ABINIT Discussion Forums

Dear Abinit users and developer,
I substituted some hydrogen atoms of a relaxed large polycyclic aromatic molecule (54 carbon atoms and 18 hydrogen)molecule with oxygen groups (i.e. hydroxyl -OH and carboxyl -COOH ) then try to relax the system again with ionomov=2, when i encountered convergence problem. The force only convergence for the first Broyd/MD run and the following Broyd/MD ran up to several hundred SCF steps without reaching convergence. I have tried LDA (TM type psp for LDA), GGA(TM-type psp .fhi), increasing both supercell size and ecut, however, nothing works.

Could anybody please shed some light on how to solve my convergence problem?


I am running abinit-7.4.3_i686_cygwin_gnu4.7 windows 64bit version. Could this be a problem of abinit configuration.

Regards,
Johnny

Sometimes, when you start far from equilibrium position, the first SCF cycles do not converge or very badly. So you can impose a smaller nstep and go to the next broyden.
You can help a litlle bit the convergence playing with input variables nline/nnsclo
Jordan

Jordan wrote:Sometimes, when you start far from equilibrium position, the first SCF cycles do not converge or very badly. So you can impose a smaller nstep and go to the next broyden.
You can help a litlle bit the convergence playing with input variables nline/nnsclo
Jordan

Dear Jordan
Many thanks for your precious reply. In fact my problem is the first SCF cycles converge (first broyden), but the next broyden does not converge (see my log file). The ETOT seems oscilated around a value but can not converge. I will try your suggestion. In the meantime, i wonder based on my clarification whether you would have some additional hints for me. Thanks again.
Johnny