Processing Math: Done
No jsMath TeX fonts found -- using unicode fonts instead.
This may be slow and might not print well.
Use the jsMath control panel to get additional information.
jsMath Control PanelHide this Message


jsMath

Discussion of verifying kolyvagin for rank 3 curve 5077a

We have
\begin{verbatim}
sage: E = EllipticCurve('5077a')
sage: for p in prime_range(30):
...       print p, gcd(E.ap(p),p+1)
2 1
3 1
5 2
7 4
11 6
13 2
17 2
19 1
23 6
29 6
sage: E.heegner_discriminants_list(10)
[-7, -19, -40, -47, -55, -59, -71, -79, -84, -88]
sage: for D in E.heegner_discriminants_list(20):
...       K.<a> = QuadraticField(D)
...       if D%11 and len(K.factor(11)) == 1 and len(K.factor(23)) == 1:
...           print D, K.class_number()
-47 5
-59 3
-71 7
-104 6
-115 2
sage: K.<a> = QuadraticField(-115)
sage: p = 3; c = 11*23
sage: c
253
sage: 12*24*3
864
sage: for p in prime_range(50):
...       if len(K.factor(p)) == 1:
...          print p, gcd(E.ap(p),p+1)
2 1
3 1
5 2
11 6
13 2
19 1
23 6
47 3
\end{verbatim}

Thus if we take $D=-115$, $c=23 \cdot 47$, and $\ell=11$.  Then my
purported maybe algorithm would compute:

$$P_c\pmod{\ell} \in E(\mathbf{F}_{\ell^2})[3].$$

With luck, this will just happen to turn out to be nonzero hence
verify Kolyvagin's conjecture provably.  If it isn't we can try
$\ell=23$ or $\ell=47$ with $c'=11\cdot 23\cdot 47/ \ell$.

We would have to do computations in 
$$
  \Div(X_0(5077)_{\F_{11}}^{\ss})
$$
We have
\begin{verbatim}
sage: dimension_new_cusp_forms(5077*11,2,11)
4233
\end{verbatim}
Doing linear algebra on that space over $\F_3$ is completely
reasonable.  But computing the Kolyvagin point directly over a number
field of degree 2162 would be impossibly hard, since the height would
be crazy ginormous.    Maybe one could get the field down to
degree $2\cdot 11\cdot 23 = 506$, but even then the height is likely
to make this just impossible. 

With my new algorithm idea, though, I think it this computation is
possible.  In fact, maybe it will just be as hard as computing
a 1-dimensional eigenspace in characteristic $3$ of a sparse
matrix of size $4233$, which is computationally... possible.

This seems very feasible, but requires working with class number > 1, which in theory should be fine.

Here is some more than Jen and I did tonight:

sage: K.<a> = QuadraticField(-7)
sage: E = EllipticCurve('5077a')
sage: for p in prime_range(400):
...       if len(K.factor(p))==1:
...          print p, factor(gcd(E.ap(p),p+1))
3 1
5 2
7 2^2
13 2
17 2
19 1
31 2
41 2 * 3 * 7
47 3
59 1
61 2
73 2
83 2
89 1
97 2
101 2
103 2^3 * 13
131 1
139 2^2
157 2
167 2^3
173 3
181 1
199 1
223 7
227 2
229 2 * 5
241 1
251 2^2
257 2 * 3
269 2
271 2
283 2^2
293 3
307 1
311 2^3
313 2
349 5 * 7
353 2
367 2^4
383 3
397 2
sage: dimension_new_cusp_forms(5077*41,2,41)
16927
sage: dimension_new_cusp_forms(5077*229,2,229)
96481
sage: for D in E.heegner_discriminants_list(20):
...         K.<a> = QuadraticField(D)
...         if D%11 and len(K.factor(23)) == 1:
...              print D, K.class_number()
-47 5
-59 3
-71 7
-104 6
-115 2
-127 5
-131 5
-151 7
sage: for D in E.heegner_discriminants_list(20):
...         K.<a> = QuadraticField(D)
...         if D%11 and len(K.factor(29)) == 1:
...              print D, K.class_number()
-19 1
-40 2
-47 5
-79 5
-84 4
-104 6
-127 5
-131 5
sage: dimension_new_cusp_forms(5077*29,2,29)
11849
sage: E = EllipticCurve('5077a')
sage: for p in prime_range(50):
...       print p, gcd(E.ap(p),p+1)
2 1
3 1
5 2
7 4
11 6
13 2
17 2
19 1
23 6
29 6
31 2
37 38
41 42
43 4
47 3

dayscambridge2/sprints/5077aestimates (last edited 2009-12-02 07:08:57 by was)