One way to embed both GPU version, KOKKOS version (CUDA, OMP, std::threads) is to use shared library and only the library interface to LAMMPS. In that way we don't need to include any header files in Atomify so we can easily use different compiled LAMMPS versions.
One way to embed both GPU version, KOKKOS version (CUDA, OMP, std::threads) is to use shared library and only the library interface to LAMMPS. In that way we don't need to include any header files in Atomify so we can easily use different compiled LAMMPS versions.