https://hgpu.org/?p=6885
Parallelizing Kernel Polynomial Method Applying Graphics Processing Units