This chapter is about computing period maps associated to newforms. We assume you have read Chapters General Modular Symbols and Computing with Newforms and that you are familiar with abelian varieties at the level of [Ros86].
In Section The Period Map we introduce the period map and give some examples of situations in which computing it is relevant. Section Abelian Varieties Attached to Newforms is about how to use the period mapping to attach an abelian variety to any newform. In Section Extended Modular Symbols, we introduce extended modular symbols, which are the key computational tool for quickly computing periods of modular symbols. We turn to numerical computation of period integrals in Section Approximating Period Integrals, and in Section Speeding Convergence Using Atkin-Lehner we explain how to use Atkin-Lehner operators to speed convergence. In Section Computing the Period Mapping we explain how to compute the full period map with a minimum amount of work.
Section All Elliptic Curves of Given Conductor briefly sketches three approaches to computing all elliptic curves of a given conductor.
This chapter was inspired by [Cre97a], which contains
similar algorithms in the special case of a
newform with
.
See also [Dok04] for algorithmic methods to compute
special values of very general -functions, which can be used for
approximating
for arbitrary
.
Let be a subgroup of
that contains
for some
, and
suppose
is a newform (see Definition Definition 9.9). In this chapter we describe how to approximately compute the complex period mapping
given by
as in Section Pairing Modular Symbols and Modular Forms. As an application, we can
approximate the special values , for
using
(?). We can also
compute the period lattice attached to a modular abelian variety,
which is an important step, e.g., in enumeration of
-curves (see, e.g., [GLQ04]) or computation of
a curve whose Jacobian is a modular abelian variety
(see, e.g., [Wan95]).
Fix a newform , where
for some
. Let
be the
-conjugates of
, where
acts via
its action on the Fourier coefficients, which are algebraic integers
(since they are the eigenvalues of matrices with integer entries).
Let
(1)
be the subspace of cusp forms spanned by the
-conjugates of
.
One can show using the results discussed in Section Atkin-Lehner-Li Theory
that the above sum is direct, i.e., that
has dimension
.
The integration pairing induces a -equivariant homomorphism
from modular symbols to the -linear dual
of
.
Here
acts on
via
,
and this homomorphism is
-stable by Theorem 1.42.
The abelian variety attached to
is the
quotient
Here , and we include
the
in the notation to emphasize that
these are integral modular symbols.
See [Shim59] for a proof that
is an abelian variety (in particular,
is a lattice, and
is equipped with a nondegenerate Riemann form).
When , we can also construct
as a quotient
of the modular Jacobian
, so
is an
abelian variety canonically defined over
.
In general, we have an exact sequence
Remark 10.1
When , the abelian variety
has a canonical structure of
abelian variety over
. Moreover, there is a conjecture of Ribet
and Serre in [Rib92] that describes the simple abelian
varieties
over
that should arise via this construction. In
particular, the conjecture is that
is isogenous to some abelian
variety
if and only if
is a number field
of degree
. The abelian varieties
have this property
since
embeds in
and the endomorphism ring over
has degree at most
(see [Rib92] for details). Ribet proves that his
conjecture is a consequence of Serre’s conjecture
[Ser87] on modularity of mod
odd irreducible
Galois representations (see Section Applications of Modular Forms). Much of Serre’s
conjecture has been proved by Khare and Wintenberger (not
published). In particular, it is a theorem that if
is a simple
abelian variety over
with
a number field
of degree
and if
has good reduction at
, then
is
isogenous to some abelian variety
.
Remark 10.2
When , there is an object called a
Grothendieck motive that is attached to
and has a canonical
“structure over
“. See [Sch90].
In this section, we extend the notion of modular
symbols to allows symbols of the form
where
and
are
arbitrary elements of
.
Definition 10.3
The abelian group sym{esM_k}
of extended modular symbols of weight
is the
-span of symbols
, with
a homogeneous polynomial of degree
with integer
coefficients, modulo the relations
and modulo any torsion.
Fix a finite index subgroup .
Just as for usual modular symbols,
is equipped with
an action of
, and we
define the space of extended modular symbols of weight
for
to be the quotient
The quotient is torsion-free and fixed by
.
The integration pairing extends naturally to a pairing
(2)
where we recall from (?) that denotes
the space of antiholomorphic cusp forms.
Moreover, if
is the natural map, then
respects (2) in the sense that
for all
and
, we have
As we will see soon, it
is often useful to replace
first by
and then
by an equivalent sum
of symbols
such that
is easier to
compute numerically than
.
Let be a Dirichlet character of modulus
.
If
, let
.index{
}
Let sym{esM_k(N,eps)} be the quotient of
by the relations
, for all
,
, and modulo any torsion.
In this section we assume is a congruence subgroup
of
that contains
for some
.
