Computing Environments

  • Fortran and C/C++ interfaces provided; Python planned for a future release
  • Five compiler suites officially supported (Cray, GNU, IBM, Intel, PGI)
  • Compiles and run on computing environments as diverse as NERSC’s Cori and Edison, Argonne’s Mira and Theta, Oak Ridge’s Titan and Summitdev, generic Unix-like clusters, and PCs.

Supported Solvers

The Kohn-Sham eigenvalue problem can be explicitly solved by traditional (tri)diagonalization.  In ELSI, the massively parallel dense eigensolver ELPA facilitates the solution of symmetric or Hermitian eigenproblems on high-performance computers.  Algorithm extensions and optimizations are ongoing within the ELPA-AEO project.

The orbital minimization method (OMM) bypasses the Kohn-Sham eigenvalue problem by efficient iterative algorithms which directly minimize an unconstrained energy functional using a set of auxiliary Wannier functions.  The Wannier functions are defined on the occupied subspace of the system, reducing the size of the problem.  The density matrix of the system is then obtained directly, bypassing the need to calculate Kohn-Sham orbitals.

NTPoly is a massively parallel library for computing the functions of sparse, symmetric matrices based on polynomial expansions.  For sufficiently sparse matrices, most of the matrix functions can be computed in linear time.  Distributed memory parallelization is based on a communication avoiding sparse matrix multiplication algorithm.  Various density matrix purification algorithms which compute the density matrix as a function of the Hamiltonian matrix are implemented in NTPoly.

PEXSI is a Fermi operator expansion (FOE) based method which expands the density matrix in terms of a linear combination of a small number of rational functions (i.e. a pole expansion).  Evaluation of these rational functions exploits the sparsity of the Hamiltonian and overlap matrices using selected inversion to enable scaling to 100,000+ of MPI tasks for calculation of the electron density, energy, and forces in electronic structure calculations.

SLEPc-SIPs is a parallel sparse eigensolver for real symmetric generalized eigenvalue problems.  SLEPc-SIPs solves the eigenvalue problem with distributed spectrum slicing method and it is currently available through the SLEPc library built on top of the PETSc framework.