Hi Tony, Ireneuz,

Comparisons are a bit difficult, because I don't know which
observations are being discarded. MPC does tell us that two
non-gravitational parameters (A1 and A2) were used for their
more recent orbits. (Non-gravs couldn't be computed for the
older orbits). JPL tells us that 2255 observations were used,
with the third A3 parameter solved for.

Neither site provides residuals. I'm seeing the trend Tony
mentioned for "current" observations. That trend remains if you
use A1 and A2, and basically goes away with A3 included.

The problem is that (as best I understand it) you really
shouldn't use A3 in a case like this. In fact, you should
use A3 sparingly at best. A3 corresponds to an "out-of-orbit-plane"
non-gravitational force, one at right angles both to the comet's
motion and to the line connecting it to the sun. So yes,
including A3 makes the residuals much nicer. But I don't know
if it corresponds to the actual forces going on here. It may
simply be an example of how, if you add enough parameters, you
can always get lower residuals. Or, as John von Neumann once
put it: "Give me enough free parameters, and I can model an
elephant. Give me one more parameter, and I can show that the
elephant flies."

You can get a good fit for current observations by giving the
most recent data much lower sigmas than the older observations.
That way, you'll have an orbit matching current data closely,
but with the older data still giving decent long-term behavior.
This _is_ a comet, though, and will do unpredictable things.

-- Bill

On 09/15/2014 07:45 AM, Ireneusz Włodarczyk astrobit@... [find_orb] wrote:
>
>
> It depends on how the nongravitational parameters are computed.
>
> For example, without nongravitational parameter rms=0.810",
> using the MPC's nongravitational parameter we have rms=0.640",
> and with the JPL NASA's ngrv parameters we have rms=0.534".
>
> Computations were made using all 2311 observations of comet C/2012X1 and the OrbFit software.
>
> Best regards,
> Ireneusz Wlodarczyk