Suppose
, so
and
is
an integer such that
,
and consider the extended modular symbol
.
Let
denote
the integration pairing from Section Pairing Modular Symbols and Modular Forms.
Given an arbitrary cusp form
,
we have
The reversal of summation and integration is justified because
the imaginary part of is positive so that the sum
converges absolutely. The following
lemma is useful for computing the above infinite sum.
Lemma 10.4
(3)
Proof
See Exercise 10.1
In practice we will usually be interested in computing the period map
when
is a newform. Since
is a
newform, there is a Dirichlet character
such that
. The period map
then
factors through the quotient
, so it suffices to
compute the period map on modular symbols in
.
The following proposition is an analogue of [Cre97a, Prop. 2.1.1(5)].
Proposition 10.5
For any ,
and
,
we have the following relation in
:
Proof
By definition, if is a modular symbol
and
, then
.
Thus
, so
The second equality in the statement of the proposition now follows easily.
In the case of weight and trivial character,
the “error term”
(4)
vanishes since is constant and
. In general this term does
not vanish. However, we can suitably modify the formulas
found in [Cre97a, 2.10] and still obtain an algorithm
for computing period integrals.
Algorithm 10.6
Given ,
and
presented as a
-expansion to some precision,
this algorithm outputs an approximation to
the period integral
.
It would be nice if the modular symbols
of the form for
and
were to generate a large subspace
of
.
When
and
,
Manin proved in [Man72]
that the map
sending
to
is a surjective
group homomorphism. When
, the author
does not know a similar
group-theoretic statement. However, we have the following
theorem.
Theorem 10.7
Any element of can be written in the form
for some and
Moreover,
and
can be chosen so that
,
so the error term (4) vanishes.
Figure 10.1
“Transporting” a transportable modular symbol.
\begin{figure}
\begin{center}
\begin{picture}(230,150)(0,0)
\put(-10,10){\line(1,0){200}}
\put(10,10){\circle*{3}}
\put(10,0){`\infty`}
\qbezier(10,10)(50,40)(75,10)
\put(49,25){\vector(1,0){1}}
\put(45,28){`P`}
\put(75,10){\circle*{3}}
\put(75,0){`\gamma \infty`}
\qbezier(75,10)(130,50)(140,10)
\put(120,30){\vector(1,0){1}}
\put(115,34){`Q`}
\put(140,10){\circle*{3}}
\put(140,0){`\beta \infty`}
\put(20,90){\circle*{3}}
\put(17,96){`\alpha`}
\qbezier(20,90)(50,120)(55,60)
\put(40,100){\vector(1,0){1}}
\put(40,104){`P`}
\put(55,60){\circle*{3}}
\put(60,54){`\gamma \alpha`}
\qbezier(55,60)(70,180)(100,140)
\put(87,150){\vector(1,0){1}}
\put(85,154){`Q`}
\put(100,140){\circle*{3}}
\put(100,146){`\beta \alpha`}
\comment{
\qbezier(10,10)(-10,30)(10,50)
\qbezier(10,50)(40,70)(20,90)
\put(5,45){\vector(2,3){1}}
\qbezier(75,10)(90,30)(75,40)
\qbezier(75,40)(60,50)(55,60)
\put(69,45){\vector(1,-1){1}}
\put(80,35){\vector(-1,2){1}}
\qbezier(140,10)(160,30)(140,50)
\qbezier(140,50)(90,80)(110,90)
\qbezier(110,90)(130,100)(100,120)
\qbezier(100,120)(90,130)(100,140)
\put(110,90){\vector(-1,-1){1}}
}
\put(100,50){\vector(-1,4){10}}
\put(50,40){\vector(-1,4){10}}
\put(140,155){\parbox[t][4in][l]{2in}
{
The modular symbol
can be “transported” to
provided that
The author and Helena Verrill prove this theorem
in [SV01].
The condition
that the error term vanishes means that one can replace
by any
in the expression for the modular symbol
and obtain an equivalent modular symbol. For this reason,
we call such modular symbols transportable, as illustrated
in Figure 10.1.
Note that in general not every element of the form
must lie in
. However, if
, then
does lie in
. It would be interesting to know under what
circumstances
is generated by symbols of the form
with
. This sometimes fails for
odd; for example, when
, the condition
implies
that
has an eigenvector with eigenvalue
, and hence
is of finite order. When
is even, the author can see no
obstruction to generating
using such symbols.
Let .
Consider the Atkin-Lehner involution
on
,
which is defined by
Here we take the positive square root if is odd.
Then
is an involution when
is even.
There is an operator on modular symbols, which we also denote
, which is given by
and one has that if and
, then
If is a Dirichlet character of modulus
, then
the operator
sends
to
. Thus if
,
then
preserves
. In particular,
acts on
.
