A.4 Parallel runs across different file systems

In parallel runs on distributed computers (clusters), FHI-aims expects its input files in the directory from which it is invoked. Its standard output can be redirected by hand to any given location, and other (optional, see keyword output) output files are again written in the same directory from which FHI-aims is invoked.

This procedure works well on most standard cluster and/or high-performance computing architecture available today, but you must make sure that the directories for input and output are visible and readable / writable on all the nodes across which FHI-aims is parallelized for a given run.