1.
Remark. (i) The map
lim
←−
O
Fn
−¦0¦ −→H
1
Iw
(Q
p
, Q
p
(1)) −→Q
p
, u →t
k+1
exp
∗
Γ
Qp
χ
k
κ(u)
is the CoatesWiles homomorphism.
(ii) Since ζ(1 + k) = 0 if k ≤ −1 is even, the above formula implies that
the extensions of Q
p
by Q
p
(k + 1) constructed via cyclotomic elements are
nontrivial and are even not de Rham.
(iii) dim
Qp
H
1
(G
Qp
, Q
p
(k)) = 1 if k = 0, 1.
Corollary 8.4.6. Nontrivial extensions of Q
p
by Q
p
(k) are not de Rham if
k ≤ 0 is odd.
Exercise. (i) Prove that this is also true for k ≤ −1 even by taking a general
element of D(Q
p
(1))
ψ=1
.
(ii) For [K : Q
p
] < ∞, prove the same statement.
8.4. EXPLICIT RECIPROCITY LAWS AND PADIC LFUNCTIONS155
8.4.5 Kato’s elements and padic Lfunctions of mod
ular forms.
Now we come to see the relations with modular forms. Suppose
f =
∞
¸
n=1
a
n
q
n
∈ S
k
(N), k ≥ 2
is primitive. So Q(f) = Q(a
1
, , a
n
, ) is a ﬁnite extension of Q, and
Q
p
(f) = Q
p
(a
1
, . . . , a
n
, . . . ) is a ﬁnite extension of Q
p
.
Theorem 8.4.7 (Deligne). There exists a representation V
f
of G
Q
of di
mension 2 over Q
p
(f), nonramiﬁed outside Np, such that if Np, for ϕ
the arithmetic Frobenius at (ϕ
(e
2πi
p
n
) = e
2πi
p
n
), then
det(1 −Xϕ
−1
) = 1 −a
X +
k−1
X
2
.
Remark. A Q
p
(f)representation V of dimension d is equivalent to a Q
p
representation of dimension d [Q
p
(f) : Q
p
] endowed with a homomorphism
Q
p
(f) →End(V ) commuting with G
Q
. Therefore, D
cris
(V ), D
st
(V ), D
dR
(V )
are all Q
p
(f)vector spaces.
Theorem 8.4.8 (FaltingsTsujiSaito). (i) V
f
is a de Rham representa
tion of G
Qp
with HodgeTate weights 0 and 1 −k, the 2dimensional Q
p
(f)
vector space D
dR
(V
f
) contains naturally f, and
D
0
dR
(V
f
) = D
dR
(V
f
), D
k
dR
(V
f
) = 0, D
i
dR
(V
f
) = Q
p
(f)f if 1 ≤ i ≤ k −1.
(ii) If p N, then V
f
is crystalline and
det(X −ϕ) = X
2
−a
p
X + p
k−1
.
If p[ N but a
p
= 0, then V
f
is semistable but not crystalline and a
p
is the
eigenvalue of ϕ on D
cris
(V ); if a
p
= 0, then V
f
is potentially crystalline.
Remark. If V is a representation of G
K
, µ ∈ H
1
Iw
(K, V ),
Γ
Kn
χ
k
µ ∈ H
1
(G
Kn
, V (k)),
then this is also true for
aΓ
Kn
χ
k
µ for all a ∈ Γ
K
and for
Γ
K
φ(x)χ
k
µ, with
φ : Γ
K
→Z
p
being constant modulo Γ
Kn
.
156 CHAPTER 8. (ϕ, Γ)MODULES AND PADIC LFUNCTIONS
Theorem 8.4.9 (Kato). There exists a unique element z
Kato
∈ H
1
Iw
(Q
p
, V
f
)
(obtained by global methods using Siegel units on modular curves), such that
if 0 ≤ j ≤ k −2, φ is locally constant on Z
∗
p
∼
= Γ
Qp
with values in Q(f), then
exp
∗
Z∗p
φ(x)x
k−1−j
z
Kato
=
1
j!
¯
Λ(f, φ, j + 1)
f
t
k−1−j
where
¯
Λ(f, φ, j + 1) ∈ Q(f, µ
p
n),
f
t
k−1−j
∈ Fil
0
(D
dR
V
f
(k −1 −j))
.
Our goal is to recover L
p, α
(f, s) from z
Kato
(recall L
p, α
is obtained from
µ
f, α
∈ T
vp(α)
(Z
p
) before). We have Exp
∗
(z
Kato
) ∈ D(V
f
)
ψ=1
, but the question
is how to relate this to D
cris
(V
f
), D
st
(V
f
).
If p [ N, let α be a root of X
2
−a
p
X +p
k−1
with v
p
(α) < k −1; if p N,
let α = a
p
= 0 (in this case pα
2
= p
k−1
). In both cases, take β = p
k−1
α
−1
.
Thus, α, β are eigenvalues of ϕ on D
st
(V
f
).
Assume α = β (which should be the case for modular forms by a conjec
ture). Deﬁne Π
β
=
ϕ−α
β−α
to be the projection on the βeigenspace in D
st
(V
f
)
and extend it by B
†
log, K
linearity to
B
†
log, K
1
t
⊗
K
0
D
st
(V
f
) −→B
†
log, K
⊗
B
†
K
D
†
(V
f
).
Theorem 8.4.10. (i) Π
β
(f) = 0;
(ii)
Π
β
Exp
∗
(z
Kato
)
=
Zp
[ε]
x
µ
f, α
Π
β
(f)
t
k−1
.
Remark. µ
f, α
exists up to now only in the semistable case, but z
Kato
exists
all the time. So a big question is:
How to use it for padic Lfunction?
This action might not be possible to undo. Are you sure you want to continue?