Dear mgeo,

I am unsure if this helps, but have you checked if

../configure.sh -mpi -hdf5 -elpa TRUE

works?

I hope this helps.
Best,
Thomas

Dear SREERAMPK,

Thank you very much for reading our paper and for your inquiry regarding the use of FLEUR for phonon calculations.

Indeed, FLEUR is capable of calculating phonon dispersion both through Finite Displacement (using the phonopy tool) and perturbation theory (DFPT). A guide for calculating phonon dispersion is available in our Git repository under the juphon folder, which can be accessed here:
https://iffgit.fz-juelich.de/fleur/fleur...?ref_type=heads

You can also read the README.md file in the juphon folder of FLEUR on your machine.


To perform a DFPT calculation, it is necessary to link the FLEUR executable to the LibXC library. This can be done during configuration by using the flag -libxc TRUE. The LibXC library is required for DFPT calculations as it provides the necessary exchange-correlation kernels.

Additionally, there are specific tags that need to be modified compared to a standard ground state calculation. These adjustments are outlined in the guide within our Git repository. Furthermore, parameters such as the k-mesh, Kmax, and Gmax may require fine-tuning depending on the system being studied.

Phonon calculations are a relatively recent feature in FLEUR, and as such, you may encounter bugs. We have successfully tested phonon dispersion for systems consisting of up to 4 atoms (including metals, semiconductors, insulators, and magnetic systems). However, certain systems may not be accurately described. For instance, the LO-TO splitting for polar materials at the Gamma point is not yet implemented. Additionally, systems with many local orbitals might encounter issues. Currently, NoCo-phonons, spin-orbit coupling, and Hubbard-U are not implemented. Phonon calculations can only be performed using the LDA approximation at this time.

It is also important to note that DFPT calculations are computationally more expensive than ground state calculations. For each q-point, 3 × n_atoms self-consistent calculations are required. Therefore, we recommend keeping the system size relatively small.

As always, we encourage you to critically evaluate your DFT results. Should any issues arise, please do not hesitate to reach out, and I will do my best to assist you.

Best regards,
Thomas