Several solutions for problems I met

[Elody]

Thanks for your great work! I have met several problems during compiling and find out corresponding solutions to share with everyone.

1. About CUDA version
   I have tried CUDA-9.0, CUDA=10.1 and CUDA-11.1, turns out CUDA-11.1 works for me
   After compiling successfully in CUDA-9.0 and CUDA-10.1, it breaks down when I tried to import package:

mesh_intersection-0.1.0-py3.7-linux-x86_64.egg/bvh_cuda.cpython-37m-x86_64-linux-gnu.so: undefined symbol: _ZNK2at6Tensor6deviceEv

2. double_vec_ops.h not found
   Turns out it's in this repo: ./include, move it to ./src

3. AT_CHECK
   Adding following lines in bvh.cpp before using AT_CHECK

#ifndef AT_CHECK
#define AT_CHECK TORCH_CHECK
#endif

[OOF-dura]

Thanks for your great work! I have met several problems during compiling and find out corresponding solutions to share with everyone.

1. About CUDA version
   I have tried CUDA-9.0, CUDA=10.1 and CUDA-11.1, turns out CUDA-11.1 works for me
   After compiling successfully in CUDA-9.0 and CUDA-10.1, it breaks down when I tried to import package:

mesh_intersection-0.1.0-py3.7-linux-x86_64.egg/bvh_cuda.cpython-37m-x86_64-linux-gnu.so: undefined symbol: _ZNK2at6Tensor6deviceEv

1. double_vec_ops.h not found
   Turns out it's in this repo: ./include, move it to ./src
2. AT_CHECK
   Adding following lines in bvh.cpp before using AT_CHECK

#ifndef AT_CHECK
#define AT_CHECK TORCH_CHECK
#endif

Are you able to install it using conda?