| Laser indirectly driven Inertial Confinement Fusion (ICF) is one hopeful method of thecontrolled thermonuclear fusion reaction and has attracted more attentions from manycountries including China. Using numerical methods to simulate capsule implosion physicalprocesses can evaluate the quality of the radiation symmetry. Indirectly driven radiationsymmetry is a key indicator in the process of experimental implosion and ignition. As higheraccuracy of numerical calculation is required, we need to discrete the target surface densely,which severely holds back the efficiency of radiation symmetry evaluation. This paper focuson how to accurately and efficiently analyze the radiation symmetry evaluation in Laserindirectly driven ICF.In the process of radiation symmetry, we need to discrete the radiating surface andestablish the balance model. In order to achieve high accuracy, the grids are often dense,which will lead to a large Energy Balance Model. As a result, CPU serial solution is time-consuming. To solve this problem, we propose the Jacobi iteration parallel algorithm underthe CUDA architecture. By mapping the calculation processes of Jacobi iteration to GPUthreads, the energy balance model can be solved efficiently. In the process of building energybalance model, the view factors need be re-calculated frequently. By the parallel technologyof GPU multiprocessor, the efficient calculation of view factor is realized, which acceleratesthe building of the model.The parallel algorithm is proved valid and efficient by several typical cases of SG laserfacility model. By compared with the results of the CPU computation, symmetry evaluationcould achieve a speed-up ratio about1000when discrete scale up to106. At the same time,the calculation error of GPU relative to the CPU is around10-9.As an application of the above research, the parallel algorithm is applied to thesimulation system IRAD3D for ICF under the CUDA architecture. we implement the parallel solution method of energy balance model in the indirect driven of ICF. The results show thatthe algorithm in this paper has promising future in engineering. |