difference between 4.x and 5.x

Message

autieri · #1 Post by **autieri** » Mon Nov 28, 2011 8:06 pm

Dear all,

I used an old version of VASP and I obtained some results for my work. Now, I used the new version and I obtained similar results but not identical. For example, for the same material I have a magnetization of 3.85, when i used the old version but now I have 3.99 Bohr magneton. What are the biggest difference between the two code??

thanks
Carmine

boris · #2 Post by **boris** » Thu Dec 08, 2011 1:09 am

Hi

Do you have f elements in your calculation?

A lot of stuff changed between the 4.x and 5.x, and as usual the changes remain a mystery, nothing is included in the manual etc.

One of the changes that I know of is the compensation charge in the PAW approach. The max value is set to 4 in vasp 4 whereas in vasp 5 it can go up to 6.

This has a significant influence on f element system. The rest should remain unchanged.

autieri · #3 Post by **autieri** » Fri Dec 09, 2011 11:30 am

Hi,
I am doing LSDA+U with d-electrons.
Are you speaking about the flag LMAXMIX??

thanks again
Carmine

boris · #4 Post by **boris** » Mon Dec 12, 2011 9:32 pm

No it's not related to the LMAXMIX tag.

Someone on the forum had the same problem with f elements, with a difference in the magnetic moment. We found two reasons:

1. The Dudarev approach is bugged in 4.6, but I don't know exactly where
2. The compensation charge in the PAW is truncated up to L=4 in 4.6 instead of L=6 in 5.2.

So in the end, 5.2 is more accurate, you should trust the 5.2 values.

B.
<span class='smallblacktext'>[ Edited Mon Dec 12 2011, 09:33PM ]</span>

seaso · #5 Post by **seaso** » Fri Feb 24, 2012 4:07 am

the vasp's manual is just so bad