Installing CUDA for deep learning while having IGP for display

Note for installing CUDA for deep learning, while keeping the integrated graphic processor handling display.
(Tested on Debian Stretch 9.3)

Credit goes to:
https://unix.stackexchange.com/questions/218163/how-to-install-cuda-toolkit-7-8-9-on-debian-8-jessie-or-9-stretch
https://www.linkedin.com/pulse/installing-nvidia-cuda-80-ubuntu-1604-linux-gpu-new-victor
https://devtalk.nvidia.com/default/topic/991849/-solved-run-cuda-on-dedicated-nvidia-gpu-while-connecting-monitors-to-intel-hd-graphics-is-this-possible-/

1. Download CUDA installation runfile.
(`cuda_9.1.85_387.26_linux.run` in my case)
(This may take a while, 1.6GB)

2. Install requirements
sudo apt-get install linux-source sudo apt-get source linux-image-$(uname -r) sudo apt-get install linux-headers-$(uname -r) sudo apt-get install gcc g++ libxi-dev libglu1-mesa libglu1-mesa-dev libxmuu-dev linux-headers-amd64 linux-source
3. Install NVIDIA driver and CUDA
(As root, although the runfile will prompt for sudo password if necessary.)
chmod +x cuda_9.1.85_387.26_linux.run ./cuda_9.1.85_387.26_linux.run --no-opengl-libs
(y)es to "install on an unsupported configuration"
(y)es to "install NVIDIA Accelerated Graphics Driver"
(n)o to "run nvidia-xconfig"
(y)es to "CUDA Toolkit"