kernel_config# MCEq.config.kernel_config = 'numpy'# euler kernel implementation (numpy/MKL/CUDA/accelerate). With serious nVidia GPUs CUDA a few times faster than MKL autodetection of fastest kernel below