kira code

Piet Hut (piet@guinness.ias.edu)
Sat, 26 Nov 94 17:32:24 EST

Greetings!
just a brief update of the work that Steve McMillan, Jun
Makino, and I have done here in Princeton over the last week.

We now have the first working version of our KIRA code.
KIRA is roughly comparable to Sverre Aarseth's NBODY3.
It does not have KS regularization, but it does introduce local
coordinate systems for dense clumps of stars. What is new here is
that these local coordinate systems are constructed in the form of a
binary tree (i.e. two branches at each node) for each clump. All
other particles are placed in a flat top-level tree.

KIRA contains an algorithm for automatically changing the trees on all
levels, guaranteeing an accurate treatment of any type of close
encounters, no matter how many particles are involved in whatever
configuration. This, by the way, explains the name: Kira is an
Integrator with a Recursive Algorithm.

So far, we have included analytic Kepler treatment, both of close
hyperbolic encounters (for a limited period around periastron passage)
as well as for nearly isolated binaries, which are integrated by using
the analytic Kepler orbit expressions.

We plan to make the KIRA code (written in C++) available through
anonymous ftp at IAS in Princeton, together with a mini version of our
starlab environment, as soon as we have cleaned things up a bit,
hopefully within a few months.

Cheers,
Jun, Steve, Piet