Professional Documents
Culture Documents
Textbook Algebraic Geometry and Number Theory Summer School Galatasaray University Istanbul 2014 1St Edition Hussein Mourtada Ebook All Chapter PDF
Textbook Algebraic Geometry and Number Theory Summer School Galatasaray University Istanbul 2014 1St Edition Hussein Mourtada Ebook All Chapter PDF
https://textbookfull.com/product/dynamics-and-analytic-number-
theory-proceedings-of-the-durham-easter-school-2014-1st-edition-
badziahin/
https://textbookfull.com/product/math-223a-algebraic-number-
theory-notes-alison-miller/
https://textbookfull.com/product/algebraic-number-theory-a-brief-
introduction-1st-edition-j-s-chahal/
https://textbookfull.com/product/algorithmic-and-experimental-
methods-in-algebra-geometry-and-number-theory-gebhard-bockle/
Algebraic and Differential Methods for Nonlinear
Control Theory Elements of Commutative Algebra and
Algebraic Geometry Rafael Martínez-Guerra
https://textbookfull.com/product/algebraic-and-differential-
methods-for-nonlinear-control-theory-elements-of-commutative-
algebra-and-algebraic-geometry-rafael-martinez-guerra/
https://textbookfull.com/product/a-gateway-to-number-theory-
applying-the-power-of-algebraic-curves-1st-edition-keith-kendig/
https://textbookfull.com/product/3264-and-all-that-a-second-
course-in-algebraic-geometry-intersection-theory-combinatorial-
enumerative-geometry-final-draft-1st-edition-david-eisenbud/
https://textbookfull.com/product/abstract-algebra-applications-
to-galois-theory-algebraic-geometry-representation-theory-and-
cryptography-second-edition-celine-carstensen-opitz/
https://textbookfull.com/product/competitive-math-for-middle-
school-algebra-probability-and-number-theory-1st-edition-vinod-
krishnamoorthy/
Progress in Mathematics
321
Hussein Mourtada
Celal Cem Sarıoğlu
Christophe Soulé
Ayberk Zeytin
Editors
Algebraic Geometry
and Number Theory
Summer School, Galatasaray University,
Istanbul, 2014
Progress in Mathematics
Volume 321
Series Editors
Antoine Chambert-Loir, Université Paris-Diderot, Paris, France
Jiang-Hua Lu, The University of Hong Kong, Hong Kong SAR, China
Yuri Tschinkel, Courant Institute of Mathematical Sciences, New York, USA
Algebraic Geometry
and Number Theory
Summer School, Galatasaray University,
Istanbul, 2014
Editors
Hussein Mourtada Celal Cem Sarıoğlu
Institut de Mathématiques de Department of Mathematics
Jussieu - Paris Rive Gauche Dokuz Eylül University
Paris, France İzmir, Turkey
Christophe Soulé Ayberk Zeytin
Institut des Hautes Études Scientifiques Department of Mathematics
Bures-sur-Yvette, France Galatasaray University
Istanbul, Turkey
Mathematics Subject Classification (2010): 11E16, 11E41, 11M41, 11R23, 11S40, 13D02, 14C17, 14E30,
14G40, 14J60, 14J70, 55N25, 55R10, 11R29, 14M17
List of Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
K. Büyükboduk
p-adic Variation in Arithmetic Geometry: A Survey . . . . . . . . . . . . . . . . . 1
I.˙ Coskun
¸
The Birational Geometry of Moduli Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . 29
O. Debarre
On the Geometry of Hypersurfaces of Low Degrees
in the Projective Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
G. Freixas i Montplet
The Riemann–Roch Theorem in Arakelov Geometry . . . . . . . . . . . . . . . . 91
L.W. Tu
Computing the Gysin Map Using Fixed Points . . . . . . . . . . . . . . . . . . . . . . 135
Z. Wojtkowiak
On -adic Galois L-functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
A. Zeytin
Class Number Problems and Lang Conjectures . . . . . . . . . . . . . . . . . . . . . . 211
Progress in Mathematics, Vol. 321, vii–x
c 2017 Springer International Publishing
Preface
The CIMPA summer school “Algebraic Geometry and Number Theory” (AGNT)
was held at Galatasaray University, Istanbul, between 2 and 13 June 2014. Sub-
jects ranging from Arakelov geometry and Iwasawa theory to classical projective
geometry and birational geometry were covered. All these active research domains
were presented by leading experts in their respective fields. The main aim of the
lectures was to introduce these contemporary research topics to graduate students.
