Hi,
When I run a dimer calculation, I firstly get the initial and the final states, and then use the "modemake.pl" to generate the MODECAR file between these two configurations. My question is how to generate the POSCAR in the dimer run. If I kown the transition state approxmately, can I specify the positions of the atoms manually? Or should I use the nebmake.pl to interpolate one image between the inital and final states?
Another question is if a dimer run is stopped because of power cut, can I just copy CONTCAR to POSCAR to continue the calculation?
Thank you very much for your kind attention.
Yi-An
A question on dimer run
Moderator: moderators
Yes you can specify the position of the atoms manually, using the nebmake script, the interpolate script, or by first doing a short NEB calculation for a few iterations and then use the neb2dim script.
Restart the dimer by first running the 'vfin.pl dir' , where dir is the directory into which the run will be archived. The remaining files will be ready for restart. This will copy the contcar to poscar and the newmodecar to modecar, as well as archiving the files from the previous calculation.
Restart the dimer by first running the 'vfin.pl dir' , where dir is the directory into which the run will be archived. The remaining files will be ready for restart. This will copy the contcar to poscar and the newmodecar to modecar, as well as archiving the files from the previous calculation.
Thank you for your reply.
I find the efficiency of my used method (use the modemake.pl to generate the MODECAR and use the nebmake.pl to interpolate the guessed saddle point) is much lower than your mentioned method (first run a neb calculation, and then use the neb2dim.pl to generate the POSCAR and MODECAR). Can you give me any suggestions?
Thanks.
Yi-An
I find the efficiency of my used method (use the modemake.pl to generate the MODECAR and use the nebmake.pl to interpolate the guessed saddle point) is much lower than your mentioned method (first run a neb calculation, and then use the neb2dim.pl to generate the POSCAR and MODECAR). Can you give me any suggestions?
Thanks.
Yi-An
There is a trade-off between the two approaches.
In one, you can generate a guess at the saddle as being 1/2 way between initial and final states, and start a dimer search right away.
If this is a bad assumption, even a crude NEB can give you a better approximation to the saddle with a few additional force calls.
An analysis of this trade-off for a model system is shown in this publication:
http://theory.cm.utexas.edu/henkelman/p ... 0_9978.pdf
In one, you can generate a guess at the saddle as being 1/2 way between initial and final states, and start a dimer search right away.
If this is a bad assumption, even a crude NEB can give you a better approximation to the saddle with a few additional force calls.
An analysis of this trade-off for a model system is shown in this publication:
http://theory.cm.utexas.edu/henkelman/p ... 0_9978.pdf