BRMIX: very serious problems the old and the new charge density differ old charge density

Message

cajuan@criba.edu.ar · #1 Post by **cajuan@criba.edu.ar** » Thu Sep 02, 2010 8:36 pm

Running one Fe (110) surface we get this error message in the STDOUT:

BRMIX: very serious problems
the old and the new charge density differ
old charge density: 192.00000 new 186.62159
0.558E+01

reading the outcar no error is printed

somebody could help me?

cpp6f · #2 Post by **cpp6f** » Sun Sep 05, 2010 2:40 am

This is one of the symptoms of the "incorrect number of electrons" problem I just posted about. Are you using the intel compiler? That is the only compiler I have trouble with. Is the number of electrons printed in the outcar for the first scf step even correct? For me it is not.

#3 Post by **admin** » Sun Sep 26, 2010 1:11 pm

please make sure that you did not set NELEC in INCAR.
please also provide some more details of the run that produced the error. At which point of the run did this error show up? Did you choose NBANDS sufficiently high? Maybe it helps to decrease the charge and spin density mixing parameters.

ucapcrw · #4 Post by **ucapcrw** » Tue Nov 02, 2010 4:42 pm

I get the same error when running a graphite intercalate calcuation (KC14).

The system is 14 C atoms and 1 K atom in a hexagonal unit cell. SYMPREC=1E-06, a plane wave cut off of 500eV and EDIFFG=-1E-07.

For my KPOINTS I use:

Code: Select all

Automatic
0
Gamma
7.Â Â 7.Â Â 7.
0.Â Â 0.Â Â 0.

And get the error in the stdout:

Code: Select all

BRMIX:Â veryÂ seriousÂ problems
Â theÂ oldÂ andÂ theÂ newÂ chargeÂ densityÂ differ
Â oldÂ chargeÂ density:Â Â Â Â 63.00066Â newÂ Â Â 62.99998

And in OUTCAR I get on the very last line:

Code: Select all

-----------------------------------------Â IterationÂ Â Â Â 1(Â Â 34)Â Â ---------------------------------------


Â Â Â Â POTLOK:Â Â VPUÂ timeÂ Â Â Â 0.00:Â CPUÂ timeÂ Â Â Â 0.02
Â Â Â Â SETDIJ:Â Â VPUÂ timeÂ Â Â Â 0.00:Â CPUÂ timeÂ Â Â Â 0.01
ErrorÂ EDDDAV:Â CallÂ toÂ ZHEGVÂ failed.Â ReturncodeÂ =Â 145Â 5Â 2

If I run the same calculation with a KPoint grid of 6x6x6 or 8x8x8 or 5x5x5 it's fine, but not for 7x7x7. Is this anything I should worry about?

Thank you in advance,
Charlie

alex · #5 Post by **alex** » Wed Nov 03, 2010 8:06 am

Hexagonal cell? Try Gamma instead of Monkhorts scheme in KPOINTS. Should work best for all k-point sets.

Cheers,

alex