Accordingly the lectures have incorporated the main ideas and techniques together
with motivating examples and guiding problems in this variety of subjects.
Each individual lecturer provided a detailed outline of the lecture together
with a list of prerequisites. In addition, lecturers have pre-organized the lectures
so as to use the two-week time in most efficient way.
The young participants have shown a considerable interest to the summer
school. We had to regretfully decline more than half of the applications to our
school. Those who participated have been actively involved in the lectures and
helped providing a lively atmosphere throughout the school. The final three days
of the summer school afternoon sessions were reserved for research talks. The talks
were mostly delivered by the young participants.
We are thankful to Galatasaray University not only for the financial, but also
for the logistic support. We would also like to thank CIMPA and TÜBİTAK for
their financial support.
Our gratitude goes also to the members of the scientific committee (Olivier
Debarre (École Normale Supérieure, Paris, France), Fouad El Zein (IMJ, Paris,
France), Monique Lejeune (Université de Versailles, France), Kamal Khuri-Makdisi
(LSMS, Beirut, Lebanon), Rahim Zaare-Nahandi (University of Tehran, Iran),
Chris Peters (Université de Grenoble, Institut Fourier, France), Christophe Soulé
(IHÉS, France), Loring W. Tu (Tufts University, Medford, USA), A. Muhammed
Uludağ (Galatasaray University, Turkey)) and those of the organizing committee
(Hakan Ayral (Galatasaray University, Turkey), Merve Durmuş (Yeditepe Univer-
sity, Turkey), Hussein Mourtada (Université de Paris VII, France), İrem Portakal
(Galatasaray University, Turkey), İsmail Sağlam (Koç University, Turkey), Celal
Cem Sarıoğlu (Dokuz Eylül University, Turkey), Ayberk Zeytin (Galatasaray Uni-
versity, Turkey)) for their efforts.
The last named editor was supported by TÜBİTAK grants 113R017 and
114R073 during the summer school and during the preparation of this volume.
viii Preface
other moduli spaces and gives a guide to the literature in this subject. The text
contains many exercises.
In his chapter, Zeytin studies a very classical topic: integral binary quadratic
forms. A new approach to this subject using certain bipartite ribbon graphs called
çarks is given. The central result of the paper is the one to one correspondence
between integral points of an affine hypersurface admitting a certain group action
and indefinite binary quadratic forms of fixed discriminant. The author also studies
these hypersurfaces along the lines of Kobayashi hyperbolicity. More precisely, by a
x Preface
result of Demailly these hypersurfaces are Kobayashi hyperbolic and have finitely
many rational points as the class group is finite. These results are in accordance
with a conjecture of Lang relating arithmetic to hyperbolicity.
Abstract. The main goal of this survey is to provide a general overview of the
theme of p-adic variation, both from a historical and technical view point. We
start off with Kummer’s work and Iwasawa’s treatment of cyclotomic fields,
which eventually paved the way to the modern p-adic variational techniques.
These methods have proved extremely powerful and enabled us to gain access
to some of the most important problems in mathematics, such as the Bloch-
Kato conjectures and Langlands’ Programme. We will point at a variety of
concrete applications in this vein.
Mathematics Subject Classification (2010). 11G05; 11G07; 11G40; 11R23;
14G10.
Keywords. Iwasawa theory, Modular motives.
1. Introduction
These notes grew out of a mini lecture series I gave in the CIMPA Summer School
“Algebraic Geometry and Number Theory” that took place in June 2014 in Istan-
bul. Evident from the scope of these notes, I was certainly over-ambitious for the
material I would like to introduce and I was not actually able to touch upon any
of the matters I hope to discuss in these notes. In this regard, this survey is rather
a continuation of what I have covered in these lectures.
The main goal of this survey is to provide a general overview of the theme
of p-adic variation, both from a historical and technical view point. Needless to
say, this survey is bound to be humble, due to the very large landscape that
the topic governs (from Bloch–Kato conjectures to the Langlands program) and
the ignorance of the author. However, we hope that it will still be useful as an
introductory guide to the literature for beginners.
Due to time and space restrictions, we shall not provide almost any proofs,
but rather refer the reader to the appropriate piece of the literature. We will still
discuss the idea behind various constructions whenever we are able to.
