Dear colleagues,
I am currently investigating the dynamical stability of SrIrH₈ under 100 GPa using first‐principles phonon calculations. The crystal structure was fully relaxed with very stringent convergence criteria (ENCUT = 800 eV, EDIFF = 1×10⁻⁸ eV, KSPACING = 0.04 Å⁻¹, ISIF = 3, NSW = 600). Despite this, small imaginary frequencies persist in the phonon dispersion obtained from even a 3 × 3 × 3 supercell.
I would appreciate any advice on best practices to eliminate numerical artifacts in high‐pressure phonon calculations.