My Community

I have a completed DFPT (IBRION=8) job with the VCA method turned on. The system is 80at.%Ti, 20at.% V. Typically I use PHONOPY to process the output (using phonopy --fc vasprun.xml). However, with the VCA turned on, it looks like some of the dynmat hessian elements (specifically, it looks like only the diagonal elements, but I could be wrong) are reported as ********. I've attached the files required in the forum rules but of course, I'd be happy to provide any other job files/details that might be important here. Any help in understanding and/or fixing this would be greatly appreciated.

Dear likith,

Thank you for bringing up this issue!

Indeed, we could reproduce the reported behavior. Unfortunately, the VCA method has not been previously tested with DFPT (IBRION=7, and 8) or finite differences (IBRION=5, and 6) . Therefore, we are very sorry for the inconvenience, but this combination of features is currently not supported. It has been added to our list of future improvements, though this might take considerable time until it is part of a release.

We highly appreciate your report and have added an appropriate comment in the documentation of the VCA tag. Sorry, this response is not more helpful.

Best regards,
Marie-Therese

Ps: Here is another thought. A different option to simulate doping is to actually replace 20% of Ti with V in a supercell calculation. This is perhaps an even better model for the true physical system and nicely agrees with your plan to compute phonon properties.

Hi Marie-Therese,
We are currently exploring options to simulate alloys with smaller quantities of alloying elements (at.%), and VCA was one of the options we were considering in order to be able to avoid the big supercells that are required for these simulations. But, we will of course have to make some decisions going forward.
Thank you very much for addressing this,
Likith