My Community

Hi everyone,

we want to investigate the influence of the density gradient on DFT-MD simulations under warm, dense matter conditions. Such an analysis was done for Quantum simulations at https://pubs.acs.org/doi/full/10.1021/acs.jctc.3c00934 (see Fig. 2).

Accordingly, I want the reduced density gradient (s-parameter) for a finite grid of x-y-z positions. Currently, I am using VASP 6.4.2. In xclib_grad.F (line 250) is an option to print the (s-parameter). The output, however, isn’t helpful since the positions aren’t printed (and during a first test I got negative values which doesn't make sense to me).

Do you see any option to get the output I’m looking for?

Many thanks for your help,

Armin Bergermann

Dear Armin,

thank you for reaching out to us on the official VASP forum.

As you already noticed, this is currently not supported in VASP. The method in question you referring to has been moved in VASP 6.4.3 to xc_driver.F . I thing adding output there is not very helpful during MD simulations as this part of the code is called every electronic scf steps and would produce huge amount of output data. You ideally want to maybe write this information at the end of subroutine ELMIN in electron.F , but I think this has not been done before. I will ask around in the team if someone tried this already.

Best regards,
Alex H.

Thank you very much!

You are welcome. I asked around and we think that the fastest way to get to the gradient of the density is via the tag https://www.vasp.at/wiki/index.php/LAECHG which will write out 3 files. VASP will try to recreate the all electron charge density on the fine fft mesh in real space. See https://www.vasp.at/wiki/index.php/CHGCAR#Format for details on how this mesh is build. Then you should be able to get nice gradients via spline interpolation on that mesh.

Internally, we had once some feature branch that allows to calculate these properties, in some preliminary rough way, but it never made it to a release. We will get back to that and plan to bring this into a future release of VASP, but unfortunately we need some time to polish this and test it. So right now the approach above will be the fastest way to calculate the property.

Best,
Alex H.