BigDFT

Key information

• In addition to the traditional cubic-scaling DFT approach, the wavelet-based approach has enabled the implementation of an algorithm for DFT calculations of large systems containing many thousands of atoms, with a computational effort which scales linearly with the number of atoms. This feature enables electronic structure calculations of systems which were impractical to simulate even very recently.

• Uses dual space Gaussian type norm-conserving pseudopotentials including those with non-linear core corrections, which have proven to deliver all-electron precision on various ground state quantities.

• Its flexible poisson solver can handle a number of different boundary conditions including free, wire, surface, and periodic. It is also possible to simulate implicit solvents as well as external electric fields. Such technology is employed for the computations of hybrid functionals and time-dependent (TD) DFT.

Features

• Pseudopotentials

• Boundary Conditions

• Electronic structure

Libraries

The compilation of the code suite relies on the splitting of the code components into modules, which are compiled by the \texttt{bundler} package. This package lays the groundwork for developing a common infrastructure for compiling and linking together libraries for electronic structure codes, and it is employed as the basis for the ESL bundle.

I/O requirements

The largest data files output by BigDFT are those containing the support functions. Thanks to the strict localization of these support functions, the file size remains constant as the system size increases (approximately 1 MB per file), and thanks to the fragment approach, the number of files should also not increase significantly so that even very large systems will have small data requirements. We therefore also have minimal I/O requirements: for BigDFT, I/O is limited to the start and end of a calculation, where there is reading and writing of the support functions to disk. The size and number of files will not (or only to a very limited extent) increase with system size. Currently there may be multiple MPI tasks reading from disk; so far this has not proven to be a bottleneck, but should this prove to be a problem when running at scale it would be straightforward to modify the code so that only one MPI task reads each file and communicates the data. Due to the small file sizes of the generated data, analysis will not require significant computational resources and will be limited to tasks such as calculating pair distribution functions and analyzing atomic coordination. These can either be performed as a post-processing step on a desktop computer or during the simulation.

Diffusion

The code is developed by a few individuals, ranging from 6 up to 15 people. The active code developers are, or have been, located in various groups in the world, including EU, UK, US, and Japan. For these reasons, in addition to production calculations aimed at scientific results, the code has been often employed as a test-bed for numerous case-study in computer science and by hardware/software vendors, to test the behavior of novel/prototype computer architectures in realistic runtimes.

References

[1] Ratcliff et al., Flexibilities of wavelets as a computational basis set for large-scale electronic structure calculations. J. Chem. Phys. 21 May 2020; 152 (19): 194110.

[2] Mohr et al., Accurate and efficient linear scaling DFT calculations with universal applicability, Phys. Chem. Chem. Phys., 2015,17, 31360-31370.