Dimer force higher than EDIFFG when job finishes

yeehaw · Post by **yeehaw** » Tue Oct 01, 2013 4:21 pm

Recently a dimer job of mine finished with the following in the DIMCAR:

Step Force Torque Energy Curvature Angle
...
9 0.20467 1.42277 -759.00017 -6.35806 1.43794
9 0.20467 1.31133 -759.00017 -6.40175 1.10612
10 0.18025 1.15579 -759.00084 -6.65171 1.29859
10 0.18025 1.13254 -759.00084 -6.68295 0.87490
10 0.18025 0.80498 -759.00084 -6.69760 0.92665
11 0.15338 0.85427 -759.00134 -6.70237 0.65621
12 0.15338 --- -759.00134 --- ---

However EDIFFG was set to -0.03 in the INCAR, and the maximum force on any of the relaxed atoms in the system did fall below 0.03 eV/Å (per OUTCAR). I am surprised by the large remaining force indicated in DIMCAR. Is this discrepancy normal? Is there a tag I can set in INCAR to directly control the convergence of the force in DIMCAR?

Thanks!