On Wed, Dec 14, 2011 at 19:06, Martin Albrecht <martinralbre...@googlemail.com> wrote: > I should mention that we do test random matrices and that some of our > algorithms depend on the input (e.g., they are rank sensitive). So perhaps at > least for some tests we run, it makes a bit of sense? But we should at least > also output the minimum as well I guess.
What comes to my mind here are the quantiles. E.g. sort your timings by value, split them into 4 parts and output the 5 values where the splits occur. The first one is the min, the last one the max, and the one in the middle where most of the timings are. Nice plots of it are called box-plots. This captures everything more accurately than the min/mean/sd scheme. (The reason why quantiles aren't used that much in the past is just that sorting is much harder than sum+division.) H -- 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
