Re: [find_orb] (6995) Minoyama

Bill Gray Feb 6, 2013

Hi Alessandro,

Hmmm... a bit of a puzzling case here.

> If you solve the orbit using all observations, with _no_perturbing_
> asteroid_ you have the following residuals for the 1956 observations:

I tried generating ephemerides for 1956. I don't have your observations,
but expect that computing orbits with the astrometry available from MPC,
running from 1978 to 2012 and then computing ephemerides for 1956 should
allow me to see any effects due to asteroid perturbations.

What I saw was that toggling asteroid perturbers on or off changed
the predicted location for 1956 by a few tenths of an arcsecond. Seems
strange that you're seeing so much more!

> What is strange with (6995) is another thing.
>
> If you choose Venus as a perturber, the overall residual improves
> from 0.599 to 0.589 but the specific individual residuals of the 1956
> observations deteriorate :
>
> ...In practice, turning Venus on, destroys the benefits of taking into
> account the (big) asteroids.

I've seen some odd effects of this sort, where turning off a perturber
causes residuals to drop. This does _not_ mean that you're better off
ignoring that planet! You can be certain that the actual universe never
shuts off perturbers.

> (2) - Find_Orb is right, Venus is a strong perturber and this suggests
> us that there are probably other unknown asteroid perturbers which almost
> counterbalance the Venus effect.

As it happens, I was indeed unable to resist the urge to add the 300
asteroid perturbers in Jim Baer and Steve Chesley's BC-405 ephemerides.

I ended up using the method suggested in my most recent e-mail on the
subject. BC-405 provides elements at 40-day intervals. Find_Orb
computes the position for each object at the start and end of those
intervals and stores that data (i.e., it only has to be computed once).
During an integration, Find_Orb checks to see if our object is in the
"box" defined by the coordinates of each object at the start and end
of those forty days. It turns out that usually, there are zero or one
perturbers out of the 300 that need to be checked (though I've seen
situations where four needed to be checked).

The end result is that a "full step" with asteroid perturbers turned
on now takes about twice as long as a "full step" without asteroid
perturbers; i.e., the asteroid perturbers double the computational
load. (I've ideas for improving this, but really should get back to
other projects...) I tried it first on (50278) 2000 CZ12, an
object that was strongly perturbed by (15) Eunomia :

http://tech.dir.groups.yahoo.com/group/mpml/message/17235

Find_Orb defaulted to a pretty ugly orbit for this object. Adding in
asteroids didn't help previously, since Find_Orb was limited to Ceres,
Pallas, and Vesta. With BC-405 in the mix, I'm now getting a perfectly
reasonable result.

Anyway... I've tried it with (6995), with no significant effect. So
I don't think this object is experiencing another asteroid perturber.

-- Bill