The next proposition shows how to compute the pairing
under
certain restrictive assumptions.
It generalizes a result of [Cre97b] to
higher weight.
Proposition 10.8
Let be a cusp form which is
an eigenform for the Atkin-Lehner operator
having eigenvalue
(thus
and
is even).
Then for any
and any
, with the property that
, we have
the following formula, valid for any
:
Here .
Proof
By Proposition Proposition 10.5 our condition on
implies that
.
We describe the steps of the following computation below.
For the first equality, we break the path into three paths,
and in the second, we apply the -involution to the first
term and use that the action of
is compatible with
the pairing
and that
is an eigenvector
with eigenvalue
.
In the following sequence of equalities we combine the first two terms
and break up the third; then we replace
by
and regroup:
A good choice for is
,
so that
.
This maximizes the minimum of the imaginary parts
of
and
, which results in series that converge
more quickly.
Let .
The polynomial
satisfies .
We obtained this formula by viewing
as
the
symmetric product of the
-dimensional
space on which
acts naturally. For example,
observe that since
,
the symmetric product of two eigenvectors for
is an eigenvector
in
having eigenvalue
.
For the same reason, if
, there need
not be a polynomial
such that
.
One remedy is to choose another
so that
.
Since the imaginary parts of the terms ,
and
in the proposition are all relatively large, the sums
appearing at the beginning of Section Approximating Period Integrals converge
quickly if
is small. It is important to
choose
in Proposition Proposition 10.8 with
small; otherwise
the series will converge very slowly.
Remark 10.9
Is there a generalization of Proposition Proposition 10.8
without the restrictions that and
is even?
Suppose is an elliptic curve and let
be the corresponding
-function. Let
be the root number of
, i.e.,
the sign of the functional equation for
, so
, where
.
Let
be the modular form associated to
(which exists by [Wil95, BCDT01]).
If
, then
(see Exercise 10.2). We have
If , then
. If
, then
(5)
For more about computing with -functions of elliptic curves,
including a trick for computing
quickly without directly
computing
, see [Coh93, Section 7.5] and
[Cre97a, Section 2.11]. One can also find higher derivatives
by a formula similar to (5) (see
[Cre97a, Section 2.13]). The methods in this chapter for
obtaining rapidly converging series are not just of computational
interest; see, e.g., [Gre83] for a nontrivial
theoretical application to the Birch and Swinnerton-Dyer conjecture.
Fix a newform ,
where
for some
.
Let
be as in (1).
Let be any
-linear map with the same kernel
as
; we call any such map a rational period mapping
associated to
. Let
be the period mapping associated
to the
-conjugates of
. We
have a commutative diagram
Recall from Section Abelian Varieties Attached to Newforms that the cokernel of
is the abelian variety
.
The Hecke algebra acts on the linear dual
by .
Let
be the kernel of the
ring homomorphism
that sends
to
.
Let
Since is a newform, one can show that
has dimension
.
Let
be a basis
for
, so
We can thus compute , hence a choice of
.
To compute
, it remains to compute
.
Let denote the space of cusp forms with
-expansion in
.
By Exercise 10.3
is a -vector space of dimension
.
Let
be a basis for this
-vector space.
We will compute
with respect to the basis of
dual to this basis.
Choose elements
with the following properties:
Given this data, we can compute
and
We break the integrals into real and imaginary parts because this
increases the precision of our answers.
Since the vectors and
,
, span
, we have computed
.
Remark 10.10
We want to find symbols satisfying the
conditions of Proposition Proposition 10.8. This is usually possible
when
is very small, but in practice it is difficult when
is large.
Remark 10.11
The above strategy was motivated by [Cre97a, Section 2.10].
Using modular symbols and the period map, we can compute all elliptic
curves over of conductor
, up to isogeny. The algorithm in
this section gives all modular elliptic curves (up to isogeny),
i.e., elliptic curves attached to modular forms, of conductor
.
Fortunately, it is now known by
[Wil95, BCDT01, TW95]
that every
elliptic curve over
is modular, so the procedure of this section
gives all elliptic curves (up to isogeny) of given conductor. See
[Cre06] for a nice historical discussion of this
problem.
Algorithm 10.12
Given , this
algorithm outputs equations for all
elliptic curves of conductor
, up to isogeny.
[Modular Symbols]
Compute using Section Explicitly Computing .
[Find Rational Eigenspaces]
Find the -dimensional eigenspaces
in
that correspond to elliptic curves.
Do not use the algorithm for decomposition from
Section Decomposing Spaces under the Action of Matrix, which is too complicated
and gives more information than we need. Instead, for the first
few primes
, compute all eigenspaces
, where
runs through integers with
.
