Useing vasp4.6 and 5.2 results in different energies

Message

dyc_2008 · #1 Post by **dyc_2008** » Sun Sep 02, 2012 5:15 am

dear friends

for obtaining the energy of a same supercell, two versions of vasp, vasp4.6 and vasp5.2, have been employed with the same functionals and other parameters, however, the result shows that there exists an energy difference of about 60 meV.

i want to consult that if such a discrepancy is reasonable or some errors occur?

how can i get the precise energy of this structure?

is this later vasp5.2 more credible?

thank you so much

boris · #2 Post by **boris** » Mon Sep 03, 2012 10:43 am

Hi

Are you using the latest 4.6.32 version?

If you are using a 4.6 version prior to 4.6.32, it is probably because of the compensation charge in PAW. In 4.6, it is truncated up to L=4 wheras in 5.2 it is truncated up to L=6. This affects total energies, especially for d- and f-electron systems.

If you are using the 4.6.32 version, then I don't know

dyc_2008 · #3 Post by **dyc_2008** » Mon Sep 03, 2012 3:57 pm

4.6.26 version

boris · #4 Post by **boris** » Wed Sep 05, 2012 11:16 am

So yes this is probably the reason.

I had the same difference in total energies, starting from the exact same input file.

It's no big deal, it's just a shift in the total energy. If you use the 4.6.32 version you'll have the same results as with the 5.2