FHI-aims 221103_1 Release – CHANGELOG

FHI-aims, version 221103_1

How to understand this CHANGELOG (and FHI-aims):

Here, we list a rough overview of recent changes in FHI-aims. A complete record of is available directly through the FHI-aims GitLab. Every user is welcome and encouraged to gain access to the FHI-aims GitLab. The GitLab repository is also the location where bug fixes, improvements etc. will be made available.

A final general note: Please have a look at the growing Wiki in the FHI-aims GitLab. You’ll find helpful information about compiling and running FHI-aims: https://aims-git.rz-berlin.mpg.de/aims/FHIaims/-/wikis/home

Changes since release version 221103

• Fix for SOC eigenstate_density cube output.
• Fix for the range-separated hybrids (lc_wpbeh) when alpha is set to 0.

Changes since release version 210716_3

Exchange-hole dipole moment (XDM) dispersion model

Alberto Otero de la Roza

• current implementation of the XDM dispersion model in FHI-aims.
• the xdm keyword with a few options is made available to activate the calculation of the XDM dispersion energy and derivatives.
• The B86b exchange functional is also implemented. The shorthand xc keywords are:
  • b86bpbe for the GGA (B86b exchange with PBE correlation)
  • b86bpbe-25 and b86bpbe-50 for 25% and 50% exact exchange hybrid functionals, respectively
• reference: https://doi.org/10.1063/1.4705760

Nuclear-electron orbital (NEO) method for isolated and periodical systems

Jianhang Xu

• NEO method enables modeling selected nuclei quantum mechanically at the same level of theory as the electrons, treating nuclear–electronic correlation in the framework of multicomponent DFT
• it is framework for quantizing protons in electronic structure calculations beyond the Born–Oppenheimer ansatz
• details and example input can be found in src/neo/README.md and src/neo/neo.in_example
• reference: https://doi.org/10.1063/5.0088427

Further improvements hybrid density functionals

Florian Merz

• Compression of the RI-coefficients (similar to the Coulomb matrix). This results in a new default calculate_fock_matrix_version 5.
• Refactoring and optimization of hybrid density functional band structure calculations
• Several bug fixes for large (large in terms of basis set and/or number of atoms) systems.
• Note: The correct behavior for range-separated hybrids including long- and short-range EXX contribution (Pulay mixing of short-range and long-range) is only working correctly for calculate_fock_matrix_version 4 or 5. For more details see below in “Handling of range-separated hybrids through Libxc”.

CC-aims: An interface to the Couple cluster for solids (Cc4s) code

Evgeny Moerman

• CC-aims is an interface to the coupled cluster code Cc4s for molecules and solids
• Cc4s is able to run e.g. MP2, CCSD. More details can be found here: https://cc4s.github.io/user-manual/
• FHI-aims and Cc4s also implement a prototype workflow for the computation of coupled-cluster band structures via the equation of motion coupled-cluster method
• documentation (including tutorial) for CC-aims: https://moerman1.gitlab.io/fhi-cc4s/

RPA Forces

Muhammad Tahir

• formalism which enables calculating the analytical gradients of particle-hole random-phase approximation (RPA) for molecules
• relax molecular structures at the RPA level using both Gaussian-type orbitals and numerical atomic orbitals
• reference: https://doi.org/10.48550/arXiv.2109.00742

COSMO (COnductor-like Screening MOdel) solvation method

Yi Yao

• industry standard solvation model (cf. https://doi.org/10.1039/P29930000799)
• code works in both serial and parallel and can compute the COSMO forces

HIP support for FHI-aims (GPU)

Werner Lipsunen, Ville Havu

• What is HIP: a C++ Runtime API and Kernel Language that allows developers to create portable applications for AMD and NVIDIA GPUs from single source code.
• The HIP code was directly ported from the original CUDA code. There are still some issues with the HIP code and this is an experimental release.

Atomic Simulation Interface (ASI)

Pavel Stishenko

• includes functions for export and import of data structures that are used in electronic structure calculations and for classical molecular dynamics simulations
• a uniform, generic and efficient interface for connecting various computational chemistry and materials science codes in multiscale simulation workflows, such as QM/MM, QM/ML, QM/QM
• Specification of functions, data types and calling conventions for export and import of density matrices, overlap and Hamiltonian matrices, electrostatic potential, atomic coordinates, charges, total energy and forces.
• documentation: https://pvst.gitlab.io/asi/
• examples can be found in the regression tests: regression_tests/testcases/ASI

RI-density restart and interface to ML framework SALTED

Alan Lewis

• the keyword ri_density_restart (and associated keywords) allows users to generate the coefficients which expand the electron density into the auxiliary basis functions used in AIMS.
• calculations can be restarted (also on different k-grids) by reading in these coefficients (or coefficients generated by SALTED)
• the overlap matrix of auxiliary functions and projections of the density onto the auxiliary functions can also be output for use in training a SALTED model. There is a worked example in the public SALTED repository (https://github.com/andreagrisafi/SALTED/tree/legacy) which explains the workflow of using AIMS + SALTED.
• reference: https://doi.org/10.1021/acs.jctc.1c00576

Range-separated hybrids through Libxc and benchmarking of the wb97 xc functional series

Yi Yao, Florian Merz

• Fixes a bug that caused poor convergence for range-separated hybrids containing both long- and short-range exact exchange contributions. Also xc energy contributions are now correctly calculated. This bug has likely effected most range-separated hybrids accessible through Libxc.
• As test case, the wb97 xc functional series as implemented in Libxc has been benchmarked. Key ingredient for this series of functionals is the range-separated exact exchange contribution.

Performance Improvements periodic GW (k-space algorithm)

Yi Yao

• GPU acceleration for the serial and parallel linear algebras via linear algebra accelerator packages of cublas, COSMA, and SLATE.

Further developments

• contour deformation with analytic continuation of W (CD-WAC) (Ramon Lorenzo)
• loc-NAO-VCC basis sets for correlated PBC calculations added to species_defaults
• optimization and bug fixing of band Mulliken analysis (Sebastian Kokott)
• Extended k-grid for DOS calculation with tetrahedron method (Hagen-Henrik Kowalski)
• update of LibXC to version 5.1.7 (Yi Yao)
• update of ELSI to version 2.9.0 (Sebastian Kokott)