Page 1 of 1

acell optimization

Posted: Tue Jun 11, 2013 7:55 am
by sevket simsek
Dear abinit users,

I'm trying to calculate acell parameters of DyMnO3 compound. It stopped working after a few steps and asked me to raise as dilatmx= 1.09 .When I set dilatmx=1.09, calculation stopped again.
and give me error as following..

I attached my input file.
How can I solve this problem?

chkdilatmx.F90:115:ERROR
The new primitive vectors rprimd (an evolving quantity)
are too large with respect to the old rprimd and the accompanying dilatmx :
this large change of unit cell parameters is not allowed by the present value of dilatmx.
You need at least dilatmx= 1.121321E+00
Action : increase the input variable dilatmx.
MPI_ERROR_STRING: MPI_ERR_UNKNOWN: unknown error

Re: acell optimization

Posted: Tue Jun 11, 2013 11:10 am
by ilukacevic
Dear Sevket,

it may be that you are starting from far of the equilibrium structural parameter values. Try first to "improve" your inital structural prameters.

Also, why do you relax only the cell size and not the xreds, too? Please, go through the tutorial to see the proposed relaxation procedure.

By the way, in order for getwfk to function as it is intended here, you have to have at least 2 datasets.

Best regards,

Igor

Re: acell optimization

Posted: Tue Jun 11, 2013 2:43 pm
by sevket simsek
Dear Igor,
Thanks you very much your response,
1) All the parameters used in this calculation is experimental parameters. Firstly I calculated ecut, then in order to optimize acell parameters, I used experimental values such as acell, xred, angdeg...

2) I also want to calculate both acell and xred, but calculation takes a long time.

how can I solve this problem?
I still try your suggestion

thanks dear Igor again,

Re: acell optimization

Posted: Wed Jun 12, 2013 8:56 am
by ilukacevic
sevket simsek wrote:1) All the parameters used in this calculation is experimental parameters. Firstly I calculated ecut, then in order to optimize acell parameters, I used experimental values such as acell, xred, angdeg...


Doesn't matter. Theoretical results can differ from the experimental ones, depending on the approximations you use (xc approximation, for example).

sevket simsek wrote:2) I also want to calculate both acell and xred, but calculation takes a long time.


Which part is long: SCF or BFGS? And again, please, follow the tutorials 1-4, in order to see the best way to relax the unit cell.

Best regards,

Igor