https://hgpu.org/?p=2803
Accelerating the solution of families of shifted linear systems with CUDA