GPAW

GPAW
Name	GPAW
Developer	CECAM
Released	2005
Latest release	2024
Programming language	Python
Operating system	Linux
License	GPL

GPAW

Overview

GPAW is an electronic structure code for plane-wave and real-space calculations developed for ab initio simulations using density functional theory. It originated from collaborations among Centre Européen de Calcul Atomique et Moléculaire, Fritz Haber Institute of the Max Planck Society, Technical University of Denmark, and other institutions such as Lawrence Berkeley National Laboratory, École Polytechnique, and University of Cambridge; the project has ties to projects like Atomic Simulation Environment and toolchains used at Argonne National Laboratory and Sandia National Laboratories. The software targets researchers in fields represented by laboratories like Max Planck Society, CERN, Oak Ridge National Laboratory, and consortia such as PRACE and HPC-Europa3.

Features and Methodology

GPAW implements real-space projector-augmented wave and plane-wave methods within the framework of Kohn–Sham density functional theory, supporting exchange–correlation functionals from families associated with creators like John P. Perdew, Walter Kohn, and theorists linked to Nobel Prize in Chemistry winners. It includes time-dependent extensions relevant to methodologies developed at institutions like Max Planck Institute for Solid State Research and techniques akin to those used by groups at University of California, Berkeley and Massachusetts Institute of Technology. The code provides pseudopotential frameworks comparable with datasets from projects such as PSLibrary, SG15Pseudopotentials, and workflows used by centers like Tsinghua University and ETH Zurich.

Implementation and Architecture

The package is primarily implemented in Python with performance-critical kernels in C and parallelization layers leveraging MPI implementations from distributions used at Argonne National Laboratory and NCSA; it interoperates with build systems and continuous integration platforms adopted by organizations like Travis CI, GitHub, and GitLab. The architecture integrates modules comparable to those in libraries maintained by NumPy, SciPy, and visualization backends similar to tools from ParaView and VMD. Workflow orchestration can hook into resource managers and schedulers typical of infrastructures at National Energy Research Scientific Computing Center, PRACE, and XSEDE.

Applications and Use Cases

Researchers apply the code to simulate materials problems associated with studies at MIT, California Institute of Technology, Harvard University, Stanford University, and industry labs like IBM Research, Microsoft Research, and BASF. Use cases span surface science problems investigated at Max Planck Institute for Solid State Research, catalysis projects tied to Danish National Research Foundation groups, nanostructure modeling relevant to Bell Labs research histories, and spectroscopy simulations analogous to those performed at SLAC National Accelerator Laboratory and Diamond Light Source. It is also used in multi-scale pipelines that incorporate inputs from experimental facilities such as Brookhaven National Laboratory, Argonne National Laboratory, and synchrotron centers like ESRF.

Development and Community

The development community includes contributors affiliated with universities and institutes like Technical University of Denmark, Fritz Haber Institute, University of Vienna, University of Oxford, and industrial partners similar to Siemens and BASF. Governance and contribution workflows follow models adopted by projects such as Linux kernel, Firefox, and scientific collaborations like CERN Openlab; community interaction occurs on platforms used by projects like GitHub, GitLab, Stack Overflow, and mailing lists patterned after Debian and Python Software Foundation communities. Training and outreach have been conducted at summer schools and workshops associated with ICAM and conferences like European Materials Research Society meetings and American Physical Society topical sessions.

Performance and Benchmarking

Performance studies compare the software's scaling and accuracy against codes developed at institutions like Oak Ridge National Laboratory and research groups behind packages such as VASP, Quantum ESPRESSO, ABINIT, and CASTEP. Benchmarks reference hardware platforms and supercomputers like Summit (supercomputer), Fugaku, ARCHER, and cloud services used by Amazon Web Services research teams; profiling often employs tools from vendors such as NVIDIA, Intel, and AMD and parallel debuggers and profilers common at Argonne Leadership Computing Facility. Published comparisons appear in journals and proceedings associated with publishers like American Physical Society, Nature Communications, and Journal of Chemical Physics.

Licensing and Availability

The software is distributed under a copyleft license model used by projects such as GNU Project software, with source code hosted on platforms associated with GitHub and mirrors reflecting practices of Debian and OpenBSD. Binary packages and containers are prepared for ecosystems maintained by organizations like Conda-Forge, Docker Hub, and institutional repositories at European Molecular Biology Laboratory and Max Planck Society to facilitate adoption in research groups at University of California, Berkeley, University of Cambridge, and national labs including Lawrence Livermore National Laboratory.

Category:Scientific software