> It's probably worth keeping in mind that originally, Iverson did not
> want any length errors on arrays, but because length errors catch so
> many problems they have been incorporated in the language.
How do you know this? If there is a citation to a paper backing this up
that'd be good.
On Sat, Nov 3, 2012 at 7:30 AM, Raul Miller <rauldmil...@gmail.com> wrote:
> Brian Schott already answered your question?
>
> Remember we are working with polynomials. So, from a results
> perspective, these are equivalent:
>
> 1 2 1 p. y
> 1 2 1 0 0 p. y
> 1 2 1 0 0 0 0 p. y
>
> You will probably notice a relationship between these zeros and the
> leading zeros in decimal numbers and the leading ones in array shape?
>
> It's probably worth keeping in mind that originally, Iverson did not
> want any length errors on arrays, but because length errors catch so
> many problems they have been incorporated in the language.
>
> In contexts where length errors are not appropriate, we have a variety
> of mechanisms available to us:
>
> We can find the length of both arrays and use take ({.) with their
> maximum (best in explicit contexts) -- x dyad&((x>.y)&{.)
>
> We can join the arrays using ,: and then reduce them (dyad/)@,:
>
> We can use sparse arrays with an arbitrarily large array index x
> dyad&(9e9{.$.) y -- note that _ does not work here, as a length, note
> also that if we take this approach we will need to extract the array
> size somehow, later, if we ever want a dense array, and finally note
> that this use of "array shape" starts feeling more like the concept of
> "type" popular in some other languages (as opposed to "dependent type"
> -- here, it's just something arbitrary which distinguishes between
> otherwise identically appearing structures). We have some other
> issues with sparse arrays, also...
>
> And, of course, we can often find algorithmically relevant ways of
> handing array length. These tend to be related to the purpose of our
> algorithms.
>
> I hope this helps,
>
> --
> Raul
>
> On Fri, Nov 2, 2012 at 5:36 PM, Linda Alvord <lindaalv...@verizon.net>
> wrote:
> > I answered the wrong message. What about the final two zero's in the
> > result?
> >
> > Linda
> >
> >
> > -----Original Message-----
> > From: programming-boun...@forums.jsoftware.com
> > [mailto:programming-boun...@forums.jsoftware.com] On Behalf Of Raul
> Miller
> > Sent: Friday, November 02, 2012 10:46 AM
> > To: programm...@jsoftware.com
> > Subject: Re: [Jprogramming] Taylor series
> >
> > On Fri, Nov 2, 2012 at 5:18 AM, Linda Alvord <lindaalv...@verizon.net>
> > wrote:
> >> Raul, I haven't gotten to t. yet, but I did manage not to use (f*g) or
> > p.
> >>
> >> f=: 1 2 1&p.
> >> g=: 1 3 3 1&p.
> >> x=: 10%~i=: i.8
> >> ]c=: (f*g) t. i NB. This still has problems
> >> 1 5 10 10 5 1 0 0
> >
> > What problems?
> >
> > 1 2 1 +//.@:(*/) 1 3 3 1
> > 1 5 10 10 5 1
> >
> > Thanks,
> >
> > --
> > Raul
> > ----------------------------------------------------------------------
> > 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