2 K. Büyükboduk
2. Kummer congruences
(For the basic notions we refer to in this section that the reader may be unfamiliar
with, as well as basic results we shall be quoting, we advice the reader to have a
look at [Jan96, Lan94, Was97].)
Kummer’s attempt on Fermat’s Last Theorem did not result its complete1
resolution, its impact in arithmetic geometry has been immense. His starting point
was to consider the Fermat equation
Xp + Y p = Zp
(where XY Z = 0 and X, Y, Z are integers) as an equality that takes place in the
cyclotomic field Q(e2πi/p ), by rewriting the same equation in the form
X p = (Z − Y ) · (Z − e2πi/p Y ) · · · (Z − e2(p−1)πi/p Y ) . (2.1)
Note that each factor Z − e Y (k = 0, 1, . . . , p − 1) is an element of the ring2
2kπi/p
Z[e 2πi/p
]. One than tries to utilise the following rather elementary idea: If X, Y, Z
are all sufficiently reduced (and massaged, in a sense which we will not make
precise), then all the factors above will be pairwise coprime. Their product being
a pth power will force (by unique factorisation into primes in the ring Z[e2πi/p ]?)
1 In a certain sense of the word it did eventually succeed, as Kummer’s work paved the way for
the idea of p-adic variation, by which Andrew Wiles’ [Wil95, TW95] work was certainly heavily
influenced.
2 This ring happens to be ring of integers of the cyclotomic field Q(e2πi/p ), see [Lan94, §IV.1,
Theorem 3]. Exactly for this reason it is a Dedekind ring [Lan94, ] and possesses all the nice
properties we will be using in this section.
p-adic Variation in Arithmetic Geometry: A Survey 3
each of these factors to be pth powers. It is then not so hard to reach a contradiction
at this point, and deducing Fermat’s Last Theorem.
We all very well know that something has to go wrong in the discussion
above. This is also secretly indicated there: For p ≥ 23, the rings Z[e2πi/p ] will
never be Unique Factorization Domains [Uch71], i.e., there never will be unique
factorization into primes in these rings. This is the flaw in the argument above
and Kummer’s contributions start with his attempt to mend this problem.
Kummer first defined what he called ideal numbers (called simply ideals of a
ring nowadays, after Dedekind) and proved the following theorem.
Theorem 2.1 (Kummer). Suppose K is a number field and OK is its ring of inte-
gers. The ring OK is a Dedekind ring. In particular, every ideal I ≤ OK may be
written uniquely (up to ordering) as a product of (finite number of) prime ideals.
Definition 2.2. Given a number field L, we let JK denote the group of fractional
ideals of K and PK ≤ JK the subgroup of principal fractional ideals. The quotient
ClK := JK /PK is called the ideal class group of K.
Remark 2.3. We will not go as far as providing the definitions of the concepts that
appear in Definition 2.2, but will rather refer the reader to the basic references we
have given at the start of this section. However, let us observe that the collection
IK of integral ideals of OK form a semigroup. It turns out that the collection of
fractional ideals turns out to be the group of fractions of this semigroup, and as a
matter of fact, this fact is equivalent to Kummer’s theorem above.
We may again attempt to run the argument above after writing the equation
(2.1) as an equality taking place in JK :
Xp = Z − Y · Z − e2πi/p Y · · · Z − e2(p−1)πi/p Y
where A stands for the ideal generated by A. Theorem 2.1 now shows that each
principal ideal Z − e2kπi/p Y = Apk is a pth power of an ideal Ak ≤ Z[e2πi/p ]. In
order to be able to go back to the realm of elementary number theory, one would
hope in the first place that each Ak = αk is principal. In fact if this was really
so, we would have had
Z − e2kπi/p Y = αpk · uk , k = 0, 1, . . . , p − 1 (2.2)
2πi/p ×
for some uk ∈ Z[e ] , i.e., each factor Z − e 2kπi/p
Y would be a pth power
up to a unit in the ring Z[e2πi/p ]. This is a statement close enough to what we
erroneously attempted to work out above and it turns out that (2.2) does still
yield a contradiction which proves FLT.
The question that remains is therefore the following: Is there a way to decide
whether or not the ideals Ak are in fact principal, given the fact that their pth
powers are principal? Note that the first obstruction for an affirmative answer to
this question is already identified as part of Definition 2.2 above: The ideal class
group ClZ[e2πi/p ] , whose non-triviality means that the ring Z[e2πi/p ] possesses non-
principal ideals. The following is one of the main theorems of classical number
theory.
