[metapost] bug in arctime calculation

[luigi scarso]

On Sun, Aug 16, 2020 at 5:09 PM Hans Hagen <j.hagen at xs4all.nl> wrote:

> On 8/16/2020 2:19 PM, luigi scarso wrote:
>
> > Both binary and decimal uses and epsilon = 1E-52
> > buth double has epsilon = 2^-52 and this matches with the binary search.
> > Locally I have set  epsilon = 2^-52 for both binary and decimal
> > and the results  are now:
> I had been staring at that but wondered if
>
> @d epsilonf                pow(2.0,-52.0)
>


>
> was to be uised instead.


 epsilonf  is used only in the "localized" mp_<number_system>_cross_point
but in this case there is only one  mp_solve_rising_cubic that uses epsilon
which is, for binary and decimal, 10^-52 << 2^-52 .
Probably  I should  set
@d epsilon  pow(2.0,-173.0)
given that
1E52*(2^-173) =~0.83523897190381113942




-- 
luigi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://tug.org/pipermail/metapost/attachments/20200816/bee5fca1/attachment.html>