Hi Andrew,

> Nice to hear of all the updates Bill, I especially like the
> linking widely separated arcs news.

This is proving to be a beast of a problem (which, of course,
makes it more interesting). I've gotten involved in it in part
because of the Daily Orbit Updates showing dozens, sometimes
hundreds, of new linkages. Many of these are blatantly obvious,
once you know about them. Others link a few-nighter in 1997 to
a few-nighter in 2011 and are less obvious.

My current goal is that one should be able to load up the
latter sort of case and try out guesses of the "I think the
object completed three orbits between 1997 and 2011... well,
maybe two... OK, let's try four", and have Find_Orb converge
to a least-squares solution for that assumption.

Next goal would be to have something that would comb through
the unnumbered object files and suggest that objects A and B might
be linked. Since there are about 10^5 unnumbered objects, there
are about 10^10 possible ways of linking them; but there are also
ways to recognize that some of those pairings can't possibly work.
Fascinating as this is, it may be a while before I can make much
progress on it.

> I'll also mention that the astrometry for "Natural outer irregular
> satellites of the giant planets" is now available on this page
> http://minorplanetcenter.net/iau/ECS/MPCAT-OBS/MPCAT-OBS.html
> which may give some interesting test cases!

It does indeed. Over the years, I've done a _lot_ with the
irregular satellites, and have gotten solutions for all of them.
They are _not_ apt to load up automatically (some do, but that's
pure luck). All of them do converge to good solutions with a
bit of fiddling about. I give an example at

http://www.projectpluto.com/find_orb.htm#S11

This sort of thing works with most irregulars. I've also been
able to get orbits for the Galileans and for some inner satellites
of Saturn and Uranus, and for Triton:

http://home.gwi.net/~pluto/mpecs/pseudo.htm

In all of these cases, one has to start out with a short arc
of data (about a quarter of an orbit around the primary), and
try some Vaisala orbits at various distances near the primary.
Eventually, you get something where Herget steps converge;
you can then gradually add in other observations. It takes
a little practice. I've hopes of making almost any heliocentric
case "automatic", but few plans to automate natural satellites.

One reason for this is that there aren't all that many natural
satellites, and having to do some user intervention isn't so painful
for them. Someday, I'd like to be able to generate orbits for
all comets and/or all asteroids. If a small percentage of these
requires "hand work", I'd be apt to go insane.

-- Bill