4 K. Büyükboduk
Theorem 2.4. For a number field K, the ideal class group ClK is always finite.
The size of this group is traditionally denoted by hK and called the class
number of K. We now have an element [Ak ] ∈ ClZ[e2πi/p ] (the class of ideals that
contains Ak ) and we are hoping that for some reason this class should better be the
trivial class (i.e., the class that contains only principal ideals). We are given that
[Apk ] = [Ak ]p is the trivial class in the finite group ClZ[e2πi/p ] . This immediately
yields the following:
Proposition 2.5. If p does not divide the order hZ[e2πi/p ] of the ideal class group of
the pth cyclotomic field, then each of the ideal classes [Ak ] are trivial. In turn, all
ideals Ak are principal and Fermat’s Last Theorem for exponent p holds true.
This motivates the following definition:
Definition 2.6. A prime number p is called regular if p does not divide hZ[e2πi/p ] ,
and called irregular otherwise.
If we wish to draw a concrete conclusion towards FLT from Proposition 2.5,
we would need a procedure to decide when a prime p is regular. That is exactly
where magic happens and Kummer proves the following magnificent result (Theo-
rem 2.8 below). Before stating this theorem, we first need to introduce completely
new object, from a far apart domain (at least at first sight) of Mathematics, which
surprisingly will
be relate to the discussion above.
∞
Let ζ(s) = n=1 1/ns denote the Riemann zeta function. This is a complex-
valued function analytic (and non-vanishing) on the half-plane Re(s) > 1. Riemann
proved that it admits an analytic continuation to the whole complex plane (except
for a simple pole at s = 1) and verifies a functional equation3 relating the values
ζ(s) and ζ(1 − s). Since the days of Riemann, ζ(s) (and its analogues in other
contexts, which we shall also briefly talk about later on in Section 4) has attracted
a lot of attention and analysing ζ (or its ‘cousins’) to extract various arithmetic
information has become one of the central themes in number theory. Kummer’s
theorem we are about to present is a result also along these lines. Before we do
so, the following proposition (due to Euler, but formally proved by Riemann) is
a strong indication that the Riemann ζ-function, albeit being a purely analytic
object, might have reflections on the algebraic world:
Proposition 2.7. The values of ζ(s) at negative odd integers are all (non-zero)
rational numbers; whereas its values at a negative even integer are always zero.
It is not hard to compute these values explicitly, either: ζ(−1) = −1/12,
ζ(−3) = 1/120, ζ(−5) = −1/252, ζ(−7) = 1/240, ζ(−9) = −1/132, ζ(−11) =
691/32760, . . .
We are now ready to state Kummer’s results:
3 This functional equation had already been observed by Euler, before Riemann had formally
function is known as the Hasse–Weil zeta function of an elliptic curve. The celebrated theorem
of Gross–Zagier computes the first derivative of this zeta function at the value s = 1 in terms of
the height of a Heegner point.
6 K. Büyükboduk
• The second conjecture is known as Gauss’ class number one problem and it
was solved by Heegner, Baker and Stark [Hee52, Bak67, Sta67].
• The conjecture G3 is wide open till this day.
If we were to sloganize Gauss’ conjectures in one sentence, it would be the state-
ment that the class numbers of quadratic fields behave rather erratically in families
of extensions of fixed degree5 . Iwasawa on the other hand observed that situation
changes dramatically if one considers a certain tower extensions, the behavior of
the (p-parts of) class numbers along such families may be characterized only in
terms of the tower one considers. The main point we would like to emphasize at
this point is that Iwasawa’s conclusion (Theorem 3.2 below) is due to the fact that
the (pro-p parts of) ideal class groups along this tower in fact interpolate p-adic
analytically and Iwasawa theory aims to study this sort of p-adic analytic variation
in arithmetic families.
To state Iwasawa’s theorem let us introduce the tower we shall be interested
in. The main reference for this portion of our survey is [Was97, Chapters 7 and
13]. Let K be any number field and let K∞ be any Zp extension of K. That
is to say, the Galois group of K∞ /K is an infinite Galois extension with group
isomorphic to Zp . Every number field has at least one Zp extension. This we
can see using Class Field Theory6 but also more directly: Consider the extension
Q(μp∞ ) := ∪∞ n=1 Q(μpn ) obtained by adjoining the rational field all p-power roots
of unity. Observe that
Gal(Q(μp∞ )/Q) = lim Gal(Q(μpn /Q)) ∼
= lim(Z/pn Z)× ∼
= Z×
p.