Intersect these eigenspaces to find the eigenspaces that
correspond to elliptic curves. To find just the new ones,
either compute the degeneracy maps to lower level or find
all the rational eigenspaces of all levels that strictly
divide
and exclude them.
[Find Newforms]
Use Algorithm 9.14
to compute to some precision each newform associated
to each eigenspace
found in step (2).
[Find Each Curve]
For each newform
found in step (3), do the following:
[Period Lattice] Compute the corresponding
period lattice
by computing the image of
,
as described in Section Computing the Period Mapping.
[Compute ]
Let
. If
, swap
and
, so
.
By successively applying generators of
,
we find an
equivalent element
in
, i.e.,
and
.
[-invariants] Compute the invariants
and
of the lattice
using the following
rapidly convergent series:
where , where
is as in step
(b).
A theorem of Edixhoven
(that the Manin constant is an integer) implies that
the invariants
and
of
are integers,
so it is only necessary to compute
to large
precision to completely determine them.
[Elliptic Curve] An elliptic curve with
invariants and
is
[Prove Correctness]
Using Tate’s algorithm, find
the conductor of . If the conductor is not
, then recompute
and
using more terms of
and real numbers
to larger precision, etc. If the conductor is
,
compute the coefficients
of the modular form
attached to the elliptic curve
, for
.
Verify that
, where
are the coefficients
of
. If this equality holds, then
must be isogenous
to the elliptic curve attached to
, by the Sturm
bound (Theorem 9.18) and Faltings’s isogeny theorem.
If the equality fails for some
, recompute
and
to larger precision.
There are numerous tricks to optimize the above algorithm.
For example, often one can work separately with
and
and
get enough information to find
, up to isogeny
(see [Cre97b]).
Once we have one curve from each isogeny class of curves of
conductor , we find each curve in each isogeny class (which is
another interesting problem discussed in [Cre97a]), hence
all curves of conductor
. If
is an elliptic curve, then any
curve isogenous to
is isogenous via a chain of isogenies of prime
degree. There is an a priori bound on the degrees of these
isogenies due to Mazur. Also, there are various methods for finding
all isogenies of a given degree with domain
. See
[Cre97a, Section 3.8] for more details.
In this section we briefly survey an alternative approach to finding curves of a given conductor by finding integral points on other elliptic curves.
Cremona and others have developed a complementary
approach to the problem of computing all elliptic curves of given
conductor (see [CL04]). Instead of computing all
curves of given conductor, we instead consider the seemingly more
difficult problem of finding all curves with good reduction outside a
finite set of primes. Since one can compute the conductor of a
curve using Tate’s algorithm
[Tat75, Cre97a, Section 3.2],
if we know all curves with good reduction
outside
, we can find all curves of conductor
by letting
be
the set of prime divisors of
.
There is a strategy for finding all curves with good reduction
outside . It is not an algorithm, in the sense
that it is always guaranteed to terminate (the modular symbols method
above is an algorithm), but in practice it often works. Also,
this strategy makes sense over any number field, whereas the modular
symbols method does not (there are generalizations of modular
symbols to other number fields).
Fix a finite set of primes of a number field
. It is a theorem
of Shafarevich that there are only finitely many elliptic curves with
good reduction outside
(see [Sil82, Section IX.6]). His
proof uses that the group of
-units in
is finite and Siegel’s
theorem that there are only finitely many
-integral points on an
elliptic curve. One can make all this explicit, and sometimes in
practice one can compute all these
-integral points.
The problem of finding all elliptic curves with good reduction outside
of can be broken into several subproblems, the main ones being
determine the following finite subgroup of :
find all -integral points on certain elliptic curves
.
In [CL04], there is one example, where they find all
curves of conductor by finding all curves
with good reduction outside
. They finds
curves of
conductor
that divide into
isogeny classes.
(Note that
.)
One can also find curves by simply enumerating Weierstrass equations.
For example, the paper [SW02] discusses a database
that the author and Watkins created that contains hundreds of millions
of elliptic curves. It was constructed by enumerating Weierstrass
equations of a certain form. This database does not contain
every curve of each conductor included in the database. It is,
however, fairly complete in some cases. For example, using the Mestre
method of graphs [Mes86], we verified in [JBS03] that
the database contains all elliptic curve of prime conductor , which implies that the smallest conductor
rank
curve is composite.
Exercise 10.1
Prove Lemma 10.4.
Exercise 10.2
Suppose is a newform
and that
. Let
.
Prove that
[Hint: Show that .
Then substitute
for
.]
Exercise 10.3
Let be a power
series whose coefficients
together generate a number field
of
degree
over
. Let
be the complex vector space spanned
by the
-conjugates of
.
Exercise 10.4
Find an elliptic curve of conductor using Section
All Elliptic Curves of Given Conductor.