[QE-users] memory problem with phonon calculations

[Hongsheng Liu]

Dear QE users,
    I recently performed some phonon calculations with QE6.1 for water
adsorption on metal surface. I select only water molecules to be used in
the linear response calculation with the keyword nat_todo. For a 6*6
supercell (220 atoms) the phonon calculation can run but for the 8*8
supercell (385 atoms) the phonon calculation can not run. It seems there is
problem with the memory. But the geometry relaxation can run with 8*8
supercell. So my question is:
1. Why the phonon calculation needs so much memory? Do the fixed atoms also
participate in the diagonalization of the dynamical matrix? Is there a way
to avoid the memory problem?
2. What is the mechanism for the phonon calcuation? Is the 'linear response
calculation' a analytic method? Is there any numeric method to calculate
the phonon?
Sorry, I have no idea about the phonon calculation. Thank you very much.

All the best,
Hongsheng
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20181204/1ecf5342/attachment.html>