←− ←−
n n
We have a decomposition
Z× ∼ ×
p = (Z/pZ) × (1 + pZp )
∞
(via the map ai pi → a0 and the Teichmüller character ω to split this surjection)
i=1
which yields the decomposition
Gal(Q(μp∞ )/Q) ∼
= Δ × Γ.
Here Δ = Gal(Q(μp )/Q) ∼ = (Z/pZ)× and Γ = Gal(Q(μp∞ )Δ /Q) ∼ = 1 + pZp ∼
= Zp ,
where Q(μp∞ ) is the fixed field of Δ (viewed as a subgroup of Gal(Q(μp∞ )/Q))
Δ
and the isomorphism 1+pZp ∼ = Zp is obtained via the p-adic logarithm. The exten-
sion Qcyc := Q(μp∞ )Δ is called the cyclotomic Zp -extension7 of Q. The following
Q(μp∞ )
z HH
zz HH Δ
z HH
z HH
zzz H
zz
z Q∞
zzz
z
zzz
zz
Q(μpn ) Z× =Δ×Γ
p
n−1
Z/p Z Γ∼ = Zp
Q(μp ) Q
QQQ
QQQ
QQQ
Δ∼
Q
=(Z/pZ)× QQQQ
Q
Q
Now given a number field K, we set Kcyc := KQcyc , the cyclotomic Zp -extension
of K. This construction validates our claim that every number field possesses at
least one Zp -extension.
Given an arbitrary Zp -extension K∞ /K as above, let Kn /K denote the
unique subfield with Gal(Kn /K) ∼ = Z/pn Z. Existence and uniqueness of Kn fol-
low from the fact that the topologically cyclic group Zp has a unique (closed)
subgroup of index pn . We call Kn the nth layer in this extension and we set8
Γ := Gal(K∞ /K) and Γn := Γ/Γp ∼
n
= Gal(Kn /K). Here is our diagram that
summarizes this picture.
K∞ P
PPP pn+1 n+1
PΓPP ∼ =p Zp
PPP
PP
Kn+1
n n+1
Γ Γp /Γp ∼
=Z/pZ
Kn
nnnnn
nn
nnnnΓnn =Γ/Γpn ∼
=Z/pn Z
nn
K
We are now ready to state Iwasawa’s theorem. Let pen denote the exact power
of p dividing the class number hKn of the nth layer of the tower K∞ /K.
8 Iwasawa [Iwa69, Iwa71, Iwa73],when he initiated the arithmetic study of towers of extensions of
the sort K∞ , called these Γ-extensions. This is the reason for our tendency to name the relevant
Galois groups Γ.
8 K. Büyükboduk
Theorem 3.2 (Iwasawa). There exists integers λ = λ(K∞ /K), μ = μ(K∞ /K) ≥ 0
and ν = ν(K∞ /K), all independent of n and an integer n0 such that
en = λn + μpn + ν
for every n ≥ n0 .
The proof of this theorem relies on the following observation. Let An denote
the pro-p part of the ideal class group of Kn . Then An is naturally a Zp [Γn ]-
module. Furthermore, the Zp [Γn ]-action on An is compatible in that, it induces a
continuous action of the ring Λ := lim Zp [Γn ] on A∞ := limn An . Serre observed
←− ←−
that the ring Λ is (non-canonically) isomorphic to Zp [[X]], formal power series ring
in one variable with coefficients in Zp . This ring has many fantastic properties: It
is a complete local Noetherian ring, and it is regular therefore a UFD. Moreover,
the module theory over Λ is relatively easy and there is a structure theorem for
finitely generated Λ-modules, resembling the structure theorem for modules over
PIDs. All these facts assembled together shows that A∞ is a finitely generated
torsion Λ-module and there is a Λ-module map
A∞ −→ Λ/f Λ × Λ/pμ Λ (3.1)
(where the product is over a finite number of polynomials f ∈ Λ of positive degree,
such that the quotients Λ/f Λ are free Zp -modules) with finite kernel and cokernel.
This fact together with Iwasawa’s control theorem that the maps (A∞ )Γpn → An
have finite cokernels proves the theorem (with μ as in (3.1) and λ the sum of the
degrees of the polynomials f that appear in (3.1)). Here (A∞ )Γpn stands for the
n
largest quotient9 of A∞ on the which Γp acts trivially.
