Re: Re: [sage-devel] timeit for mathematicians

Tom Boothby's did something much like this once... On Dec 20, 2011 9:25 AM, "David Roe" wrote: -- 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 ht

Re: Re: [sage-devel] timeit for mathematicians

That sounds awesome, and not too hard to do. I'll add it to my wishlist as well. David On Tue, Dec 20, 2011 at 02:25, Nicolas M. Thiery wrote: > On Fri, Dec 16, 2011 at 10:12:35AM -0800, Harald Schilly wrote: >>    done, it's here: http://trac.sagemath.org/sage_trac/ticket/12168 > > Speaking of

Re: Re: [sage-devel] timeit for mathematicians

On Fri, Dec 16, 2011 at 10:12:35AM -0800, Harald Schilly wrote: >done, it's here: http://trac.sagemath.org/sage_trac/ticket/12168 Speaking of timing code, one tool I have been dreaming about for a while is: sage: def random_data(n): ... sage: def my_algorithm(data):

Re: Re: [sage-devel] timeit for mathematicians

On Friday, December 16, 2011 6:42:09 PM UTC+1, William wrote: > > Do it! > done, it's here: http://trac.sagemath.org/sage_trac/ticket/12168 H -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-

Re: Re: [sage-devel] timeit for mathematicians

On Dec 16, 2011 9:25 AM, "Harald Schilly" wrote: > > > > On Thursday, December 15, 2011 1:06:57 AM UTC+1, William wrote: >> >> One heads up here -- many factoring algorithms are nondeterministic. > > > Hi, thank's for pointing this out! But I think at that point one has to ask what one wants to me

Re: Re: [sage-devel] timeit for mathematicians

On Wed, Dec 14, 2011 at 3:39 PM, Harald Schilly wrote: > On Wed, Dec 14, 2011 at 23:50, Harald Schilly > wrote: >> I hacked something together that plots a histogram... > > I made a second version, that uses the timeit command directly and > scales the x-axis manually (i.e. including the 0) > >

Re: Re: [sage-devel] timeit for mathematicians

On Wed, Dec 14, 2011 at 23:50, Harald Schilly wrote: > I hacked something together that plots a histogram... I made a second version, that uses the timeit command directly and scales the x-axis manually (i.e. including the 0) It just times the factor function, but feel free to edit the .sage fil

Re: Re: [sage-devel] timeit for mathematicians

On Thu, Dec 15, 2011 at 00:10, William Stein wrote: > > What were you timing? > well, it's in the attached sage file, nothing serious: factor(2**111-1), 10 inner loops and 200 repetitions. if you have a better example to test, go ahead or tell me :) H -- To post to this group, send an email t

Re: Re: [sage-devel] timeit for mathematicians

On Wed, Dec 14, 2011 at 2:50 PM, Harald Schilly wrote: > I hacked something together that plots a histogram. Well, it looks odd, > maybe there is a big obvious bug -- but nevertheless, it's not symmetric and > in no way a normal distribution! What were you timing? > H > > > -- > To post to this

Re: Re: [sage-devel] timeit for mathematicians

I hacked something together that plots a histogram. Well, it looks odd, maybe there is a big obvious bug -- but nevertheless, it's not symmetric and in no way a normal distribution! H -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send

Re: Re: [sage-devel] timeit for mathematicians

On Wed, Dec 14, 2011 at 10:11 AM, Harald Schilly wrote: > 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.) A nice s

Re: Re: [sage-devel] timeit for mathematicians

On Wed, Dec 14, 2011 at 19:06, Martin Albrecht 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 outpu

Re: Re: [sage-devel] timeit for mathematicians

On Wednesday 14 December 2011, Volker Braun wrote: > On Wednesday, December 14, 2011 12:18:58 AM UTC, Martin Albrecht wrote: > > It uses the "Student-t" method to estimate the certainty > > This is precisely what the python documentation warns about (see Nils' > post). The timings are not normal d

Re: [sage-devel] timeit for mathematicians

On Wednesday, December 14, 2011 2:36:24 AM UTC+1, Volker Braun wrote: > The timings are not normal distributed. Yes, I just wanted to add the same. mean and standard deviation only made sense if you have a normal distribution. This t-test martin mentioned only checks if the assumption about t

Re: [sage-devel] timeit for mathematicians

On Tue, Dec 13, 2011 at 5:36 PM, Volker Braun wrote: > On Wednesday, December 14, 2011 12:18:58 AM UTC, Martin Albrecht wrote: >> >> It uses the "Student-t" method to estimate the certainty > > This is precisely what the python documentation warns about (see Nils' > post). The timings are not norm

Re: [sage-devel] timeit for mathematicians

On Wednesday, December 14, 2011 12:18:58 AM UTC, Martin Albrecht wrote: > > It uses the "Student-t" method to estimate the certainty > This is precisely what the python documentation warns about (see Nils' post). The timings are not normal distributed. Especially if your code is deterministic it

Re: [sage-devel] timeit for mathematicians

On Tuesday 13 December 2011, William Stein wrote: > Thoughts?Are there any experts out there in code benchmarking done > from a more mathematically sophisticated perspective than just one > number? It's entirely possible I'm making some stupid mistake in > suggesting the above. Carlo Wood