Portable parallel implementation of symmetry-adapted perturbation theory code

We describe a parallel implementation of symmetry-adapted perturbation theory of intermolecular interactions. Utilizing only sequential BLAS operations on distributed data structures and simplest message-passing calls for inter-process communication, the code is highly portable and runs efficiently on a variety of parallel systems, including the high-end shared- and distributed-memory machines, as well as on Linux clusters. A part of the code is a parallel implementation of the coupled cluster singles and doubles method, which has been shown to scale efficiently on as many as 64 processors.