UT theoretical chemistry code forum

Hi Everyone,

Following is one of the result from DIMER calculation.
Step Force Torque Energy Curvature Angle
25 0.13218 2.72457 -472.57147 -3.49910 4.12687
25 0.13218 3.40552 -472.57147 -3.55196 1.57487
25 0.13218 1.37301 -472.57147 -3.64305 1.82167
25 0.13218 1.75477 -472.57147 -3.50525 1.01309
26 0.13218 --- -472.57147 --- ---


INCAR FILE Parameters:
EDIFF = 1E-7 ; EDIFFG = -0.03

It said in OUTCAR file : reached required accuracy - stopping structural energy minimisation

But still the force is quite high. Do I continue with this result? If I want to lower the force, what actions do i need to take?

That looks strange, and may indicate a problem with what is being printed. The force should change between step 25 and 26, and yet it doesn't. Can you post the calculation so that I can reproduce the problem and see what's going on? Also, just check the OUTCAR to see what the final force is. My guess is that the calculation has converged to 0.03 eV/Ang.

Actually, I just realized that what is likely happening is that the dimer code is printing |F| whereas the convergence criterion (EDIFFG) is based upon the maximum force on any individual atom. For a system with many atoms, |F| can be significantly larger than the EDIFFG criterion.

Anyway, you can lower EDIFFG to -0.01 or -0.005 to converge more tightly.

Thanks Dr. Henkelman. Attached is the DIMCAR file from my calculation.

Following is the last line from OUTCAR FILE

FORCE total and by dimension 0.089317 0.024019

Could you please help me to understand- which is the force value?
-- last one from DIMCAR file or last value from OUTCAR File (mentioned above)?

There are different ways to measure the magnitude of the force. The magnitude of the entire force vector |F| is reported in the DIMCAR file. The criterion that EDIFFG looks at is the maximum force on any single atoms (the by dimension force). The fact that this by dimension force dropped below 0.03 eV/Ang means that the calculation properly achieved the convergence that you specified.