Comparative study of CUDA GPU implementations in Python with the fast iterative shrinkage-thresholding algorithm for LASSO

A general-purpose GPU (GPGPU) is employed in a variety of domains, including accelerating the spread deep natural network models; however, further research into its effective implementation needed. When using compute unified device architecture (CUDA), which has recently gained popularity, situation analogous to use GPUs and memory space. This due lack gold standard for selecting most efficient approach CUDA parallel computation. Contrarily, as solving least absolute shrinkage selection operator (LASSO) regression fully consists basic linear algebra operations, computation GPGPU more than other models. Additionally, optimization problem often requires fast calculations. The purpose this study provide brief introductions approaches numerically compare computational efficiency with that iterative shrinkage-thresholding algorithm LASSO. contributes providing standards computation, considering both ease implementation. Based on our comparison results, we recommend implementing Python, either dynamic-link library or PyTorch algorithms.