These are less than a factor of two different, and on my system noise
in timing can swamp any computational difference, despite the size
issues:
ts'~.;<@(*{.)\.>:i.1000'
0.0176966 1.49499e7
ts'~.,*/~>:i.1000'
0.0174547 2.62336e7
But that size difference might be worth worrying about in extreme cases.
I imagine that the time difference might be more stable on larger
arguments, and might favor the shorter expression on smaller
arguments.
--
Raul
On Mon, Jun 25, 2012 at 1:06 PM, R.E. Boss <r.e.b...@planet.nl> wrote:
>
> ts'~.;<@(*{.)\.>:i.1000'
> 0.0072987262 14949888
>
> ts'~.,*/~>:i.1000'
> 0.011011744 26233600
>
> (~.,*/~>:i.1000)-:~.;<@(*{.)\.>:i.1000
> 1
>
>
> R.E. Boss
>
>
>> -----Oorspronkelijk bericht-----
>> Van: programming-boun...@jsoftware.com
>> [mailto:programming-boun...@jsoftware.com] Namens R.E. Boss
>> Verzonden: maandag 25 juni 2012 18:57
>> Aan: 'Programming forum'
>> Onderwerp: Re: [Jprogramming] Multiplying with folds
>>
>>
>> ;<@(*{.)\.>:i.5
>> 1 2 3 4 5 4 6 8 10 9 12 15 16 20 25
>>
>>
>> ~.,*/~>:i.5
>> 1 2 3 4 5 6 8 10 9 12 15 16 20 25
>>
>>
>> R.E. Boss
>>
>>
>> > -----Oorspronkelijk bericht-----
>> > Van: programming-boun...@jsoftware.com
>> > [mailto:programming-boun...@jsoftware.com] Namens Raul Miller
>> > Verzonden: maandag 25 juni 2012 18:51
>> > Aan: Programming forum
>> > Onderwerp: Re: [Jprogramming] Multiplying with folds
>> >
>> > Why do you want this?
>> >
>> > If it's efficiency, note that the size of the dense array */~N is
>> > typically less than the size of a linked list which only represents
>> > the products of unique combinations of pairs (if we assume that linked
>> > list pointer size is approximately the same size as a number), and
>> > also note that typical cpu cache implementation heavily favors regular
>> > operations.
>> >
>> > If it's because you need unique values, you might consider using:
>> >
>> > ~.,*/N
>> >
>> > If it's for some other purpose, the topic might have inherent interest...
>> >
>> > Thanks,
>> >
>> > --
>> > Raul
>> >
>> > On Mon, Jun 25, 2012 at 12:37 PM, David Vaughan <purpleblue...@gmail.com>
>> > wrote:
>> > > I'm looking for a verb to achieve the same thing as just the unique items
>> > > of */~ N, i.e. n0*(n0 ... nk) n1*(n1 ... nk) n2*(n2 ... nk) ... nk*nk.
>> > >
>> > > I can easily create the result, but I want to be able to calculate it
>> > > more
>> > > efficiently, without having to do the multiplications that will
>> > > ultimately
>> > > be redundant (which would be ~.,*/~ N).
>> > >
>> > > I'm not particularly good with folds and similar concepts, so I expect
>> > > the
>> > > solution is relatively simple. Can anyone help?
>> > >
>> > > Many thanks.
>> > > ----------------------------------------------------------------------
>> > > For information about J forums see http://www.jsoftware.com/forums.htm
>> > ----------------------------------------------------------------------
>> > For information about J forums see http://www.jsoftware.com/forums.htm
>>
>> ----------------------------------------------------------------------
>> For information about J forums see http://www.jsoftware.com/forums.htm
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
