This archive contains the data from the publication "Probing the Mott-insulating behavior of Ba2MgReO6 with DFT+DMFT" by Maximilian E. Merkel, Aria Mansouri Tehrani, and Claude Ederer, currently published on the arxiv under https://doi.org/10.48550/arXiv.2312.09839 and under review at Physical Review Research.
All our results are stored data.tar.gz. The figures and data in the table are generated with .ipynb notebooks. The number of the respective figures/tables are part of the file name. If data is passed between notebooks, .pkl files are used that are generated by one notebook and read in by another.
Additionally, the archive data_dft_para_rand.tar.gz contains the data we obtained from the authors of doi.org/10.1103/PhysRevMaterials.5.104410. Also here, we added two .ipynb notebooks in there to extract the relevant data into two .pkl files for our Figs. 5 and 7.
This section contains some information on how to reproduce the results contained in this archive. As specified in publication, we used VASP 6.4.1 and postprocessed the data with multipyles 1.1.0.
Most folders include an input folder with the relevant input files and a job script as jobfile.sh or run.sh.
In our calculations, we have mostly used docker with two different images that are not part of this archive because they contain the VASP source code:
vasp:6.4.1 with VASP 6.4.1 and Wannier90 3.1.0 installedtriqs_mpich:unstable_vasp5 with solid_dmft installed as specified below. The VASP version of this image is never used.For solid_dmft, we used commit 4712080 (which corresponds to the branch soc_real at the moment of publication of this Materials Cloud Archive entry). It can be obtained from GitHub but the python files from that commit are also added to this Materials Cloud Archive entry as solid_dmft_soc_real.tar.gz to ensure reproducibility.
We used the libraries TRIQS, DFTTools and CT-HYB from commits based on the respective 3.1.x branch.
The DFT calculations are in the folders containing 0scf, 1bands, 2wannier, 8ed, and 9crpa. nonmag and soc are used to designate if SOC was turned on. Note that also for the soc calculations, the magnetic moment is numerically zero throughout all DFT calculations. In general, calculations sometimes use results from calculations with lower number, for example 9crpa links to files from 8ed and 8ed copies files from 0scf.
For rerunning our calculations, please insert a POTCAR file composed of the following files from VASP into respective input folder:
The DMFT input is generated from wannierization of the DFT results in 2wannier. We first run VASP with wannier90.win to write out all initial projections potentially needed later (Re-d and O-p).
We now manipulate the Wannier90 output using the python scripts crop_amn_file.py and read_projectors_and_write_to_h5.py that are contained in dmft_projectors.tar.gz here on Materials Cloud Archive. This is done in the script postproc.sh that modifies the Wannier90 files and creates a new set of input files (amn, eig, mmn, win) for the frontier-orbital projectors (indicated by a _corr) using the script dmft_projectors/crop_amn_file.py.
We then run Wannier90 for the seeds wannier90 (VASP default, localized orbitals) and wannier90_corr (output from crop_amn_file.py, frontier orbitals) to obtain the respective Wannier orbitals. Finally, we run the converter to obtain an h5 archive in the run_conv.py script, which uses the Wannier90Converter from triqs_dft_tools for different spin-orbit coupling strengths between 0 and 0.6 eV. It also writes the projector matrices for frontier and localized Wannier orbitals to the h5 archive with dmft_projectors/read_projectors_and_write_to_h5.py.
These resulting h5 archives are then the input to our DMFT calculations.
All DMFT calculations are in folders containing 5dmft. They simply use the h5 archives obtained before and the generate_jobfile_scan_standard.sh script that writes out the correct config file for solid_dmft.