Introduction

FHI-aims (“Fritz Haber Institute ab initio molecular simulations”) is a computer program package for computational materials science based only on quantum-mechanical first principles. The main production method is density functional theory (DFT) [145, 172, 77] to compute the total energy and derived quantities of molecular or solid condensed matter in its electronic ground state. In addition, FHI-aims allows to describe electronic single-quasiparticle excitations in molecules using different self-energy formalisms (e.g., $GW$ and MP2), and wave-function based molecular total energy calculation based on Hartree-Fock and many-body perturbation theory (e.g., MP2, RPA, SOSEX, or the more encompassing renormalized second-order perturbation theory, RPT2).

Online tutorials for FHI-aims can be found at

https://fhi-aims-club.gitlab.io/tutorials/tutorials-overview/

or via a link at

https://fhi-aims.org/

The basic physical algorithms in FHI-aims concerning ground state DFT and applications are described in

This is an Open Access paper and everyone should be able to obtain a copy at:
https://doi.org/10.1016/j.cpc.2009.06.022 .
Please cite this reference if you use FHI-aims.

However, FHI-aims is not just a product of this basic reference. Many more developments make this code a reality. For each individual FHI-aims run, a list of references describing the specific methods used is given at the end of the FHI-aims standard output. Please give credit in your publications if you can. FHI-aims is a scientific code, written by and for scientists. The primary recognition for their work is credit in the form of appropriate reference to their work.

Some particularly important papers (also worth reading!) follow below.

A list of methodological publications for specific methods in FHI-aims can also be found at

https://fhi-aims-club.gitlab.io/tutorials/basics-of-running-fhi-aims/references/

When making use of / reference to scalability, please refer to and cite

and also to the large-scale eigenvalue solver ELPA:

Any application making use of functionality beyond LDA, GGA, or mGGA – i.e., Hartree-Fock, hybrid functionals, MP2, RPA, $GW$, etc. – should please refer to and cite

DFT calculations in FHI-aims employ density functionals from the Libxc library

S. Lehtola, C. Steigemann, M. J. T. Oliveira, and M. A. L. Marques, Recent developments in LIBXC – a comprehensive library of functionals for density functional theory, SoftwareX 7, 1 (2018).

which should be cited in any DFT paper.

Finally, we’re quite proud that FHI-aims performed extremely well in the precision benchmark of 15 leading electronic structure codes known as the “Delta Project”, https://molmod.ugent.be/deltacodesdft – see Reference [195] in Science Magazine for details. Numerical reliability – high precision – in everyday applications, applicable up to very large production problems – continues to be a top priority and is, in fact, one of the key reasons why FHI-aims was written in the first place.

In the present documentation, we do not repeat the basic physical algorithms; rather, the focus is on the actual use of the methods in FHI-aims for a given task, including a full description of all input and output possibilities.

The rest of this document is organized as follows:

• •

  In Chapter 1, a “quickstart” description attempts to give you all the necessary (but not more) information to get FHI-aims up and running on your own computer system, up to the first test run.

• •

  Chapter 2 explains the basic input files and input philosophy very briefly. Some important remarks on choosing the numerical accuracy are summarized here.

• •

  Chapter 3 gets into the gory details, summarizing all available input keywords and their meaning, sorted roughly by their expected use.

• •

  A large chapter 4 is dedicated to some frequently required “meta-tasks” of electronic structure theory: Not just setting up a specific set of input files for a given run, but actually extracting some of the frequently required information from those runs. For the more complex tasks (e.g., a transition state search), we attempt to provide scripts that perform a series of well-defined runs automatically, the use of an external visualization tool, etc.

• •

  In chapter 5 we provide a description of the aitranss (ab initio transport simulations) package which is a project under continuous development at the Institute of Nanotechnology of the Karlsruhe Institute of Technology (KIT), Germany, since 2002. When combined with FHI-aims, aitranss provides a post-processor module that enables calculation of the electron transport characteristics of molecular junctions based on a Landauer formalism in a Green’s function formulation.

• •

  In the appendices, we suggest further reading, more on building the code from source, and we also address some issues (“troubleshooting”) that are either beyond our control (operating-system related issues come to mind), or simply require some level of experience to address.

Electronic structure theory (and FHI-aims) is extremely versatile but many of the most interesting applications require complex workflows. We cannot possibly document them all on our own. Please consider sending us hands-on descriptions of any complex workflows that worked for you, and we would gladly include them in this manual (obviously, we’ll happily include references to your work).

In any case, we hope that this manual will be helpful for your specific purposes. We welcome feedback, in particular regarding issues from production settings that we might not yet have thought of / experienced ourselves. In any event: Happy computing with FHI-aims!