Remark 3.3. It is a folklore conjecture that the μ-invariant μ(Kcyc /K) vanishes for
any number field K. Note that this conjecture implies via (3.1) that the interpolated
class group A∞ is a finitely generated Zp -module. Washington and Ferrero proved
in [FW79] that this is the case for abelian extension of Q. The general case remains
widely open.
3.1. p-adic zeta functions and Iwasawa’s main conjectures
We recall the significant role that the special values of the zeta function played
in Kummer’s results1 Theorem 2.8. It turns out that Iwasawa’s theory has also
a counterpart in the zeta-world. The first hint towards the fact that the zeta-
values interpolate into a p-adic analytic function was secretly spelled out as part
9 It n
is helpful to notice that this quotient is given by A∞ /(γ p −1)A∞ , where γ is any topological
generator of Γ.
1 Equally importantly, one should also take a note of Dirichlet’s analytic class number formula,
cf., [Was97, Chapter 4]. These express the value of the Dedekind zeta function ζK (s) at s = 0 (or
equivalently, the residue at s = 1 by its functional equation) in terms of the class number and the
unit regulator of the field K. Due the space restrictions, we do not dwell further on this topic.
Note however that the Birch and Swinnerton-Dyer conjecture (as well as its p-adic variant) that
we will discuss in Section 4 are direct analogues of this class number formula in view of the very
general conjectures of Bloch and Kato.
p-adic Variation in Arithmetic Geometry: A Survey 9
of the torsion subgroup E(K)tors (in fact, for example in the case when K = Q,
we understand what all the possibilities this group could possibly be, thanks to
Mazur’s phenomenal theorem [Maz77]). On the other hand, the Mordell–Weil rank
rlag is still mysterious and understanding this invariant is the subject of many
major open problems in number theory. One of these is the celebrated conjecture
of Birch and Swinnerton-Dyer, which we are to introduce next.
Just as we may attach a Dirichlet character an L-series, we may also3 asso-
ciate E an L-function, called the Hasse–Weil L-function of the elliptic curve E
and denoted by L(E/K, s). We refer the reader to [Sil09, Section C.16] for a defini-
tion of this object. And just the same way these Dirichlet L-series control various
arithmetic data associated to the Dirichlet character in question, we may inquire
whether or not the special values of L(E/K, s) have an arithmetic significance.
The Birch and Swinnerton-Dyer conjecture (BSD for short) claims that they do
and in fact furnishes us with a efficient algorithm to at least determine whether a
given elliptic curve carries an infinite number of K-rational points or not.
Conjecture 4.1 (Birch and Swinnerton-Dyer). The order of vanishing of L(E/K, s)
at s = 1 agrees with the Mordell–Weil rank ralg (K) of E/K. In particular, the
Mordell–Weil group E(K) is finite if and only if L(E/K, s) = 0.
Although we shall not record it here, we remark that the full BSD makes also
a strikingly precise prediction for the ralg (K)th derivative of L(E/K, s) in terms
of the arithmetic invariants of the elliptic curve E.
Remark 4.2. For a general number field K, we do not know whether L(E/K, s) is
defined at s = 1 or not: The defining (infinite) Euler products for L(E, s) converges
only on the right half-plane Re(s) > 2. When K = Q, it follows from Wiles’ proof
of the Taniyama–Shimura conjecture in [Wil95, TW95] that L(E/Q, s) admits an
analytic continuation to the full complex plane. For a general totally real number
field K, Taylor [Tay06] has proved that L(E/K, s) has a meromorphic continuation
to C and this is pretty much all we know for a generic (i.e., non-CM) elliptic
curve. Both of these spectacular results are corollaries of various modularity lifting
theorems, which dwell heavily on p-adic variation. We shall take the a few baby
steps towards these results in the next section.
We now discuss what extend of our knowledge towards BSD. We first make
a very honest remark: Almost nothing is known over non-abelian number fields
K and even over the field K = Q, nothing at all is known when the order of
vanishing ran (K) := ords=1 L(E/K, s) is greater than 1. This discussion should
make the limits of our exposition explicit.
4.1. The case K = Q and ran = 1
When the analytic rank is at least one, the Birch and Swinnerton-Dyer conjecture
predicts that there should be points of infinite order in the Mordell-Weil group of
3 The underlying theme is the notion of a motive: Associated to each of these objects, one may
define a motive, to which one attaches an L-function in a natural way. See [Del79, Ser86].
