On 17 April 2012 11:28, Jeroen Demeyer <jdeme...@cage.ugent.be> wrote: > Consider the file > http://boxen.math.washington.edu/home/jdemeyer/integral_points.sage > (based on a doctest in ell_rational_field.pyx): > > E = EllipticCurve([-879984,319138704]) > P1 = E.point((540,1188)) > P2 = E.point((576,1836)) > P3 = E.point((468,3132)) > P4 = E.point((612,3132)) > P5 = E.point((432,4428)) > t = cputime() > pts = E.integral_points([P1,P2,P3,P4,P5]) > print "Time:", cputime(t) > > With sage-4.8, this takes about 35 seconds. With sage-5.0.beta10, the > same test takes about 50 seconds. I'm not aware of any serious changes > to the elliptic curve code (apart from the Cremona Database upgrade). > Any clues? >
There is a major change in the code which computes integral points, from faster code which gave wrong answers to slower (or at least longer-running) code which is better. BUT that is quietly rotting at #10973 so certainly is not the cause of this. I am not aware of any changes to the code in the integral_points function itself but will take a look. Using verbose=True one sees that all but a negligible part of the time is taken in finding Z-linear combinations of those points which are integral with coefficients bounded in absolute value by 6. This is somthing which (a long time ago) I spent a lot of effort making fast.... John > > Jeroen. > > -- > To post to this group, send an email to sage-devel@googlegroups.com > To unsubscribe from this group, send an email to > sage-devel+unsubscr...@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/sage-devel > URL: http://www.sagemath.org -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
