arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1907.09934astro-ph.HEphysics.plasm-ph
收藏

湍流磁场的内插及其对扩散cosmic ray传播的影响

Interpolation of turbulent magnetic fields and its consequences on diffusive cosmic ray propagation

L. Schlegel, A. Frie, B. Eichmann, P. Reichherzer, J. Becker Tjus

摘要:

带电粒子通过磁场传播的数值模拟求解运动方程通常导致在离散定义的磁场的情况下使用插值,通常在均匀网格结构上给出。然而,插值方法会影响网格间距尺度上的磁场特性,因此插值程序的选择会改变结果。与此同时,它提供了一种影响,即\空间粒子分布的误差。我们比较了三种不同的插值例程-三线性,三次和最近邻插值-在湍流磁场的情况下,并表明,使用三线性插值没有好处。我们表明,相比之下,最近邻插值提供了最好的性能,即\需要最少的CPU时间,并导致最小的错误。此外,我们优化了一种算法的性能,该算法产生一个连续的无网格湍流磁场超过一个数量级。这种连续方法对于大粒子数的模拟是可行的,其精度仅受所用波模数的限制。我们表明,通过使用100多个波模式的扩散系数的形式的空间粒子分布的扩散行为是确定的误差小于几个百分点。

英文摘要:

Numerical simulations of the propagation of charged particles through magnetic fields solving the equation of motion often leads to the usage of an interpolation in case of discretely defined magnetic fields, typically given on a homogeneous grid structure. However, the interpolation method influences the magnetic field properties on the scales of the grid spacing and the choice of interpolation routine can therefore change the result. At the same time, it provides an impact, i.e.\ error, on the spatial particle distribution. We compare three different interpolation routines -- trilinear, tricubic and nearest neighbor interpolation -- in the case of turbulent magnetic fields and show that there is no benefit in using trilinear interpolation. We show that in comparison, the nearest neighbor interpolation provides the best performance, i.e.\ requires least CPU time and results in the smallest error. In addition, we optimize the performance of an algorithm that generates a continuous grid-less turbulent magnetic field by more than an order of magnitude. This continuous method becomes practicable for the simulation of large particle numbers and its accuracy is only limited by the used number of wave-modes. We show that by using more than 100 wave-modes the diffusive behavior of the spatial particle distribution in form of the diffusion coefficient is determined with an error less than a few percentage.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