Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  • To get parallelization over bands, we can at the moment only use the rmm-diis eigensolver (cg and davidson don't work).
  • The number of bands must be divisible by the number of CPUs, according to Jens Jurgen.
  • At the moment there is no dipole correction in planewave mode.
  • Density mixing is still done in real-space

    GPAW Geometry Optimizations

With GPAW one can do geometry optimizations a factor of 10 faster in LCAO mode (with smaller memory requirements). Then it's necessary to "tweak" the optimization with a little bit of running in FD mode.

Plus, LCAO mode has the added feature that convergence is typically easier, according to Heine.

I think it's difficult to automate the above process in one script, since the number of cores required for LCAO is typically lower than FD (because of the lower memory usage).

But if you're limited by CPU time when doing GPAW optimizations it might be worth keeping the above in mind.

GPAW Memory Estimation

The get a guess for the right number of nodes to run on for GPAW, run the
following line interactively:

...