force and energy change after static calculation

[mariesto]

Dear All,

I need explanation about relaxation. I performed relaxation calculation. After the system relaxed, I used its CONCAR file as a POSCAR of new static calculation. I used the same number of Kpoint, value of cutoff energy, NELMIN, EDIFF, ISMEAR, SIGMA.

The forces and energy now (in static calculation) are bigger than those in the last step of relaxation calculation. I wonder the reason of these difference. What should I do then?
below are my two INCAR files and part of OUTCAR files:

1) relaxation

NELMIN = 8 # do a minimum of ten electronic steps
EDIFF = 1E-5 # high accuracy for electronic groundstate
EDIFFG = -0.01 # small tolerance for ions
NSW = 40 # 20 ionic steps should do
MAXMIX = 80 # keep dielectric function between ionic movements
POTIM = 0.2
IBRION = 2 # use RMM-DIIS algorithm for ions
NFREE = 10 # estimated degrees of freedom of the system
ISIF = 4
LGO = Fast # RMM-DIIS algorithm for electrons
LREAL = A # evaluate projection operators in real space
NSIM = 4 # blocked algorithm update, four bands at a time
ISMEAR = 0
SIGMA = 0.05
ENCUT = 550
ISPIN = 2

OUTCAR:
first tthree colums are coordinat, the last three colums are forces of the last step of relaxation.
....
7.37737 0.59840 9.39123 0.004632 0.001697 -0.001366
6.76174 0.58411 10.63033 1.096011 0.076688 1.514850
7.50423 0.55647 11.79905 -0.466400 -0.057010 3.801280
8.88756 0.51892 11.72617 2.883524 -0.099949 3.517584
6.64804 0.72064 8.22155 -0.002042 -0.000691 -0.001793
5.25959 0.88645 8.29620 -0.001294 -0.001056 -0.000907
4.63612 0.72862 9.54721 0.003600 -0.008207 -0.003445
......
FREE ENERGIE OF THE ION-ELECTRON SYSTEM (eV)
---------------------------------------------------
free energy TOTEN = -370.448432 eV

energy without entropy= -370.448431 energy(sigma->0) = -370.448431
-------------------------------------------------------

2) Static

INCAR
startparameter for this Run:
NELMIN = 8 # do a minimum of ten electronic steps
EDIFF = 1E-5 # high accuracy for electronic groundstate
NSW = 0 # 20 ionic steps should do
LGO = Fast # RMM-DIIS algorithm for electrons
LREAL = A # evaluate projection operators in real space
NSIM = 4 # blocked algorithm update, four bands at a time
ISMEAR = 0
SIGMA = 0.05
ENCUT = 550
ISPIN =2

OUTCAR:
....
7.37737 0.59840 9.39123 0.041742 0.021186 0.059220
6.76174 0.58411 10.63033 1.022889 0.083720 1.396480
7.50423 0.55647 11.79905 -0.372900 -0.068669 3.863764
8.88756 0.51892 11.72617 2.824964 -0.102414 3.663122
6.64804 0.72064 8.22155 -0.167796 -0.003430 -0.084571
5.25959 0.88645 8.29620 0.028570 -0.158013 -0.051058
4.63612 0.72862 9.54721 0.015213 0.001480 -0.082504
5.37841 0.62166 10.70320 -0.893832 0.209176 1.485869
------------
FREE ENERGIE OF THE ION-ELECTRON SYSTEM (eV)
---------------------------------------------------
free energy TOTEN = -370.425069 eV

energy without entropy= -370.425069 energy(sigma->0) = -370.425069


sincerely yours,

[admin]

please find a reply to your thread in the bugreports forum

[SKM]

Hi
i have similar problem, i could not locate the answer in Bugreports forum.
I am NOT doing either of DFT+U or Volume relaxation.
i did relax a system of two materials. (say A and B hybrid structure). and then for energy difference i.e. binding energy. I separated the relaxed systems into its two parts after relaxation (i.e. POSCAR-A and POSCAR-B) and then run static calculations.
the combined A.B system took 63 SCF stps
system Part-A, took 100 SCF steps
and System Part-B, is not converged even after 340 SCF steps.

used same INCAR files for all 3 steps with IBRION = -1 NSW=0 and added the NGX/Y/Z and NGXF/XY/XZ explicitly taken from previous OUTCAR file.

for a query on the Binding energy calculation method (the query Heading was " Binding Energy in Hybrid system" from me in "Physics Questions" forum, Martin.schilpf has replied like below:

In general, your approach is correct, but you need to be careful with the specifics. E.g., the systems might converge differently w.r.t. the energy cutoff, so it is not sufficient to just use one value for both. Other considerations are the system size convergence for the combined system, lattice mismatch at the interface, rearrangements (e.g. interdiffusion), ... Keep in mind to converge w.r.t. to the target quantity (binding energy).

As for a way to combine the calculations. I don't see how this would be easily possible. Typically it doesn't matter anyway, because the time to calculate the combined system dominates the one needed for individual systems due to the much larger unit cell.

Though i am not very clear about the reply and how to sort out, but this static calculation non-convergence is another issue.
Can anyone help in suggestion?

Regards

[martin.schlipf]

Here are a few things you can try if your electronic SCF fails:
-increase the number of steps NELM
-switch to a different algorithm ALGO = Normal (if you were using Fast)
-adjust the mixing parameters (see wiki/wiki/index.php/BMIX )

[SKM]

Thank you Martin for quick reply.

Here are a few things you can try if your electronic SCF fails:
-increase the number of steps NELM
-switch to a different algorithm ALGO = Normal (if you were using Fast)
-adjust the mixing parameters (see wiki/wiki/index.php/BMIX )

If i change algorithm for static calculations of the same structure, after the optimisation done (say for a hybrid structure), can we get comparable energies? i.e. Not much variation w.r.t algorithms?

Regards

[martin.schlipf]

If you only change the algorithm, you should get the same energies if your energy threshold is sufficiently tight. If you want to be extra careful you can run 1 step restarting from a converged calculation with the same algorithm.

[SKM]

Thank you i did this step of checking with different algorithms, it did NOT change much, only at 4th or 5th decimal, there is slight variation. So seems fine changing with Algorithms, in this case.

Regards