Re: log_int_rat

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

On 2023-11-29 22:41, Ruud H.G. van Tol wrote:
> On 2023-11-29 22:15, Ruud H.G. van Tol wrote:
> > [...]
>
> I was surprised by this performance difference:
>
> ? localbitprec(10000); my(L = 1-log(2)/log(3)); [ 1 + floor(n/L) 
> |n<-[0..10^5] ];
> cpu time = 1,061 ms, real time = 1,064 ms.
>
> ? localbitprec(10000); my(L = log(2)/log(3/2)); [ 1 + n + floor(n*L) 
> |n<-[0..10^5] ];
> cpu time = 80 ms, real time = 82 ms.
>
> What would explain that?

Unlooping the division, does the trick:

? localbitprec(10000); my(L = 1/(1-log(2)/log(3))); [ 1 + floor(n*L) 
|n<-[0..10^5] ];
cpu time = 81 ms, real time = 82 ms.

-- Ruud