12 K. Büyükboduk
E. In the particular case when the analytic rank is exactly one, Heegner points
supply us systematically with these points of infinite order: This is the celebrated
theorem of Gross and Zagier.
Let us briefly recall the definition of Heegner points on the elliptic curve E.
To start, the work of Breuil, Conrad, Diamond and Taylor [BCDT01] (generalizing
[Wil95, TW95] to give the full statement of the Taniyama–Shimura conjecture)
shows that every elliptic curve over Q admits a modular parametrization
φ : X0 (N ) −→ E,
namely a non-zero morphism between the modular curve X0 (N ) of level N and the
elliptic curve E which is defined over Q. (As a side note, we should perhaps record
that the proof of this spectacular fact is based on the theme
√ of p-adic variation
as well.) Consider a quadratic imaginary field F = Q( −D) with discriminant
−D. For the sake of simplicity we shall assume that D = 3, 4 (so as to ensure
that OF× = {±1}). Fix an integer N which verifies the Heegner hypothesis, namely
that every prime dividing N splits in K/Q. This shows that there exists and
ideal N ≤ OF for which OF /N ∼ = Z/N Z. For a square-free positive integer n
that is prime to N , consider an order O of F of conductor n. Define the ideal
Nn := N ∩ O of O. This is an invertible ideal of O and it verifies N−1 ∼
n /O = Z/N Z.
In other words, N−1n /O ⊂ N −1
O/O is a cyclic subgroup of order N . As a matter
of fact, complex analytic description of elliptic curves shows us that A := C/O is
an elliptic curve (a priori defined over C) and N −1 O/O is simply the N -torsion
submodule A[N ] of the elliptic curve E. The moduli interpretation of the curve
X0 (N ) then shows that we may view the pair xn := (C/O, N−1 n /O) (which we
call an enhanced elliptic curve with Γ0 (N )-level structure) as a point on the curve
X0 (N )(C). Theory of Complex Multiplication shows that the points xn are in fact
algebraic: They are defined over the ring class field F (n) of conductor n of the
imaginary quadratic field F , i.e.,
xn = (C/O, N−1
n /O) ∈ X0 (N )(F (n)) .
compatible with the leading term formula predicted as part of the full BSD which
we have not made explicit here).
Theorem 4.3 (Gross–Zagier). Under the Heegner hypothesis, the point yF has in-
finite order iff L (E/F, 1) is non-zero, i.e., iff the analytic rank of E/F is exactly
one. In particular,
ran (E/F ) = 1 =⇒ ralg (E/F ) ≥ 1 .
Not far later than this work, Kolyvagin assembled a powerful machine (called
the method of Euler systems) which takes as an input the full collection of Heegner
points yn and gives out the following more precise version of Theorem 4.3.
Theorem 4.4 (Kolyvagin, see [Kol90, Kol91a, Kol91b]). Suppose that the field
F verifies the Heegner hypothesis and ran (E/F ) = 1. Then the algebraic rank
ralg (E/F ) also equals one, as predicted by the BSD conjecture. Moreover, the Tate–
Shafarevich group is finite as well.
Remark 4.5. Although these statements are given over an imaginary quadratic
field F , one may choose this auxiliary field in a suitable way (and use along the
way the results of [BFH90, MM91]) to formulate consequences towards the BSD
conjecture for E/Q: For r = 0 or 1,
ran (E/Q) = r =⇒ ralg (E/Q) = r
and the Tate–Shafarevich group (over Q) is finite.
Remark 4.6. The reader may rightfully worry that until this point in this section,
we have not made a single reference to the Iwasawa theory of elliptic curves nor
any sort of p-adic variational technique. Although Kolyvagin’s methods rely on p-
descend techniques, they do not have p-adic analytic flavour. In contrast, though,
the converse theorems to Kolyvagin’s theorems (due to Rubin [Rub92] in the CM
case, and for a non-CM elliptic curve established in [SU14, Ski14a, Ski14b, Wan12,
Ven16b] under varying set of assumptions ([SU14, Wan12] treats the case r = 0))
very strongly appeal to techniques from Iwasawa theory.
