• No results found

A Chabauty computation on a curve of genus 2

N/A
N/A
Protected

Academic year: 2021

Share "A Chabauty computation on a curve of genus 2"

Copied!
4
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

A Chabauty computation on a curve of genus 2 Nils Bruin

In these notes, we redo part of a Chabauty argument that explicitly determines all rational points on a curve of genus 2. The curve in question comes from Flynn, Poonen, Schaefer, Cycles of Quadratic Polynomials, Duke 90, 1997.

The argument we give here is slightly different from the computation done in the original article. The advantage of the present approach is that it more easily gener- alises to non-hyperelliptic curves.

1. Picard group of a curve

For a projective curve C, the group variety we can use is quite a bit more compli- cated than for an affine curve. First we describe the group. Here, we’ll concentrate on one particular curve of genus 2. Much of the methods we describe apply in general.

C : y2 = x6+ 8x5+ 22x4+ 22x3+ 5x2+ 6x + 1

Again, this is just an affine model. What we really mean, is the complete curve in weighted projective space:

C : Y2 = X6+ 8X5Z + 22X4Z2+ 22X3Z3+ 5X2Z4+ 6XZ5+ Z6

and well, write ∞+ = (1 : 1 : 0) and ∞ = (1 : −1 : 0). There are 6 rational points on this curve:

±, (0 : ±1 : 1), (−3 : ±1 : 1)

Definition: A divisor on a curve is a formal linear combination of points in C(Q) X

P

nPP.

A divisor is defined over Q if the sum is galois-stable, i.e., nPσ = nP. The degree of a divisor isP nP.

Definition: A divisor is called principal if the multiplicities describe the pole and zero orders of a function on C:

(x) = (0, 1) + (0, −1) − ∞+− ∞

The principal divisors form a subgroup of the divisor group. Note that on a projective curve, functions have the same number of poles as they have zeros, so principal divisors are of degree 0.

Definition: The Picard group of a curve is the group of divisors modded out by principal divisors. An element of the picard group is called a divisor class.

Theorem: If a curve has points everywhere locally then every galois-stable divisor class can be represented by a galois-stable divisor.

Definition/Theorem: The Jacobian of a curve is a g-dimensional projective group variety of which the rational points represent the galois-stable divisor classes of degree 0. More specific for genus 2 curves, every (rational) divisor class can be represented in the form

[P + Q − 2∞]

where P, Q are either rational points or quadratic conjugate points.

1

(2)

2

Abel-Jacobi map:

C → J

P 7→ [P − ∞] Complicated computation:

Pic0(C/Q) = hG = [∞+− ∞]i ' Z Some elements of Pic0(C/Q):

−6G [2(0, 1) − 2∞]

−5G [(−3, −1) − ∞]

−4G [(−3, −1) + ∞+− 2∞]

−3G [(0, 1) − ∞]

−2G [(0, 1) + ∞+− 2∞]

−G [(0, −1) + (−3, −1) − 2∞]

0 [∞− ∞]

G [∞+− ∞]

2G [2∞+− 2∞] 3G [(0, 1) + (−3, 1) − 2∞] 4G [(0, −1) − ∞] 5G [(0, −1) + ∞+− 2∞] 6G [(−3, 1) − ∞] 7G [(−3, 1) + ∞+− 2∞] 8G [2(0, −1) − 2∞] 9G [P + P − 2∞] where α =√

33 and

P = (1 : 101α − 580 : α − 6) and P = (1 : −101α − 580 : −α − 6)

Note that if we reduce 9G modulo 3 then we get 9G ≡ 0 mod 3. Furthermore, we have C(F3) = {(1 : ±1 : 0), (0 : ±1 : 1)}, so we see that under the Abel-Jacobi-map, C(Q) maps into

{0, G, 4G, 6G} + h9Gi 2. Logarithms for Gm

Recall from last time that we used

Log(1 + z) = z − z2/2 + z3/3 − z4/4 + · · ·

to convert an equation in Gm around 1 into an equation around 0 in A1. The obser- vation to be made here is that this log has a better interpretation as

Log(1 + z) = Z z

0

1 1 + tdt.

