On the GPU Parallel Computing for Sommerfeld Integral Tails

The Sommerfeld integrals (SIs) in layered media are notoriously difficult to calculate due to their oscillatory nature, and their evaluation can require a significant amount of computational resources. This letter investigates the acceleration of numerical integrations of SIs by using the(GPU) processor. With the advent of high-performance computing architectures, such as graphics processing units (GPUs), it has become possible to accelerate SI evaluation, leveraging the parallelism inherent in these architectures. By distributing the workload across a large number of processing units, GPUs can significantly reduce the computational time required to evaluate these integrals. Implemented using the CUDA platform, up to 24 times speedup is achieved compared to a state-of-art multicore CPU with OpenMP parallelism. We render the proposed method can be applied to a wide range of problems in electromagnetics, acoustics, and optics, among other fields.