Page 1 of 1

Question related to TS_Extract.sh and DIMCAR

Posted: Mon Oct 04, 2021 1:42 pm
by spbatchu
Hello Dr. Henkelman

I have questions related to a couple of topics; one topic being TS_Extract.sh and the other being the DIMCAR file. The questions are as follows:

1) regarding the output displayed upon using TS_Extract.sh,
i) What does E-E0 mean? What value is chosen as E0?
ii) What exactly do Angle, k previous and k next mean?

2) regarding DIMCAR
i) Whenever I look at my DIMCAR after my dimer calculation is complete, I notice that the F Max displayed in the DIMCAR is different from that in the OUTCAR. The F Max in the OUTCAR matches with the ediffg but the F Max in DIMCAR doesn't. Any specific reason behind this?
ii) What exactly do Torque and Angle mean/correspond to in the DIMCAR file?
iii) In the DIMCAR file, I notice that the values of torque, curvature, and angle are not displayed in the last iteration. Any specific reason behind this?

Your answers will be of great help. Thanks in advance

Re: Question related to TS_Extract.sh and DIMCAR

Posted: Tue Oct 05, 2021 4:45 pm
by graeme
I'm not sure about the TS_Exact.sh script - I don't think that is one of ours. If I were to guess, however, E0 would be the energy of the initial image so that E-E0 would be the relative energy along the path. The angle, would likely be the angle between the previous and next image at each image, and k_prev and k_next would be the spring constants to the previous and next images.

The force displayed in the DIMCAR is the magnitude of the projected force at the center of the dimer. This can be different from what is calculated in the OUTCAR when the image is displaced from the center of the dimer. Also, F Max is the maximum force on any atom, which is used as the convergence criterion for EDIFFG. When using the dimer method, we use the same criterion - so we calculate the max force on any atom (but don't print that in the DIMCAR). When we reach that force, there is one more call to the dimer code in which is does nothing, and prints that final line, and then returns control back to vasp to also decide that the calculation is converged so that it cleans up and quits.