Study on the formation and evolution of galaxies by Large-scale numerical simulations on special-purpose computers.

1.在并行GRAPE系统上实现了Barnes-Hut树算法的高效并行实现。使用这个计算程序，现在可以使用~10^8粒子在比星系团更小的空间尺度上模拟宇宙结构的形成。以前只报道过一次这样的模拟。通过对3000多万个粒子的宇宙学模拟，我们以前所未有的精度和可靠性确定了CDM宇宙学中暗物质晕中心尖点的斜率。斜率在-1.4左右。我们没有观察到斜率有接近-1的趋势，这个斜率不依赖于宇宙学模型。我们已经找到了卫星星系轨道演化的N体模拟和半解析计算之间存在差异的原因。在最近的高精度N体模拟中，卫星星系的演化被证明比使用钱德拉塞卡动力摩擦公式进行近似计算的结果要慢得多。我们发现，这种差异并不是由于动力摩擦公式本身，而是因为人们普遍将母星系的大小作为库仑对数的上限。通过取理论上的偏好值，即卫星到母星系中心的距离，我们解决了这一差异。在卫星内部自由度的情况下，我们发现其演化速度要快于解析的动力摩擦公式。

1. We have developed a high-efficiency parallel implementation of the Barnes-Hut tree algorithm on parallel GRAPE system. Using this calculation code, it is now feasible to perform simulations of cosmological structure formation in the spatial scale smaller than clusters of galaxies using 〜 10^8 particles. Only one such simulation has been previously reported.2. By cosmological simulations with more than 30 million particles, we have determined the slope of the central cusp of the dark-matter halo in CDM cosmology with unprecedented accuracy and reliability. The slope is around -1.4. We have not observed any tendency of the slope to approach to ―1. This slope does not depend on the cosmological models.3. We have found the reason of the discrepancy between N-body simulations and semi-analytic calculations of the orbital evolution of satellite galaxies. The evolution of the satellite galaxies in recent high-accuracy N-body simulations turned out to be significantly slower than the result of approximate calculations using Chandrasekhar's dynamical friction formula. We found that the discrepancy is not due to the dynamical friction formula itself, but due to the common exercise of taking the size of the parent galaxy as the upper limit for the Coulomb logarithm. By taking the theoretical preferred value, namely the distance of the satellite from the center of the parent galaxy, we resolved the discrepancy. In the case of the satellite from internal degree of freedom, we found that the evolution is faster than that for analytic dynamical friction formula.