RMG - A REAL SPACE MULTIGRID DFT CODE

Getting Started with RMG

Download Source code and binaries are available here.

Install Binaries should readily work on their respective platforms. For source code install make sure you have required and optional dependencies. Then (Linux/UNIX):

1. tar xvzf rmg*tar.gz
2. Edit settings in CMakeLists.txt (optional)
3. cmake .
4. make rmg-cpu (default, cpu only) or make rmg-gpu (gpu-enabled)
5. (Optional) make install (requires root). This installs the binary to /usr/local/bin and examples to /usr/local/share/rmg

Setup up calculation The best way to learn RMG is by running Examples. These are included in the source, and also here. At minimum, an RMG input file needs to specify atomic coordinates and cell dimensions. Other options are setup automatically, but can be overridden by a user. Details on RMG input file options are here.

Running A typical run command is mpirun -np 6 ./rmg-cpu in.c60. The argument after "np" is the number of cores to use. Use the full path path to the binary. The last argument is an RMG input file. For clusters and supercomputers, follow the documentation specific to your target platform for details on how to submit and run MPI jobs.

Analyze results Most of the calculation data is in the .log file. Wavefunctions and charge density are saved in a wave directory.

More information Consult Documentation on our SourceForge project page. In particular, Quickstart Guide, Examples, and Tutorial (coming later) should be useful.