I am trying to duplicate the results of this study using VASP. Where they have observed that the structure of $\ce{Cs_2SnI_6}$ switches from Fm-3m to I2/m upon application of pressure more than 3.3 GPa. They have also validated there experimental results using DFT but for that they have calculated the energy of various possible space groups and found that I2/m has lesser energy as compared to Fm-3m above 3.3 GPa. I want to do this using Geometry relaxation directly.

Directly running geometry relaxation with Pressure does not work because VASP maintains the symmetry. I also tried running with ISYM=0 that did not help either.

Another idea is to distort the structure manually as has been done here but this switches the symmetry from Fm-3m to let's say some X symmetry.Now this X symmetry is maintained even through the geometry relaxation.

What do you think is the best method to solve this issue ??

  • $\begingroup$ Let me ask another question and think about that, If I model iron as hcp, will it flip to bcc during structural relaxation? If not why put a thought first. I can argue based on your question that it should but it never do so. $\endgroup$ Oct 28, 2022 at 11:10
  • $\begingroup$ @Pranavkumar So basically I know that would happen because of the optimization being stuck at global minima. So in short basically what I am trying to achieve is not possible. The only way to predict the structure is to create a set of initial geometries(in the ideal case this will be all the possible configurations and will posses the true symmetry) and then see which one posses the least amount of energy. That should be the physically realizable structure, right ?? $\endgroup$
    – Chan
    Oct 28, 2022 at 15:00
  • $\begingroup$ overall you got the point! $\endgroup$ Oct 29, 2022 at 3:21
  • $\begingroup$ @Pranavkumar thanks $\endgroup$
    – Chan
    Oct 29, 2022 at 6:55
  • 1
    $\begingroup$ Chan, perhaps you or @Pranavkumar can write an answer now that you feel the problem is solved. We don't need to close the question if it gets answered :) $\endgroup$ Oct 29, 2022 at 13:29

1 Answer 1


Potential energy of a system depends on relative positions of different atoms which is basically a n-dimesional energy landscape. Imposing certain symmetry on the crystal system represents same symmetry in energy landscape as well. For example, It is enough to use 1D energy-volume relationship for cubic system to find minimum energy configuration while for hexagonal system, two dimensional surface is required. From n-dimesnional hypersurface which is unknown to us and computational algorithm as well, we choose certain starting point on that hypersurface. Assuming a convex smooth hypersurface, minimization algorithm tries to find energy minima local to the starting point. Once it reaches local minima algorithm stops, now question is whether that local minima is a global minima or not, it is difficult to say, until other configurations are known.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .