> NB. there must be some optimisation here?!
> ts'(**<.@:**)(100000$i:5)--:ct'
> 0.0108771 3.67232e6
There are no optimizations. The expression is fast
because it avoids the most expensive operation
executed by the other expressions -- the monad <.
on floats.
BTW, the result of the expression is a float and
is not necessarily an exact integer:
] b=: (**<.@:**) 5+-:ct
5
b-5
2.84217e_14
----- Original Message -----
From: Mike Day <[EMAIL PROTECTED]>
Date: Monday, October 23, 2006 11:12 am
Subject: Re: [Jprogramming] fuzzy <.0
> ... but the timings aren't so good - perhaps that
> doesn't matter.
>
> ts'<.`0:@.(0=*)("0) (1000000$i:5)--:ct'
> 0.995482 5.82e7
>
> NB. similar avoiding the "0 test:
> ts'(<.*0:~:*) (1000000$i:5)--:ct'
> 0.279156 3.04111e7
>
> NB. there must be some optimisation here?!
> ts'(**<.@:**)(100000$i:5)--:ct'
> 0.0108771 3.67232e6
> M
>
> bill lam wrote:
> > bill lam wrote:
> >
> >> I also try <.`0:@.(1:=>:)("0)
> >>
> >
> > another candidate may be <.`0:@.(0:=*)("0) as monad * is
> tolerant when comparing
> > with 0.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
