How much time need for phonon calculation in quantum espresso for a crystal with 80 atoms?
it is not exactly clear. actually it depend on other parameters, may be too long or not.
Dear Esmaeil Rahmatpour ,
It depends on a few things.
- tr2_ph - This is the convergence threshold. The smaller it is, the longer the calculation will take.
- alpha_mix - This controls the mixing of the potential, i.e. vnew(in) = alpha_mix*vold(out) + (1-alpha_mix)*vold(in). The smaller it is, the more iterations will be needed to reach tr2_ph and, therefore, the longer the calculation will take.
- The number of cores - If you are using one core, then I expect this calculation to take several months. If you are using 100 cores, then I expect this calculations to take on the order of days. These are very rough estimates but should give you some idea.
If you can get one iteration to complete, then you can more accurately estimate the time needed by multiplying the amount of time it took for that iteration to complete by the total number of iterations (number of atoms * 3).
I hope this was helpful. Please let me know if you have any further questions.
Best,
Rob
Will this really take on the order of weeks?! I am running something similar and don't have access to 100 cores at the moment. Are there ways to parallelize this so that it goes faster. Why does this take so long when the scf was so quick(relatively)? Are there other ways to get phonon calculations done that are quicker
The amount of time required for a phonon calculation in Quantum Espresso (QE) for a crystal with 80 atoms will depend on a number of factors, such as the specific hardware being used, the convergence criteria for the calculation, and the level of accuracy required.
Phonon calculations are computationally expensive, as they require diagonalization of the dynamical matrix, which scales as N^3 with the number of atoms N in the crystal. Therefore, larger systems will generally require more computational resources and time.
In general, phonon calculations for a crystal with 80 atoms can take anywhere from a few hours to several days or more, depending on the factors mentioned above. It is important to carefully optimize the convergence criteria, such as the k-point sampling and energy cutoff, and use efficient parallelization methods to minimize the computational time.
It may be helpful to perform benchmark calculations on smaller systems or to use simplified models to estimate the expected computational time for the full system. Additionally, the specific hardware and resources available for the calculation will play a significant role in determining the actual time required.
- Badges
- Science topic