Ab Initio Random Structure Searching
Ab initio Random Structure Searching (AIRSS) is a very simple, yet powerful and highly parallel, approach to structure prediction. The concept was introduced in 2006 and its philosophy more extensively discussed in 2011.
Random structures - or more precisely, random "sensible" structures - are generated and then relaxed to nearby local energy minima. Particular success has been found using density functional theory (DFT) for the energies, hence the focus on "ab initio" random structure searching. The sensible random structures are constructed so that they have reasonable densities, and atomic separations. Additionally they may embody crystallographic, chemical or prior experimental/computational knowledge. Beyond these explicit constraints the emphasis is on a broad, uniform, sampling of structure space.
AIRSS has been used in a number of landmark studies in structure prediction, from the structure of SiH4 under pressure to providing the theoretical structures which are used to understand dense hydrogen (and anticipating the mixed Phase IV), incommensurate phases in aluminium under terapascal pressures, and ionic phases of ammonia.
The approach naturally extends to the prediction clusters/molecules, defects in solids, interfaces and surfaces (interfaces with vacuum).
The AIRSS package is tightly integrated with the CASTEP first principles total energy code. However, it is relatively straightforward to modify the scripts to use alternative codes to obtain the core functionality, and examples are provided.
Contact: airss@msm.cam.ac.uk
Obtaining AIRSS
The AIRSS package is released under the GPL2 licence. You can download the full source code below. Note that you will require a Unix-like environment in order to build the code. Windows users can install a Linux subsystem if required. Build instructions are contained within the source:
As well as the full source, you can access an online version of the AIRSS code which lets you try some of the AIRSS examples in-browser. Read more about this online version in the associated news item, or access it below:
Access the online version of AIRSS
Documentation
The AIRSS source contains a number of worked examples to assist the user in learning how to use the code. Online AIRSS documentation is also under construction at https://airss-docs.github.io/ (with thanks to James Walsh, UMass Amherst)
Ephermeral data derived potentials
From v0.9.3 AIRSS can exploit EDDPs to potentially dramatically accelerate structure search. The EDDP package can be downloaded here.
History
Earlier versions of AIRSS: