This project is motivated by the recent explosion in high-quality data, from some of the most powerful telescopes on earth and space-based observatories, on stellar clusters in many galaxies. Important examples of star clusters include globular clusters, spherical systems containing typically 105 - 107 stars densely packed within radii of just a few light years, and galactic nuclei, even denser systems with up to ~109 stars contained in similarly small volumes, and often surrounding a supermassive black hole at the center.
Understanding the dynamical evolution of such dense star clusters is of critical importance to many key unsolved problems in astrophysics. It connects directly to our understanding of star formation, as well as galaxy and supermassive black hole formation and evolution. Dynamical interactions in dense star clusters play a key role in the formation of many of the most interesting and exotic astronomical sources, such as bright X-ray and gamma-ray sources, radio pulsars, and supernovae. As tracers of the formation history of galaxies throughout the Universe, going back to the earliest times of the first structure formation, globular clusters also play a crucial role in our modern understanding of cosmology.
Although N-body simulations have been performed for a large number of stars for collision-less systems, the methods employed are unsuitable for dense stellar systems where the evolution is dominated by relaxation. The inclusion of all relevant physics for collisional systems and the steep N2 scaling have limited N-body simulations of dense stellar systems to N ~ 105. However, the typical number of stars in globular clusters and galactic nuclei is 1 to 5 orders of magnitude larger than that.
The Cluster Monte Carlo code which is based on the orbit-averaging method, can simulate the evolution of systems containing up to a few million stars. We developed and use a distributed memory parallel code which is capable of being run on GPUs and large supercomputing clusters, using MPI (Message Passing Interface) for inter-process communication. Our code is highly scalable due to the various parallelization and optimization strategies that we implemented, and these could be easily adapted for other similar large particle astrophysics codes. We have also demonstrated that our code combined with a direct N-body code can be particularly useful for accurately simulating dense stellar systems that are known to possess two nearly de-coupled components a very dense central “core” and a more sparse outer sparse “halo”. We have developed a first-of-its-kind hybrid code SCHMoCK (Star Cluster Hybrid Monte Carlo Kira), which uses the direct N-body method to simulate the core stars and MC to treat the ones in the halo. Our hybrid code can produce the same results as a GPU-accelerated direct N-body code while running at least an order of magnitude faster.
Project Members:
Alok Choudhary
Vicky Kalogera
Wei-keng Liao
Gokhan Memik
Meagan Morscher
Bharath Pattabiraman
Fred Rasio
Carl Rodriguez
Stefan Umbreit
Kevin Broh-Kahn
Publications:
K. J. Joshi, F. A. Rasio, and S. P. Zwart. Monte carlo simulations of globular cluster evolution. I. method and test calculations. The Astrophysical Journal, 540(2):969, 2000.
K. J. Joshi, C. P. Nave, and F. A. Rasio. Monte Carlo simulations of globular cluster evolution. II. Mass spectra, stellar evolution, and lifetimes in the galaxy. The Astrophysical Journal, 550(2):691, 2001.
J. M. Fregeau and M. A. Gurkan and K. J. Joshi and F. A. Rasio. Monte Carlo Simulations of Globular Cluster Evolution. III. Primordial Binary Interactions. The Astrophysical Journal, 593:772, 2003.
J. M. Fregeau and F. A. Rasio. Monte Carlo simulations of globular cluster evolution. IV. Direct integration of strong interactions. The Astrophysical Journal, 658(2):1047, 2007.
S. Chatterjee, J. M. Fregeau, S. Umbreit, and F. A. Rasio. Monte Carlo simulations of globular cluster evolution. V. Binary stellar evolution. The Astrophysical Journal, 719(1):915, 2010.
S. Umbreit, J. M. Fregeau, and F. A. Rasio. Monte Carlo simulations of globular cluster evolution. VI. The influence of an intermediate mass black hole. The Astrophysical Journal, 750(1):31, 2012.
B. Pattabiraman, S. Umbreit, W. Liao, F. Rasio, V. Kalogera, and A. Choudhary. "GPU- Accelerated Monte Carlo Algorithm for Simulating Dense Stellar Systems", Astronomical Society of the Pacific Conference Series 2012.
B. Pattabiraman, S. Umbreit, W. Liao, F. Rasio, V. Kalogera, G. Memik, and A. Choudhary, "GPU-Accelerated Monte Carlo Simulations of Dense Stellar Systems", to appear in 2012 Innovative Parallel Computing: Foundations & Applications of GPU, Manycore, and Heterogeneous Systems (InPar 2012), San Jose, USA, May 2012.
B. Pattabiraman, S. Umbreit, W. Liao, A. Choudhary, V. Kalogera, G. Memik, and F. Rasio. A Parallel Monte Carlo Code for Simulating Collisional N-body Systems., The Astrophysical Journal Supplements 2012.
S. Chaterjee, S. Umbreit, J. M. Fregeau, and F. A. Rasio. Understanding the Dynamical State of Globular Clusters: Core-Collapsed Versus Non-Core-Collapsed. Monthly Notices of The Royal Astronomical Society, 429(4):2881, 2013.
S. Chatterjee, F. A. Rasio, A. Sills, and E. Glebbeek. Stellar Collisions and Blue Straggler Stars in Dense Globular Clusters. The Astrophysical Journal, 777(2):106, 2013.
M. Morscher, S. Umbreit, W. M. Farr, and F. A. Rasio. Retention of Stellar-Mass Black Holes in Globular Clusters. The Astrophysical Journal Letters, 763(1):L15, 2013.
M. Morscher, B. Pattabiraman, C. Rodriguez, F. A. Rasio, and S. Umbreit. The Dynamical Evolution of Stellar Black Holes in Globular Clusters. The Astrophysical Journal, 800(1):9, 2015.
Downloads:
The parallel Tausworth113 Random Number Generator we use for both our CUDA and MPI codes can be downloaded HERE.