Make barycentered TOAs work correctly about pint HOT 9 CLOSED

@paulray is working on this.

from pint.

Currently if you call model.phase(toas.table), it tries to evaluate the full model for each TOA. It actually should not evaluate the solar system delay terms for TOAs at the 'Barycenter'. This needs to be fixed so that one could have TOA tables that intermix Barycentric and normal TOAs.

from pint.

I think it is likely a one-liner (or near one-liner) that we can put in to the solar_system_geometric_delay() method to return 0 for TOAs where the observatory is Barycenter.

from pint.

Shouldn't this all already work out OK if the observatory position is set correctly (ie to 0,0,0) for the barycentered TOAs?

from pint.

Ah, but that isn't the correct barycenter position.... that is the geocenter position. And we want that as well.

from pint.

(unless you meant in barycenter coordinates!)

from pint.

Yes, I meant the vector stored along with pint TOAs ("toas['ssb_obs_pos']") which is relative to the barycenter. This is what the astrometry class uses.. if it's zero things should work fine although I agree we might want a special case just to avoid unnecessary calculation.

I don't know if barycentered TOAs (which as we've discussed before are kind of a hack to begin with) typically have had SS shapiro delay applied or not... depending on how this was done the shapiro delay part might need disabling, or changing as the observatory wasn't really at the barycenter.

from pint.

It should be easy to correct, but I am having trouble figuring out the Python to make it work (elegantly).
An example of a function that needs to be modified is below. The 'toas' argument is really the toas.table, which is grouped by observatory. So, how do we elegantly loop over the groups and make the delay = 0.0 for the 'Barycenter' key and use the computation for the rest?

def solar_system_shapiro_delay(self, toas):
    """
    Returns total shapiro delay to due solar system objects.
    If the PLANET_SHAPIRO model param is set to True then
    planets are included, otherwise only the value for the
    Sun is calculated.

    Requires Astrometry or similar model that provides the
    ssb_to_psb_xyz method for direction to pulsar.

    If planets are to be included, TOAs.compute_posvels() must
    have been called with the planets=True argument.
    """
    psr_dir = self.ssb_to_psb_xyz(epoch=toas['tdbld'].astype(numpy.float64))
    delay = self.ss_obj_shapiro_delay(toas['obs_sun_pos'],
                                      psr_dir, self._ss_mass_sec['sun'])
    if self.PLANET_SHAPIRO.value:
        for pl in ('jupiter', 'saturn', 'venus', 'uranus'):
            delay += self.ss_obj_shapiro_delay(toas['obs_'+pl+'_pos'],
                                               psr_dir, self._ss_mass_sec[pl])
    return delay

from pint.

I believe this was fixed. If not, someone re-open it.

from pint.