Hello, everyone.
I started to do calculations of DFT using Quantum Espresso, and I had a few questions about the input files.
I found tutorials in internet like this:
https://www.youtube.com/watch?v=DKHffCHSoAk&t=65s
In which an input file describes a Silicon structure with 2 atoms, and an IBRAV=2, leading to the diamond-like crystal. Check: Si_2_atoms.in
Also, I tried to do calculations of the same structure, but describing just 1 atom of Silicon and specifying the space group F d -3 m, like in : Si_1_atom_Fd-3m.in
These calculations lead to very similar results. However another known way to generate input files for QE is using the tool provided in:
https://www.materialscloud.org/work/tools/qeinputgenerator
This tool "removes" the symmetry of the system, and try to describe all atoms, so the Silicon cell is described with 8 atoms, like in: Si_Materials_Cloud.in
This last input generates a totally different result, and I wanted to confirm which is the correct form of input file that Quantum Espresso accepts.
Best regards,
Ricardo Tadeu
Hello Ricardo Tadeu,
The correct form of input file that Quantum Espresso (QE) accepts depends on the specific calculation you want to perform and the level of accuracy you need.
In general, QE can handle input files that describe a variety of crystal structures, including those with and without symmetry. If you want to include symmetry in your calculations, you can specify the appropriate space group using the SYMGRP card in the input file.
Regarding the different results you obtained with the different input files, it is not surprising to see differences in the calculated properties depending on the input file used, especially when the input files describe different crystal structures or different numbers of atoms.
To determine the correct input file to use for your specific calculation, you should carefully consider the physical system you want to study and select the appropriate crystal structure, space group, and number of atoms. It is also important to consider the convergence criteria for your calculations, such as the k-point mesh, energy cutoff, and other parameters, and ensure that these are sufficiently converged for your desired level of accuracy.
In general, it is recommended to use multiple input files and perform benchmark calculations to ensure that the results are consistent and reliable. You can also compare your results with experimental data or with other theoretical calculations to validate your results.
I hope this information helps you to select the appropriate input file for your calculations in QE.
Hi Ricardo,
xcrysden or ase are able to read QE input files and to produce the unit cell. You can check by eye if the system are the same.
Regards,
Roberto
Hello, Sudhanshu Gupta and Roberto D'Agosta
Thanks for your replies, I wanted to be sure how I could make correct calculations. I used Xcrysden to visualize them, and all of them showed the same structure. So, you can have different inputs resulting in the same structure visualization, although some of them can be wrong.
Thank you both,
Best regards
Ricardo
Dear Ricardo,
If the structures are similar the energy should come out similar. I have the feeling that if you look at the forces you would discover that the structures have different residual forces. Your first structure for example has a lattice constant larger than the other 2.
Indeed, the structures you obtain from databases are relaxed with some pseudo potential this is always a good choice to optimise the atomic positions with the pseudo you plan to use.
The second and third structure have similar energy per atom per unit cell. Force relaxation should reduce the difference to below the chemical accuracy.
I hope this help,
Roberto
Yes, that is correct. Different inputs can lead to the same structure visualization, but it is not guaranteed that all inputs that give the same visualization are correct. It is important to verify the input parameters and ensure they are physically meaningful and consistent with the expected chemical properties of the system being modeled. Additionally, it is essential to validate the results using various techniques, such as comparing with experimental data, validating the methodology and software used, and performing convergence tests to ensure the results are reliable.
- Badges
- Science topic