the energy dangle

[linglin]

I wanna calculate a strong correlated system. So I first do a structure optimization. But the energy always cannot convergence, just dangle. Is something wrong with LDA+U? Or maybe I should set a special parameter?My INCAR is like
NELM=200
ISTART=0;ICHARG=2
ISMEAR=0
IBRION=2
NBANDS=60
ISIF=3
NSW=30
ALGO=Fast
LREAL=Auto
PREC=Accurate

[forsdan]

1. If you have convergence issues when the LDA+U is not switched on, which it isn't judging from your INCAR, then it cannot be connected to the LDA+U approach. Electronic convergence rate will depend on the system conditions (such as geometry. mag. structure, xc-potential, algoritm, number of bands, etc.) while ionic convergence e.g. depend on relaxation algoritm and starting geometry.

2. If you want to perform LDA+U, you'll find the relevant tags in the manual:
http://cms.mpi.univie.ac.at/vasp/vasp/node162.html
You can also search the forum for previous threads related to the LDA+U.

Best regards,
/Dan Fors



<span class='smallblacktext'>[ Edited Sun Apr 20 2008, 02:31PM ]</span>

[linglin]

Thank you. Yes, I havn't add U in the INCAR. I mean maybe I should add the U to get a fast convergence in optimization?

[admin]

this entirely depends on your system. There may be several other reasons why a calculation does not converge, related to unreasonable input geometries, the choice of the k-mesh, too large mixing parameters, or magnetism which is not accounted for or with an unreasonable input configuration.