The important bit here is that the differential ω = 1zdz is regular on Gm (it has a pole in z = 0, ∞, but those are not on Gm). The nice thing is that (formally),

Z z1

0

1 1 + tdt +

Z z2

0

1 1 + t =

Z (1+z1)(1+z2)−1 0

1 1 + tdt

(3)

3

which gives the fundamental property of Log. If the corresponding power series converge (say, zi ∈ pZp and p > 2) then these identities really hold.

Coleman integration: These integrals can be made to work along larger paths as well, but this requires deeper results. We will make sure we only need to integrate along paths that fit inside small p-adic neighbourhoods, so that we stay inside the radius of convergence. Then integration really just corresponds to the formal power series operation.

3. Regular differentials on C

On hyperbolic curves, one can find multiple regular differentials. For instance, on C, we have

ω1 = dx

y , ω2 = xdx y

We use integration to define functions on Pic0(C) by insisting that forP[Pi− Qi] we haveP RPi

Qi ωj. This yields a group homomorphism:

J → A2

P[Pi− Qi] 7→ (P RPi

Qi ω1,P RPi

Qi ω2) In a way, this is actually how J is really defined.

4. Computing differential that vanishes along the Mordell-Weil group

Note that the points P and P are 3-adically close to ∞. To be precise, (1 : −1 + y1 : t)

is equal to P for (t, y1) = 141α − 810, 101α − 580. We know that (−1 + y1)2 = 1 + 8t + 22t2 + 22t3+ 5t4+ 6t5+ t6 so

y1 = −4t − 3t2+ t3− 2t4+ 2t5− 2t6+ O(t7) Then

ω1 = t3

−1 + y1d(1/t) = t − 4t2+ 13t3− 39t4+ 111t5+ O(t6) ω2 = t2

−1 + y1d(1/t) = 1 − 4t + 13t2− 39t3+ 111t4− 304 ∗ t5+ O(t6) Z [P +P ]

0

ω1 ≡ 12 mod 34 Z [P +P ]

0

ω2 ≡ 63 mod 34

so ω = (21 + O(33))ω1− (4 + O(332) has the property that RD

0 ω = 0 if D in the Mordell-Weil group.

(4)

4

5. Determining the number of points in a fiber of reduction Let’s see how many points in C(Q) reduce to (0 : 1 : 1) modulo 3. For this, we make once again a power series expansion:

Pt= (3t : y2 : 1) where

y2 = 1 + 9t − 18t2+ 459t3− 3402t4 + O(35, t5) If Pt∈ C(Q) for some t ∈ Z3 then [Pt− P0] ∈ J (Q), so

Z t 0

ω = 0 In fact,

Z t 0

ω ≡ 9t + 9t2 mod (33)

So indeed, by Strassman’s Theorem we see that this power series should have at most 2 zeros for t ∈ Z3. In fact, we know that t = 0, −1 are such zeroes, so we see that the points (0 : 1 : 1) and (−3 : 1 : 1) are the only two rational points on C(Q) that reduce to that point mod 3.

6. Exercise

Do the Chabauty argument for the fiber around ∞+.

Referenties

GERELATEERDE DOCUMENTEN

The slope before buckling is much larger than after buckling, so the force for a buckled beam with engineering strain zero is a good approximation for the buckling load:.. P c ≈

Since the categories of quasi- coherent nite presentation sheaves at over the base scheme and of locally free rank n sheaves are algebraic stacks [9, Theorem 4.6.2.1], we deduce

A reason why elliptic curves are import is that we can put a group struc- ture on it. Now we will assume that the base field k is algebraically closed to construct the group

Our main tools are the quantitative version of the Absolute Parametric Subspace Theorem by Evertse and Schlickewei [5, Theorem 1.2], as well as a lower bound by Evertse and Ferretti

The proof for this needs some insight in projective geometry and it’s rather involved, so I only give here the general rule for drawing complete rational Bezier curves: Make every

tact. Daardoor komt er ook weinig terecht van het zich verplaatsen in de positie van de ander en van inzicht in objectieve problemen. Hiermee kan het gedrag

By so doing, the theory helps to define interrelationships amongst concepts in kinematics addressing the principal objective for this study, “to identify mathematical

Onder het colluvium bevindt zich de C horizont, deze horizont wordt gekenmerkt door bruingeel zandige leem (3).. De dikte van het colluvium is