Zheyong Fan

OK, so > Hi @brucefan1983, thanks for the suggestions and clarifications. > > > I think for system virial (sum of per-atom virials), you also don't need to force it...

Thanks for the explanations. I have no more questions regarding the virial/stress, except that I think it would also be nice to have testing results using LAMMPS (there are options...

Hello, sorry for disturbing again. Thanks @sirmarcel for confirming that heat current for message-passing construction is a little more complicated than "local" machine-learned potentials. I also read your preprints and...

@initqp Do you think there is a potential bug for the PLUMED interface in GPUMD?

You can try to change `CFLAGS = -std=c++14 -O3 -arch=sm_60` to `CFLAGS = -std=c++14 -O3 -arch=sm_37` in `src/makefile` and try again (`make clean` and then `make`).

Then I guess CUDA code does not work in your platform at all. You can try to compile and run the folloiwng simplest CUDA code: ```c++ #include __global__ void hello_from_gpu()...

Being able to compile and run the simplest CUDA code means you have a working CUDA platform. Then did you run `gpumd` (or `nep`) from comand line directly? The error...

If possible, could you change a platform to test?

I would like to close this if there is no more discussion. I believe this is a problem related to the CUDA environment instead of GPUMD.