Yang C, Li L, Zhang Y. Development of a scalable solver for the earth’s core convection. In: Zhang W, Chen Z, Douglas CC, Tong W Proc. 2nd International Conference on High Performance Computing and Applications (HPCA 2009), Lecture Notes in Computer Science. Vol. 5938. Shanghai, China: Springer Berlin Heidelberg; 2010. pp. 497–502. 访问链接Abstract
A scalable parallel solver is developed to simulate the Earth’s core convection. With the help from the “multiphysics” data structure and the restricted additive Schwarz preconditioning in PETSc the iterative solution of the linear solver converges rapidly at every time-step. The solver gains nearly 20 times speedup compared to a previous solver using least-squares polynomial preconditioning in Aztec. We show the efficiency and effectiveness of our new solver by giving numerical results obtained on a BlueGene/L supercomputer with thousands of processor cores.
The charge-exchange spin-dipole (SD) and spin-quadrupole (SQ) strength functions of 90 Zr are calculated with and without the tensor terms of the Skyrme interaction in self-consistent HF+RPA approach. It is found that, in SD and SQ transitions, the RPA correlations associated with the tensor terms shin dramatically the strengths of ( Y l œÉ ) Œª = l ‚à? and ( Y l œÉ ) Œª = 1 modes upward and downward, respectively, and also shift the strengths of ( Y l œÉ ) Œª = l + 1 modes upward. The coupling between ( Y l = Œª ‚à?1 œÉ ) Œª and ( Y l = Œª + 1 œÉ ) Œª modes arising from the tensor correlation is noticeable. The RPA tensor correlations produce strengths of SD and SQ modes, which are distributed in a much wider energy range, and the ( Y l œÉ ) Œª = l ‚à?1 modes dominate the high energy part of the strength functions. These energy shifts and coupling effects of different modes can be understood qualitatively by expressing a finite range tensor force in a separable form.