作者: Benjamin D. Wandelt , Svetlin Tassev , Matias Zaldarriaga , Daniel J. Eisenstein
DOI:
关键词: Astrophysics 、 Embarrassingly parallel 、 Classical mechanics 、 Acceleration 、 Perturbation theory (quantum mechanics) 、 Code (cryptography) 、 Volume (computing) 、 COLA (software architecture) 、 Gravitational potential 、 Physics 、 Algorithm 、 Cluster analysis
摘要: We present sCOLA -- an extension of the N-body COmoving Lagrangian Acceleration (COLA) method to spatial domain. Similar original temporal-domain COLA, is for solving large-scale structure in a frame that comoving with observers following trajectories calculated Perturbation Theory. Incorporating code allows one gain computational speed by capturing gravitational potential from far field using perturbative techniques, while letting solve only near field. The and fields are completely decoupled, effectively localizing gravity side code. Thus, running small simulation volume can reproduce results standard run same embedded inside much larger simulation. demonstrate be safely combined COLA. used as performing zoom-in simulations. It also codes made embarrassingly parallel, thus allowing efficiently tiling interest grid computing. Moreover, useful cheaply generating large ensembles accurate mock halo catalogs required study galaxy clustering. Surveys will benefit most ones aspect ratios, such pencil-beam surveys, where easily capture effects transverse modes without need substantially increase simulated volume. As illustration method, we proof-of-concept simulations freely available sCOLA-based