[Kstars-devel] question about KSUtils::lagrangeInterpolation()

[Pablo de Vicente]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

El Sábado, 15 de Mayo de 2004 12:03, Jason Harris escribió:
> Hello,

Hi Jason,

>
> The lagrangeInterpolation() function is used to determine the exact
> moment at which an equinox or solstice occurs, by interpolating from 5
> measurements of the Sun's ecliptic longitude (as determined at noon on
> each of 5 days surrounding the target event).
>
> However, as the function is written, the interpolated Julian Day number
> is returned as a double, which is then recast as a 'long double'.
>
> Since doubles have 10-digit precision, this means the solstice/equinox
> times will only be accurate to +/1.5 minutes.  I suggest we change

Uff, I did not notice this when I wrote the code... That could explain why we 
have an error of approx. 2 minutes for the Spring equinoxes.

> lagrangeInterpolation() so that it uses long doubles internally.

I think this is the simplest solution. The Julian Day is always used in KStars 
as a long double and since the interpolation returns a Julian Day it should 
return a long double.

Can you change it, Jason?.

>
> Alternatively, we could subtract the JD of one of the measured points
> from each of them, so that ten digits would be sufficient precision.
>
> What do you think?
>
> regards,
> Jason

regards,

Pablo.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)

iD8DBQFApobkSItUpHl6kJERAhp2AJ9Y8sDFz9CzWMMOaBAGLpEoUZ4lAACg4F2f
P5GEgqsvACtGsgb/MxCveKw=
=4OyN
-----END PGP SIGNATURE-----