|
Joshua Fuhrman, Aaron Geller, Carl Rodriguez, Fred Rasio | |||||
| |||||
|
In order to understand how modest-sized star clusters evolve over millions or even billions of years we model clusters with 1K - 40K objects. We use a direct integration method of solving the N-body problem (as analytic methods have only been developed for N = 1,2 and some special cases of N =3), implemented with a custom version of the NBODY6 (Aarseth, 2003) code: NBODY6++GPU (Wang, 2015). This version of the code has been restructured to run in parallel on GPUs in addition to CPUs. We also modified the code to define the initial binary frequency and orbital parameter distributions, include supernova kicks, and modify the output format. We draw supernovae kicks from a Maxwellian with a typical velocity of 265 km/s. Supernova fallback is then calculated and the kick is reduced partially or completely. Our initial binary parameters follow Raghavan et al. (2010) and Sana et al. (2012), who empirically define the binary period eccentricity and mass ratio distributions for solar-type and high-mass stars, respectively. Importantly, we define a binary fraction that increases toward higher-mass stars, as observed, with an initial binary fraction approaching 100% for black hole-forming stars. Our current grid of models consists of 179 star clusters, not including an additional 234 clusters used for testing. We create a grid of models, where tor each grid point, we create multiple models drawn from the same initial parameter distributions (though with different initial random seeds) such that the total number of objects produced by all models at a given grid point is approximately equal. Our grid includes five values of N: 1K 5K, 10K, 20K and 40K; and two metallicities: solar and one-tenth solar. The initial half-mass radius is unanimously set at 2 parsecs. |
|||||
|
. Joshua Fuhrman |