One should also note that in the (non-generic) case when E has CM, there
was much earlier progress on the BSD conjecture, due to [CW77, RW82] (and this
approach was ultimately perfected in [Rub91]). Let us very briefly overview the
main idea (that very much dwells on Iwasawa theoretic techniques). Let E/Q be an
elliptic curve that has CM by a quadratic imaginary field F . A fundamental object
in the study of elliptic curves is the p-adic Tate module Tp (E) := limn E(Q)[pn ],
←−
the projective limit of the p-power torsion on E. This is a free Zp -module equipped
with a continuous action of the absolute Galois group GQ of Q. By a theorem of
Faltings [Fal83], the p-adic Tate module (for a single prime p) determines E up
to isogeny (even when E has no CM), so it should not come as a surprise that
we allow ourselves to forget pretty much everything about the elliptic curve E
besides this simple linear-algebra object! The fact that E has CM translates into
the fact that Gal(F (E[p∞ ])/F ) (where F (E[p∞ ]) is the smallest field through
14 K. Büyükboduk
6 In their proof, Greenberg and Stevens make fabulous use of p-adic variation: What they first
Theorem 4.11 (Kato, [Kat04] and Rubin, [Rub91]). In the situation of Conjec-
ture 4.10,
⎧
⎨ ralg , if E has good ordinary or non-split
rp ≥ multiplicative reduction at p,
⎩
1 + ralg , if E has split multiplicative reduction at p.
In fact, combined with the work of Bertrand [Ber82] on the non-vanishing of
p-adic height pairings, one may deduce that
ran = 1 ⇐⇒ rp = 1 ⇐⇒ ralg = 1
in case the elliptic curve E has CM. (We remark that a CM elliptic curve may
only have additive reduction at primes of bad reduction.)
Theorem 4.12 (Skinner–Urban, [SU14]). Suppose E does not have CM, has good
ordinary reduction at p and is semistable. Suppose further that certain p-adic height
pairing (see [Sch82, Sch85, Sch83]) is non-degenerate and the Tate–Shafarevich
group is finite. Then Conjecture 4.10 holds true for p ≥ 11.
The main idea behind the proofs of these two theorems is to extract infor-
mation regarding the big Mordell–Weil group E(Qcyc ) = ∪E(Qn ) from LE , then
specialize to n = 1 (which corresponds to the Mordell–Weil group E(Q) at the
base level). This is achieved through Iwasawa–Mazur main conjectures for the el-
liptic curve E over the cyclotomic Zp -tower. Before we give the formulation of
this conjecture, we first need to introduce Selmer groups, which are fundamental
objects of interest in arithmetic geometry.
Definition 4.13. For an arbitrary number field L and positive integer m, let
Selpm (E/L)
H 1 (Lv , E[pm ])
:= ker H (L, E[p ]) −→
1 m
Kum
v im(E(Lv )/pm E(Lv ) → H 1 (Lv , E[pm ])
be the pm -Selmer group. Here the direct product is over all primes of L and
Kum : E(Lv )/pm E(Lv )→H 1 (Lv , E[pm ])
is the Kummer map. We define the pro-p Selmer group Selp (E/L) by setting
Selpm (E/L) := lim Selpm (E/L) .
−m
→
We finally define the Iwasawa theoretic Selmer group Selp (E/Qcyc ) by
Selp (E/Qcyc ) = lim Selp (E/Qn ) .
−→n
Remark 4.14. The main interest in the Selmer group lies behind the following
classical exact sequence:
0 −→ E(L) ⊗ Qp /Zp −→ Selp (E/L) −→ IIIE/L [p∞ ] −→ 0 .
p-adic Variation in Arithmetic Geometry: A Survey 17
Remark 4.19. Although one does not deduce the following converse of Theorem 4.4
(due to Skinner [Ski14a]) directly from Theorem 4.17, its proof still is purely p-adic:
In the situation of Theorem 4.17, if ralg = 1 and the p-part of the Tate–Shafarevich
group is finite, then ran = 1. (See also [Ven16b] for a similar result in the case of
split-multiplicative reduction.)
1.F.4. Except for the limited right of replacement or refund set forth in
paragraph 1.F.3, this work is provided to you ‘AS-IS’, WITH NO
OTHER WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR ANY PURPOSE.
Please check the Project Gutenberg web pages for current donation
methods and addresses. Donations are accepted in a number of
other ways including checks, online payments and credit card
donations. To donate, please visit: www.gutenberg.org/donate.
Most people start at our website which has the main PG search
facility: www.gutenberg.org.