Dear participant of the N-body Summer School:
Welcome to Amsterdam! Although our Summer School is still three
months away, it is not too early to start preparing, in terms of
background reading. Here are some suggestions, centered around the
five major software packages that will be available during the Summer
School:
1) -- for the NEMO package, go to "http://www.astro.umd.edu/nemo/"
from where you find instructions to download, install and tinker with
the software. NEMO is a toolbox (UNIX programs and libraries to help
you build new programs) for stellar dynamics. It has tools to create
initial conditions, various integrators (treecode, PPM, direct
N-body), plotting, analyzing, and tools to import/export data to
various other popular formats. In addition to snapshots, NEMO also has
tools to operate and inter-convert to/from images, orbits and tables
to help you directly compare models to data.
2) -- for the Starlab package, go to http://www.ids.ias.edu/~starlab/
Starlab is a collection of modular software tools designed to simulate
the evolution of stars and stellar systems and to analyze the
resulting data. Starlab consists of a library of loosely coupled
programs, sharing a common flexible data structure, which can be
combined in arbitrarily complex ways to study the dynamics of binary
and multiple star systems, star clusters and galactic nuclei. Starlab
contains modules to perform a variety of tasks related to stellar
dynamics and stellar evolution, such as creation of initial models,
following the evolution of stellar systems, and visualizing and
reducing the results. Individual Starlab modules may be linked in the
"traditional" way, as function calls to C++ (the language in which
most of the package is written), C, or FORTRAN routines, or at a much
higher level--as individual programs connected by UNIX pipes. The
former linkage is more efficient, and allows finer control of the
package's capabilities; however, the latter provides a quick and
compact way of performing test simulations and managing large
production runs.
The centerpiece of Starlab is a dynamical integrator known as kira.
Designed as an independent alternative to Aarseth's NBODY4 and NBODY5
the workhorses of collisional N-body calculations for the past 25
years, kira is a high-order predictor-corrector scheme designed for
simulations of collisional stellar systems. Briefly, kira
incorporates a Hermite integration scheme, a block time step
scheduler, and a adaptive binary tree structure that allows
homogeneous treatment of all objects in the system, including
hierarchical structures of arbitrary complexity. The modular internal
organization of Starlab allows additional pieces of physics, such as
stellar and binary evolution, or hydrodynamical simulations of
particular stellar interactions, to be added, tested, and modified
without major changes in the rest of the code. While kira is designed
to operate efficiently on general-purpose computers, it achieves by
far its greatest speed when combined with GRAPE hardware.
Stellar and binary evolution are incorporated in Starlab using the
SeBa package. The evolution of single (isolated) stars in SeBa is
computed using fitting formulae for the evolution of individual stars,
modified to include the effects of binary evolution, accretion from a
stellar wind, Roche-lobe overflow, and physical mergers. The orbital
elements of the binary and the evolution of the component stars are
followed together via a robust iterative procedure, ensuring accurate
treatment of all phases of binary evolution, to the extent that the
underlying physics is presently understood.
The entire Starlab package, including kira, SeBa, and the GRAPE
interface software, is available from the Starlab Web site
"http://www.ids.ias.edu/~starlab", where on-line documentation of
Starlab, its various modules, and its companion projects may be found.
3) -- for the ACS package, you can find codes and extensive
documentation on "http://www.ArtCompSci.org" . This web site contains
more than 1,000 pages of text, so you may want to browse first, before
you decide what to read. There are two ways to get started:
3.1) you can read the 250-page volume "Moving Stars Around", that
forms a self-contained introduction to stellar dynamics,
with sample codes written in C++. Starting with the most
simple type of forward Euler integration for the 2-body problem,
the book moves on toward leapfrog and Hermite integrators for
the general N-body problem, and then ends with example experiments
including cold collapse and binary formation.
3.2) you can start with the Kali series of volumes. Here the codes
are written using the scripting language Ruby, which is somewhat
similar to Perl or Python, but designed as an object-oriented
language from the start. In volume 1, a quick
introduction to Ruby is presented, centered around
the example of integrating a two-body orbit, using forward Euler.
Other volumes present dozens of other integration schemes, for
the general N-body problem, for constant time steps, for adaptive
but still shared time steps, and for individual time steps,
including multistep methods of arbitrary order. Finally, we
present codes where individual particles can choose individual
algorithms, within the same N-body run.
4) -- for the EZ package,, you can look at the website,
"http://theory.kitp.ucsb.edu/~paxton/". EZ is a stellar evolution code
written in Fortran 90 and derived from the decades long efforts of
Peter Eggleton and associates. Stellar evolution is not the main
topic of this conference, but, as you know from MODEST, it will be a
part of future modeling efforts coupling it with hydrodynamics and
stellar dynamics. While there will not be formal presentations
covering EZ and stellar evolution, Bill Paxton will be attending the
summer school and will be happy to answer questions. He'll also
gladly talk about his newest project, Tioga, which combines Ruby, PDF,
and TeX in a package for plotting and making figures.
5) -- for the MMAS (Make Me A Star) package, go to
"http://faculty.vassar.edu/lombardi/mmas". MMAS is a package for
quickly generating stellar collision products. Detailed modeling of
collisions require lengthy hydrodynamic computations in three
dimensions that are currently unfeasible to include in cluster
simulations. However, a much less computationally expensive approach,
which MMAS provides, is to approximate the merger process (including
shock heating, hydrodynamic mixing, mass ejection, and angular
momentum transfer) with simple algorithms based on conservation laws
and a basic qualitative understanding of the hydrodynamics. These
algorithms have been fine tuned by comparing to the results of
previous hydrodynamic simulations of colliding main sequence stars.
MMAS can therefore be used in realistic dynamical simulations of star
clusters that include physical collisions.
MMAS needs as input the periastron separation of the orbit of the
colliding parent stars, as well as their internal profiles (radius,
pressure, density and optionally chemical composition versus enclosed
mass). The same profiles are then returned for the collision product.
The thermodynamic and chemical composition profiles of the collision
product models agree very well with those from hydrodynamical
calculations of stellar collisions, and the subsequent stellar
evolution of these models also matches closely that of the more
accurate hydrodynamic models.
There is also an SPH code available, you can find those on
"http://www.astro.northwestern.edu/StarCrash/"
and we will probably also cover this. Feel free to download and tinker
with this version.
These are the 5 core packages which we will use at some point during
the summerschool. There will be a larger set of reference material
and ancillary codes available, which we may wind up using in some of the
subgroups that we will likely form during the school.
It is probably a good idea to sample all five of these packages,
before you decide which ones to delve into further. We strongly
encourage you to explore at least one of those five packages before
arriving in Amsterdam, so that you will have a flying start.
If you have problems with the installation, don't spend too much time
debugging. Sent a brief message to teuben@astro.umd.edu with a
description of the problem, or attach the screen output of what seems
to go wrong, and Peter will coordinate with the respective authors
what to do about it. Most of the packages should run out of the box
on a number of unix, and possibly cygwin, versions, which should cover
pretty much anybody.
Finally, could you please send teuben@astro.umd.edu an email if you
are going to bring your own laptop, and if so, their basics (e.g Dell
8100, 1.6 GHz P4, 512 MB, 10 GB free diskspace, Redhat9.0). The most
important one he needs to know is your Unix version. Most Linux and
Mac versions of Unix are fine. If you bring a Windows machine, you
might find yourself using ssh, unless you can stomach the limitations
of cygwin.
Greetings from Bill,
Douglas,
Jamie,
Jun,
Peter,
Piet,
Simon,
Steve.
This document was generated
by Simon Portegies Zwart on April, 22 2005
using texi2html