How to compute the system with AM05?

Message

VASP001 · #1 Post by **VASP001** » Sat Jan 09, 2010 1:53 pm

It's just necessary to set GGA=AM ? Or there is need other setting for it ?

waiting for the reply!!

panda · #2 Post by **panda** » Mon Jan 25, 2010 12:16 am

For AM05, yes set GGA=AM
http://cms.mpi.univie.ac.at/vasp/vasp/node133.html

#3 Post by **admin** » Fri Feb 05, 2010 3:04 pm

you need vasp.5.2 for this XC type

nitin · #4 Post by **nitin** » Fri May 28, 2010 7:05 pm

Hi

I'm trying to relax a system with AM05 and rPBE using VASP 5.2. rPBE seems to work fine. But in AM05 i get huge number for XCENC and EBANDS as:

-V(xc)+E(xc) XCENC = 12356.96671364
eigenvalues EBANDS = 5270856.46338422

And after 5 or 6 steps these numbers become so huge that following prints in the OUTCAR:

-V(xc)+E(xc) XCENC = ******************
eigenvalues EBANDS = ******************

I've used following keywords in the INCAR file:

PREC=Normal
ALGO=FAST
GGA = AM
NSW=0
LREAL=.TRUE.
LCHARG=.FALSE.
LWAVE=.FALSE.
LVTOT=.FALSE.

Following error message is also printed:

-----------------------------------------------------------------------------
| |
| ADVICE TO THIS USER RUNNING 'VASP/VAMP' (HEAR YOUR MASTER'S VOICE ...): |
| |
| You enforced a specific xc-type in the INCAR file, |
| a different type was found on the POTCAR file |
| I HOPE YOU KNOW, WHAT YOU ARE DOING |
| |
-----------------------------------------------------------------------------

I use PAW_PBE functional provided in the package.

Any help is really appreciated.

Thank you.

forsdan · #5 Post by **forsdan** » Sat May 29, 2010 12:40 pm

First, please understand that the message in the OUTCAR file is NOT an error. The message is written to remind you that the xc-functional has been changed from its default value in the POTCAR file.

Secondly, regarding your AM05 calculation and the high energies. Try to see if one of the following procedure works:

1. Increase the numerical accuracy. (Increase ENCUT quite significantly, set PREC = Accurate, set ADDGRID = .TRUE. , set LASPH = .TRUE.)

2. First perform a static run with PBE, LDA, or rPBE. Then restart the calculation with GGA = AM and use the converged WAVECAR from the previous run as input for the initialization. I also suggest to increase the accuracy for the runs (see point 1).

3. For the GGA = AM run, use ALGO = All or ALGO = Damped with an appropriate TIME (you will have to test what works best for your systems). See the manual http://cms.mpi.univie.ac.at/vasp/vasp/I ... G_tag.html . Starting from a WAVECAR from a static PBE or LDA run, can be of relevance.

The experience I have for my systems are that procedure 1) can work in rare cases, but is quite unpredictable. Procedure 2) works in the majority of the cases while procedure 3) takes care of the remaining systems.

Best regards,
/Dan

<span class='smallblacktext'>[ Edited Sat May 29 2010, 03:55PM ]</span>

nitin · #6 Post by **nitin** » Wed Jun 23, 2010 5:56 pm

Hi forsdan,

Thanks for your suggestions.

I tried using your 3 suggestions. Suggestion 1 & 2 did not work. Both gave similar errors as mentioned in my previous post. Following suggestion 3, the calculation keep crashing with this message: "LAPACK: Routine ZPOTRF failed! 767".

best,
nitin

nitin · #7 Post by **nitin** » Wed Jun 23, 2010 6:02 pm

Also, i saw some where in this Forum that for doing AM05 calculation using PAW_PBE pseudopotentials i have to modify the parameter "LEXCH = PE" to "LEXCH = AM". is this right?

thanks

forsdan · #8 Post by **forsdan** » Thu Jun 24, 2010 11:27 am

To run AM05 calculations, you only need to set GGA = AM in the INCAR file. Do NOT change LEXCH the POTCAR file.

The ZPOTRF error usually stems from that you end up with an unreasonable density. In connection with ALGO = All or Damped I have encountered it mainly when I use the default mixing parameters. If you haven't already done so I would suggest to try linear mixing parameters as well. The error also arise in occasions when I used an inappropriate TIME setting. So you have to try out different settings (both TIME and Damped/All). For insulators LDIAG = .FALSE. can also be an option. Please read http://cms.mpi.univie.ac.at/vasp/vasp/I ... G_tag.html carefully and follow the recommendations. As I said in my previous post you will have find out what works for your systems.

I would also recommend to you to preconverge a WAVECAR file with either LDA, PBE, or rPBE with your standard settings. Then perform a continuation run with AM05 with your Damped/All and TIME settings.
Increasing the numerical accuracy (Increased ENCUT, PREC = Accurate, ADDGRID = .TRUE. , set LASPH = .TRUE.) is also something I advise you to do if you haven't done that already. To set NEMLDL to a higher value than default, say -8 or -12, can also be of importance.

Cheers,
/Dan

<span class='smallblacktext'>[ Edited Thu Jun 24 2010, 01:39PM ]</span>