Re: Ellsea

[Kevin Acres <research@research-systems.com>]

> On 15 Sep 2019, at 15:21, Bill Allombert <Bill.Allombert@math.u-bordeaux.fr> wrote:
> 
>> On Sun, Sep 15, 2019 at 03:00:57PM +0400, Kevin Acres wrote:
>> Sorry, I don’t have an example to easily send. I’d been using ellap
>> but wanted the early abort feature of ellsea.
>> 
>> Using a curve ellinit([0,0,0,-3,b], 2^285-9) with b in the range from
>> 10^7 upwards. It probably crashed before 90 iterations of b. I
>> couldn’t establish the failure point though. The parvector had a size
>> of 256 with 10 threads assigned. 
> 
> So something like
> parvector(256,b,ellsea(ellinit([0,0,0,-3,10^7+b], 2^285-9),1))
> 
> What version of PARI/GP are you using ?
> Does it happen often ?
> 
> Cheers,
> Bill

Effectively that summarizes the problem that I’m seeing. Different curves fail at different points. However it seems deterministic. I’ll get more time later in the day to get an example put together. 

I’m unsure as to whether ellsea is crashing or the return value causes problems later in the code. It’s worth noting that ellap worked flawlessly in the same piece of code. Basically I swapped out ellap for ellsea (with minor changes of course) and started to see this problem.