Dear Siesta users, I just want to share an idea that might be useful if you use Siesta to calculate phonon energies/modes.
Doing calculations of phonon energies and electron-phonon couplings I noticed that the numerical problems with using small mesh cut-offs seemed to be localized to the force on the atom being displaced. This is also consistent with the "Egg-box" picture. Using the fact that the forces on all atoms should add up to zero, I replaced the force on the atom (being displaced) by the negative of the sum of the other forces. This seems to give me good phonon energies without using prohibitively large mesh-cutoffs. I use this method regularly to calculate phonon energies. However, my interest is limited to using the phonon mode / electron phonon couplings for transport calculations and have thus not done extensive tests. Hope someone finds this idea useful - Magnus Paulsson ps. The "sum of forces=0" rule is the same as translational invariance -> three zero frequency phonons. If anyone is interested I also have some notes on how to regain rotational invariance. ds.
