论文部分内容阅读
The Dirac equation[l] is very important in many fields of physics ranging from relativistic quantum mechanics to quantum field theory.For a long time, it is a formidable task to solve the equation for arbitrary time-dependent potential.Because of high-resolution demands in both space and time to insure the convergence of calculations, numerical procedures are usually involving huge size of grids and lots of integration steps, which represents high computational costs.Some existing attempts[2-4] are limited in the low spatial dimensions.It is a question that to what extent their calculations are a good approximation of fully 3D reality.Benefiting from the great progresses on modern cluster systems or supercomputers, we develop an efficient parallel code for solving the time-dependent Dirac equation in full 3D space, adopting the fast Fourier transform (FFT) split operator method[5].It can be used on up to hundreds of thousands computational cores.The high level of parallelization is based on the highly scalable 2DECOMP&FFT library[6], which is a Fortran library including a 2D domain decomposition algorithm for 3D Cartesian data structures and a distributed FFT interface.Some calculation examples are given to demonstrate the validity of the code.It is hopeful that the code will be applied to relativistic strong field physics in the near future.