Imaginary frequency is reduced but not disappeared after the step of removing the imaginary frequency

[jgSi1113]

Dear all,

I want to check the anharmonic effect and the potential quantum atomic fluctuations of my system, where the harmonic phonon shows some imaginary modes at some non-gamma points. When I finished the harmonic dynamic matrix calculations and removed the the imaginary frequency by dyn.Symmetrize() & dyn.ForcePositiveDefinite(), I found the imaginary frequency is reduced, but still existed in a very small zone. How to deal with it? Does it mean the system is unstable?

Best & Regards!
Jianguo

[mesonepigreco]

Hi, by using dyn.ForcePositiveDefinite() you only enforce the q-points in the q-mesh to be positive, and this is enough for the SSCHA (e.g., the R point is commensurate with an even grid, so it becomes positive, but not necessary what is close to R does).

You can perform the SSCHA calculation to see what happens to the final converged result. If it still has imaginary phonons when interpolated on this path, then it is probably a signature that the q-mesh (or supercell) employed is insufficient to converge this system.

[jgSi1113]

Thanks for your reply. I will start the SSCHA calculations. Hope I can obtain some good results.

Best & Regards!
Jianguo