[SOLVED] Pressure difference in optimization and band struc

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Locked
ilukacevic
Posts: 271
Joined: Sat Jan 16, 2010 12:05 pm
Location: Dept. of Physics, University J. J. Strossmayer, Osijek, Croatia
Contact:

[SOLVED] Pressure difference in optimization and band struc

Post by ilukacevic » Wed Feb 24, 2010 11:51 am

Dear colleagues,

I'm trying to calculate the band structure at some specific pressure of VN.

First I optimized the structure at that specific pressure and then copied the relaxed structure in the band structure calculation. However, the pressure abinit gives in the end of the 1st dataset of the band structure calc. (3.3 GPa) is different from the specified one (10 GPa).

I'm using the same parameters in both calculations as can be seen from the input files I attached. Could someone help me and explain why this happens?

Thank you all in advance!

Yours,
Igor Lukacevic
Attachments
t11.in
This is a band structure input file.
(1.43 KiB) Downloaded 657 times
t11.in
This is a structural optimization input file.
(1.72 KiB) Downloaded 663 times
Last edited by ilukacevic on Wed May 23, 2012 8:57 am, edited 1 time in total.

mverstra
Posts: 655
Joined: Wed Aug 19, 2009 12:01 pm

Re: Pressure difference in optimization and band structure calcu

Post by mverstra » Wed Feb 24, 2010 1:44 pm

Need output too. Your input looks ok, but it is suspicious that it is just 1/3 of the desired pressure...

check how well converged it is, of course...
Matthieu Verstraete
University of Liege, Belgium

ilukacevic
Posts: 271
Joined: Sat Jan 16, 2010 12:05 pm
Location: Dept. of Physics, University J. J. Strossmayer, Osijek, Croatia
Contact:

Re: Pressure difference in optimization and band structure calcu

Post by ilukacevic » Thu Feb 25, 2010 4:51 pm

Dear prof. Verstraete,

I'm sending the output files. I hope that they will reveal something.

I've checked the convergence of the pressure while I was checking the convergence of the relaxation at the zero pressure. Used parameters converged the pressure very close to 0 GPa (10^(-7) GPa). When I was optimizing the lattice parameter at different pressures, abinit gave out the exact values of 10, 20,... GPa. I took as a sign of good convergence.

Hope this will help You to help me.

Yours,
Igor Lukacevic
Attachments
t11.out
This is a relaxation output at 10 GPa.
(52.25 KiB) Downloaded 484 times
t11.out
This is a band structure output at 10 GPa.
(108.75 KiB) Downloaded 441 times

mverstra
Posts: 655
Joined: Wed Aug 19, 2009 12:01 pm

Re: Pressure difference in optimization and band structure calcu

Post by mverstra » Fri Feb 26, 2010 12:46 am

You should keep the same ecutsm after the relaxation: it affects the PW set which is used, and determines the convergence to a certain extent (effectively increases ecut for the relaxation run). Same thing for dilatmx, which is not echoed in your outputs... In this case it changes the fft grid quite a bit (your dilatmx is quite high):

ngfft 32 32 32
ngfft 40 40 40

In practice, this means your relaxation was done with a higher real ecut. The pressure is very sensitive to the ecut and ecutsm/dilatmx, but you can see that the etotal is also very different between the two files.

Matthieu
Matthieu Verstraete
University of Liege, Belgium

ilukacevic
Posts: 271
Joined: Sat Jan 16, 2010 12:05 pm
Location: Dept. of Physics, University J. J. Strossmayer, Osijek, Croatia
Contact:

Re: Pressure difference in optimization and band structure calcu

Post by ilukacevic » Fri Feb 26, 2010 11:03 am

Oh, yes! Thank You a lot for reminding me about the consistency.

I'll be back with the results.

Yours,
Igor Lukacevic

ilukacevic
Posts: 271
Joined: Sat Jan 16, 2010 12:05 pm
Location: Dept. of Physics, University J. J. Strossmayer, Osijek, Croatia
Contact:

Re: Pressure difference in optimization and band structure calcu

Post by ilukacevic » Fri Feb 26, 2010 10:25 pm

Dear prof. Verstraete,

Your corrections proved to be essential. After turning on ecutsm and dilatmx, pressures in the band structure calc. were the same as in the relaxation calc.

Thank You again!

Yours,
Igor Lukacevic

Locked