A Cl-NEB and/or NEB must with a minus EDIFFG?

[cenwanglai]

I always run a NEB or Cl-NEB with the EDIFFG<0.

But, I can not make sure that the EDIFFG<0 is set to scale the the forces between images?

Can I set a EDIFFG>0 (basing on energy) for the NEB/Cl-NEB run?

[graeme]

There is only one harmonic degree of freedom between the images; you should not worry about convergence along the band when using the force criteria. If there was a reason to change the tolerance of the image spacing, you can change the spring constant. But our suggestion is to use the force-based convergence criteria with the standard spring constant. There is no need to use energy, especially since it does not provide a meaningful and reproducible convergence criterion.

[cenwanglai]

Thank you so much!