Hello all,
I've just posted a new "release" Find_Orb at
http://www.projectpluto.com/find_orb.htm
If you're running the previous release version, I strongly
recommend upgrading.
This also caused me to update the page itself; I realized that
the documentation was badly outdated on a few points.
The only update relative to the previous development version is
that Monte Carlo "noise" is now properly weighted. When Find_Orb
computes a MC variant orbit (called a "virtual asteroid" by some), a
bit of Gaussian-distributed random noise is added to each RA and dec
observation, and the orbit fitting these offset observations computed.
If some observations are given, say, twice as much weight in the
orbit fit as others, then they really should have only half the
Gaussian noise added to them. The program now handles this properly.
(And really should have right from the beginning.)
This improvement, and others that appeared in the development
version, are listed in the "revision history" (see the "Revisions"
link at the above page).
I wanted to get this posted because the program is now in a
reasonably stable state, and I'm about to add/modify various parts.
Just to give you an idea of what is apt to happen:
(1) At present, Find_Orb will load up all observations for an
object, but presents an orbit for a sub-arc. (The arc must be less
than about 45 degrees. If it's longer than that, or if Find_Orb
can't get a decent orbit, it shortens the arc until it gets one that
works.) I'll be revising the program so that if the initial orbit
doesn't cover the entire arc, Find_Orb will attempt to extend the
arc automatically. This will also mean better rejection of outliers,
probably using Peirce's criterion or similar.
(2) The program currently is single-threaded. I expect to revise
it to make better use of multi-core machines.
(3) In order to display uncertainties, an alternative format
will be available. Instead of the "standard" eight-line format
(fixed-size font is best for viewing these) :
Orbital elements:
1996 XX1
Perihelion 1998 Mar 8.969329 TT = 23:15:50 (JD 2450881.469329)
Epoch 1997 Oct 13.0 TT = JDT 2450734.5 Earth MOID: 0.0615 Ma: 0.0049
M 322.34260 (2000.0) P Q
n 0.25622619 Peri. 72.47318 -0.50277681 -0.86441429
a 2.45501241 Node 47.71084 0.79213697 -0.46159019
e 0.5741560 Incl. 0.14296 0.34602670 -0.19930484
P 3.85 H 15.8 U 8.4 q 1.04545221 Q 3.86457261
From 13 observations 1997 Oct. 12-22; mean residual 0".485.
...one will be able to select this sort of format:
Orbital elements: 1996 XX1
Perihelion 1998 Mar 8.977252 +/- 0.109 TT (JD 2450881.477252)
Epoch 1997 Oct 12.0 TT = JDT 2450733.5 Earth MOID: 0.0613 Ma: 0.0049
M 322.109164 +/- 0.0574 (More MOIDs or area/mass goes here)
Peri. 72.530699 +/- 0.0861 n 0.256058518 +/- 0.000205
Node 47.657952 +/- 0.0576 a 2.456084084 +/- 0.00131
Incl. 0.143216 +/- 0.000294 e 0.57438774 +/- 0.000178
H 16.6 G 0.15 U 7.1 P 3.85012 +/- .00021
q 1.045339477 +/- 0.000139 Q 3.866828691 +/- 0.0025
From 13 of 17 observations 1997 Oct. 12-22; mean residual 0".485.
Uncertainties in orbital elements _are_ currently calculated, by
the way. They are provided at the end of 'covar.txt' when you do a
full step, and in 'monte.txt' after you do a Monte Carlo run. It's
just that they don't show up in the orbital element display in the
Windows dialog.
Ideally, meaningless digits will be omitted. For example, the
above mean anomaly will be shown as 322.11 +/- .06.
(4) Ability to set sigmas for observed times and magnitudes.
(The sigmas for magnitudes will be used right away. Those for
times... still working on how that will work; but the structure
for setting them will be in place. I don't know of any software
that currently handles this issue, but uncertainty in time is
_very_ important for poorly-timed fast-moving objects.)
Note that some of these appear on the "wish list" at
http://www.projectpluto.com/find_orb.htm#The_future
...as well as several other "wishes" that aren't apt to make it
into this next batch of changes.
-- Bill