Hi Nick,
You don't have to guess which one seems to be quicker, you can Benchmark this
directly on whatever machine you'll be doing the computations. On my machine,
GSL is by far the fastest (3.5x faster than the next), followed by Marek's
solution, then David's. Of course, if you're only doing a few thousand, then
downloading and installing GSL if it isn't already present (or even just
running the benchmark in the first place!) will take much more than the ~10
seconds you'll save. But if you're going to be doing millions or billions of
iterations, or need access to different RNGs, GSL might be worth the extra
overhead.
best,
Derek
CODE:
use Benchmark;
use PDL::GSL::RNG;
$data = sequence(1500);
$r = PDL::GSL::RNG->new('taus');
timethese(5e4, {
David=>sub{cat(random($data),$data)->transpose->qsortvec->(1)->flat;},
Marek=>sub{$data(qsorti(random($data)));},
GSL=>sub{$r->ran_shuffle($data);}
});
OUTPUT:
Benchmark: timing 50000 iterations of David, GSL, Marek...
David: 17 wallclock secs (16.39 usr + 0.06 sys = 16.45 CPU) @ 3039.51/s
(n=50000)
GSL: 3 wallclock secs ( 2.66 usr + 0.01 sys = 2.67 CPU) @ 18726.59/s
(n=50000)
Marek: 9 wallclock secs ( 9.58 usr + 0.03 sys = 9.61 CPU) @ 5202.91/s
(n=50000)
> On Aug 6, 2015, at 11:23 AM, Nick Wright <[email protected]> wrote:
>
> Hi everyone
>
> Thank you all for your quick replies!
>
> I've given Marek and David's suggestions a try and both seem to work (David I
> had to edit yours to be $shuffled = $data -> index(qsorti(random($data))),
> was that what you meant? It gave a compilation error otherwise). Both are
> very quick and do exactly what I wanted, thank you both - David's appears to
> be a bit quicker so I expect I'll use that.
>
> Thanks for all the other suggestions as well, there's some things there I'll
> investigate once I set this running.
>
> Cheers,
>
> Nick
>
> On 6 August 2015 at 18:14, vividsnow <[email protected]
> <mailto:[email protected]>> wrote:
> hi
>
> use PDL::GSL::RNG;
> $p = sequence(100);
> $r = PDL::GSL::RNG->new('taus');
> $r->ran_shuffle($p);
> p $p;
>
> On 08/06/2015 07:59 PM, Nick Wright wrote:
> > Hi everyone
> >
> > I need help with something that I think *should* be simple with Perl/PDL
> > but I can't find the command for it.
> >
> > I basically just want to randomly re-order a 1-dimensional, moderately long
> > (1500 entries) piddle and I can't find a simple way. I need to do this for
> > the purposes of a Monte Carlo simulation, so I was hoping PDL might have an
> > efficient way to do it so I can repeat is 1000s of times quickly!
> >
> > Thanks in advance for the help!
> >
> > Cheers,
> >
> > Nick
> >
> >
> > ------------------------------------------------------------------------------
> >
> >
> >
> > _______________________________________________
> > pdl-general mailing list
> > [email protected] <mailto:[email protected]>
> > https://lists.sourceforge.net/lists/listinfo/pdl-general
> > <https://lists.sourceforge.net/lists/listinfo/pdl-general>
> >
>
> ------------------------------------------------------------------------------
> _______________________________________________
> pdl-general mailing list
> [email protected] <mailto:[email protected]>
> https://lists.sourceforge.net/lists/listinfo/pdl-general
> <https://lists.sourceforge.net/lists/listinfo/pdl-general>
>
> ------------------------------------------------------------------------------
> _______________________________________________
> pdl-general mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/pdl-general
------------------------------------------------------------------------------
_______________________________________________
pdl-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/pdl-general
