Summary: HelioLinC is an algorithm that discovers solar system objects in transient source
catalogs by propagating tracklets along hypothesized orbits and clustering the resulting
heliocentric positions at common reference epochs. For near-Earth objects (NEOs),
which frequently have non-linearly varying heliocentric ranges as a function of time r(t), the
search space of hypothesized orbits is increased compared to other orbit classes.
We can, however, dramatically reduce that space by testing the projected heliocentric
angular sweep of a tracklet against the sweep of the hypothesized orbit. If a tracklet
has a swept angle discrepancy with respect to the hypothesized orbit (a SAD tracklet),
we don’t have to propagate or cluster it which saves both substantial compute time and
reduces the number of confounding tracklets in the clustering phase space. In the limit
where we choose an object’s true r(t) for the hypothesis orbit and reject SAD tracklets,
we can reduce propagation and clustering counts by over 95% without sacrificing
object recovery rates – meaning we propagate and cluster almost nothing other than
the tracklets that conform to the hypothesis orbit. In the more realistic scenario of
searching for objects we do not know the orbits for, we need to increase the tolerance
to account for small discrepancies from the hypothesis orbit. In these instances, we are
still able to reduce the number of tracklets that need to be propagated and clustered by
approximately 85% in a DP0.3 search.
