Re: log_int_rat

["Ruud H.G. van Tol" <rvtol@isolution.nl>]

On 2024-01-22 11:25, Ruud H.G. van Tol wrote:
> [...]
> On the non-discrete code branch:
>
>   a(n) = localbitprec(logint(n+2,2)); 1 + n + floor(n*log(2)/log(3/2));
>
> with test:
>
>   {default(parisizemax,2^30); my(v=alist(2^20)); for(n=1,#v, 
> (v[n]!=a(n-1)) && print(n);break);}
>
> I received this report:
>
> > that code wrong result a(83690) = 226760 (in a 32-bit pari)
> > with no hint that it didn't know right from wrong.
>
>
> ? my(v=alist(83700)); v[83691]
> cpu time = 485 ms, real time = 490 ms.
> % 226759

My current guess is that it needs something more like
  localbitprec(2*n+1); ...

-- Ruud