I really like what I can do with extended arithmetic in J, continued
fractions and such.  I'd like to help, if I can.

I attempted to extend a Fibonacci verb using the "power of phi" method (
https://code.jsoftware.com/wiki/Essays/Fibonacci_Sequence#Power_of_phi),
which is so simple compared to many of them, but suffers from its
dependence on phi.  So, we could rewrite it to use a pre-calculated
extended phi - maybe using this
https://code.jsoftware.com/wiki/Wiki/Report_of_Meeting_2022-01-13 - by
replacing the in-line limited precision version of it in the code.
However, when I tried to do this, the verb quickly ran out of gas.  I don't
know how long some of the tasks took because I had to cancel them because
they were so slow compared to other versions of the algorithm.

Anyway, I have at least one idea for a benchmark.  Anything I can do to
help would be fine as it seems like a worthwhile project.




On Tue, Jan 25, 2022 at 8:38 PM Julian Fondren <[email protected]>
wrote:

> I'll get it done.
>
> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>
> On Tuesday, January 25th, 2022 at 7:23 PM, Henry Rich <
> [email protected]> wrote:
>
> > Please, somebody, step up to replacing J's antiquated
> >
> > extended-arithmetic support with GMP.
> >
> > Henry Rich
> >
> > On 1/25/2022 8:21 PM, Marshall Lochbaum wrote:
> >
> > > All of these solutions except F# and Raku end up calling GMP for this
> > >
> > > (which uses Miller-Rabin; so does J). I think F# uses the following
> .NET
> > >
> > > library, while Raku appears to eventually call an npm package
> > >
> > > jsbi-is-prime?
> > >
> > >
> https://github.com/Open-NET-Libraries/Open.Numeric.Primes/blob/master/source/MillerRabin.cs
> > >
> > > Even so, it's not particularly fast: I timed the Julia version at 15s
> to
> > >
> > > solve the base task. Clearly the limits are chosen to be moderately
> > >
> > > taxing for the fastest libraries out there; anything without years of
> > >
> > > optimization effort is left behind.
> > >
> > > Adding x: to your version runs out of memory for some reason, but it
> > >
> > > looks like the following would probably finish in half an hour or so.
> > >
> > > {{1+I. 1 p:y#.x:|.+./\.=i.1000}}&.> 2+i.15
> > >
> > > Marshall
> > >
> > > On Tue, Jan 25, 2022 at 05:52:40AM -0500, Raul Miller wrote:
> > >
> > > > http://www.rosettacode.org/wiki/Repunit_primes
> > > >
> > > > Conceptually, this task might be tackled using an expression like
> > > >
> > > >     ":@I. 1 p:(2+i.15) #."0 1/|.+./\\.=i.1000
> > > >
> > > >
> > > > However, some of the numbers being tested for primality here are
> > > >
> > > > rather large. Even the base 2 numbers reach 300 digits, and the base
> > > >
> > > > 16 numbers reach 1200 digits:
> > > >
> > > >     #":2x#.1000#1
> > > >
> > > >
> > > > 302
> > > >
> > > > #":16x#.1000#1
> > > >
> > > > 1203
> > > >
> > > > So... how would we approach this in J?
> > > >
> > > > 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
> >
> > --
> >
> > This email has been checked for viruses by AVG.
> >
> > https://www.avg.com
> >
> > ---------------------------------------------------------------------
> >
> > For information about J forums see http://www.jsoftware.com/forums.htm
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>


-- 

Devon McCormick, CFA

Quantitative Consultant
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to