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

Several small changes/fixes:

- Setting
`packed_matrix_threshold=0d0`

for computation dipole matrix or momentum matrix. - Using the same number of states across all task for ill-conditioned basis
- Fix allocation of some arrays when relaxation is requested for sr-lr hybrid functionals
- Fix hybrid functionals with pure Gaussian functionals
- Fix bug in memory estimation for GW calculations
- Stop hybrid functional calculations when band structure segment is request in a direction where the SCF k-grid has less than 3 k-points

*Connor Box*

- DFPT, DFPT_reduce_memory, DFPT_polarisability, DFPT_dielectric, DFPT_phonon, DFPT_phonon_reduce_memory now use a common DFPT module, currently activated by setting DFPT_centralised .true. in the control.in
- Electronic friction calculations now support DFPT, for both clusters and periodics systems.
- Refactor of electronic friction driver, including improved scaling and output of electron-phonon coupling matrix elements using ELSI.
- DFPT centralised interface is interfaced to LibXC and supports LDA, GGA, and hybrid functionals
- Framework set to switch DFPT_centralised .true. as the default in the new-future.

*Sam Hall and Dylan Morgan*

- Refactored force_occupation_basis and force_occupation_projector keywords, re-written in an extensible format as a platform for future development
- Introduced new deltascf_projector and deltascf_basis keywords
- Significantly improved the scaling performance by using ELSI to calculate the non-Aufbau occupations and chemical potential
- Further improved the scaling by utilising other optimised matrix routines from ScaLAPACK and adding support for KS_method parallel in aperiodic systems
- See https://www.archer2.ac.uk/ecse/reports/eCSE04-03/ for further details

*Yair Litman, Alaa Akkoush*

- Allows ab initio simulations of polarizability tensors including a numerical inhomogeneous local hartree potential originally calculated by TDDFT responses from metallic tip-like structures
- Repository with a how-to example, input files including the local hartree potentials for selected geometries, and a script to create tip-enhanced Raman images with this implementation in FHI-aims can be found in https://github.com/sabia-group/TERS_Tutorial

*Elisabeth Keller*

- minimal+s basis set: cost-efficient, near-minimal basis set with correction for basis set incompleteness error (currently parametrized for PBE).
- Purpose: overcomes runtime limitations and memory issues for fast (pre-)relaxation, MD, large-scale geometry optimizations.
- Where to find:
`species_defaults/defaults_2020/minimal+s/`

- Usage: Copy-paste the minimal+s species files into control.in. The corrections are automatically invoked by using the minimal+s species files.

*Francisco Delesma*

- Green-X is a collection of libraries for low-scaling RPA and GW methods.
- We use minimax grids implemented in GreenX for canonical RPA, RPA gradients, and GW calculations.
- The minimax grids are available for periodic and non-periodic computations.

*Florian Merz, Min-Ye Zhang*

- Reduced time consumption for self-energy by limiting QPE calculation to energy states near Fermi level.
- Reduced memory consumption for reciprocal lvl triples, along with detailed memory report in run and a final report.
- Bug fixes for occasional hanging.

*Pavel Stishenko, Jakob Filser*

- The new MPE solvent model (keyword
`mpe-ncps`

) uses Piecewise Solvent region representation. It increases sparsity of the MPE matrix, improves memory efficiency, and has a potential for calculations speed-up. - The MPE model with Piecewise Solvent can be used for systems with periodic boundary conditions.

*Pavel Stishenko*

- Atomic Simulations Interface (ASI) is a C-style API. It is available
if FHI-aims is build as a library. The Python wrapper
`asi4py`

is available in PIP. - ASI API supports setting atomic coordinates, lattice size, custom electrostatic potential, and density matrix.
- ASI API supports getting total energy, atomic forces, charges, stress, electrostatic potential, and density, overlap and hamiltonian matrices.

*Uthpala Herath*

- Implement static excitations method for studying electron excitations.
- Replace get_occupation_numbers_p0 with ELSI’s elsi_compute_mu_and_occ subroutine for SOC calculations.
- Increase number of empty states for SOC calculations to converge the SOC results.
- Replace subroutine check_norm with ELSI’s check_electrons.

*Muhammad Tahir*

- RPA+rSE forces for molecules
- RPA force for solids, stress is conserved
- Added evaluation of direct particle-hole RPA total energy (plasmon approach)
- Added direct phRPA+ SOSEX total energy (not dependent on frequency integration grid)
- direct-ring-ccsd from direct-particle-hole RPA total energy
- Added evaluation of analytical gradient of long-range coulomb
interaction (needed for RPA force for solids under
`have_Ewald`

tag)

- Added routines for localising molecular orbitals via optimised unitary transformations; includes localised molecular orbital visualisation and restarts (Gabriel Bramley)
- Implementation of GPU-offloading in the RI-V routines on Nvidia GPUs (CUDA) (Moritz Leucke, Roman Kempt, Antonio Delesma, Ramon L. Panades-Barrueta)
- Refinement of the Hybrid functionals auto-tuning parameters: Auto-tuning parameters are chosen more conservative to avoid out-of-memory crashes (Florian Merz)
- Written new slab identification routines (Osama Jalil)
- Adapting default choices for preconditioner: only switch on the preconditioner when a slab s identified. The preconditioner will be switched off for bulk materials by default now. Please report back, if that affects your SCF convergence significantly.
- Fixing rotational invariance issue of the on-site nuclear Hartree potential (Volker Blum, Uthpala Herath)
- Added basis sets for Periodic GW calculations that include for_aux functions (Uthpala Herath)
- Updated the default relativistic setting to
`atomic_zora scalar`

(Konstantin Lion) - Updated CI to use Docker image with Ubuntu20, Intel 2022, MPICH 3.3.2 and GNU10.5 (Uthpala Herath)
- update of LibXC to version 6.1.0 (Osama Jalil)

*Andrey Sobolev*

- New workflows for GW calculations and MD simulations
- Visualize Hirshfeld and Mulliken charges.
- Improved handling of the band structure visualization
- Where to find: gims.ms1p.org

- Embedded-Cluster Modelling With FHI-aims (Oscar van Vuren)
- MDAnalysis and FHI-aims (Oliver Beckstein)
- An overview of all tutorials can be found here