Re: precision of result when adding t_REAL with lots of cancellation

[John Cremona <john.cremona@gmail.com>]

On 13 February 2014 22:15, Peter Bruin <P.Bruin@warwick.ac.uk> wrote:
> Bonjour Karim et Denis,
>
>> * Peter Bruin [2014-02-12 13:12]:
>>> Bonjour Denis et pari-dev,
>>>

>
> I am not familiar enough with descent on elliptic curves to say if there
> is any way to avoid such large number field elements (compared to the
> apperently modest input size), or if one can somehow keep track of these
> elements in a more compact factored form.  Maybe Denis can enlighten us?
>

Bonjour.   The number field elements in question are representatives
of a class in K^* / (K^*)^2 for some number field K.  I know that
there are some very good tricks to obtain good representatives, asn
have been assuming that Denis's code uses these, but it would be good
to have that confirmed.  Otherwise, a reminder of the paper in which
these tricks are described?

John

> Best wishes,
>
> Peter
>