doubble nudging

[kai]

Could anyone tell me a bit more regarding the tag "LDNEB", implemented as a double nudging (variable spring constant)? how vasp would change spring constant if i use this tag to apply to the connection between different images?

I am facing a peculiar difficulty to further converge a set of climbing-NEB that has been precoverged to 0.05 eV/A. Extremely tight ediff and multiple optimizer have been tried. The problem appears that the limiting image (not the transition state) was flucturating betweeen two configurations.

I am thinking of turning on the double nudging to see if the varible spring helps with this situation. Do I need to change the spring setting from the default -5 when using the LDNEB tag?

thanks

[graeme]

The double nudging scheme is a way to keep a portion of the spring forces so that the band tends to stay straighter as it converges. It was developed in the Wales group, and found to be useful for long paths in biological systems. There are, however, some convergence problems when the force gets close to zero, and we have not found advantages using the method with diffusion paths in solid state systems.

The original method is described here:
S. A. Trygubenko and D. J. Wales, J. Chem. Phys. 120, 208 (2004).

and some comparisons with other methods here:
D. Sheppard, R. Terrell, and G. Henkelman, J. Chem. Phys. 128, 134106 (2008).

[kai]

thanks, Graeme; I am testing it and would post the result. This is regarding the molecular dissociation on silicon surface.

The whole reaction appears to involve of two separate saddle points en route from initial to final states (two bumps in 1D energy profile). Quite interestingly, these two saddle points seems to be quite close in terms of activation energy.