Parallel tempering & metadynamics

[rwindiks]

Dear VASP experts,

is the implementation of parallel tempering of VASP 6.4.3 compatible with any of the advanced MD techniques such as metadynamics, slow growth dynamics, simulations within the blue-moon ensemble? Can metadynamics simulations be reasonably performed in conjunction with parallel tempering?

Thanks.

René

[martin.schlipf]

We have not tried these features in combination so it is possible that this will not work right away. We will explore this and get back to you.

[rwindiks]

To help you a bit to exlpore: It looks as in case of metadynamics simulations all Gaussian-like bias potentials are written to one file HILLSPOT which is located in the main folder and not in the sub-folders 01, 02, etc. That is already indication that both features are not 100% compatible. However, every MD simulation writes its own file REPORT in the sub-folders. Hence, in principle it is possible to re-construct the files HILLSPOT for each individual MD simulation of the parallel tempering calculation.

Nonetheless, it would be very helpful if files such as n files HILLSPOT are written in each subfolder, whereby n is the number of individual VASP calculations.

Thanks.

René

[martin.schlipf]

If that is the only issue that should be easy to fix. VASP has usually a prefix DIR_APP(1:DIR_LEN) that is used when opening files that ensures that they go to this subdirectory. I can see in dynconstr.F that this has been commented out for one access to the HILLSPOT file. Perhaps adding that back would already solve the issue.