Plenty of standard plane-wave codes are able to compute correlation energies on top of DFT solutions using the random-phase approximation (RPA). I was wondering what the typical capabilities are in terms of scaling (Default RPA is O(N^4), but approximations can lower it) and the largest systems the codes report to be treatable. I am most strongly focused on open-source codes.
-
1$\begingroup$ I am not an expert in this, but I have some references that I rely on for accurate energy calculations including DMC, RPA, etc. I think none of them are open source. But you can track more papers from these that may lead you to what you want: (i) RPA on hBN: (1) Wu t al. J. Chem. Phys. 2016, 144, 164118. (2) water-CNT interactions: Al-Hamdani et al. J. Chem. Phys. 2017, 146, 094701. (3) Hongo et al. J. Chem. Theory Comput. 2015, 11, 907–917. $\endgroup$– gogoJul 2, 2020 at 3:45
-
$\begingroup$ Thanks very much. If you transform this into answer, I'd be more than happy to upvote ;) $\endgroup$– Michael F. HerbstJul 3, 2020 at 6:41
-
2$\begingroup$ @gogo Can you transform it into an answer? It would be really nice to get the unanswered queue cleared, and this question is more than 30 days old, so you'll get the "revival" badge for bringing the question back to life. $\endgroup$– Nike DattaniJul 4, 2020 at 14:53
-
$\begingroup$ @gogo it's a bounty now $\endgroup$– Cody AldazJul 19, 2020 at 21:46
-
$\begingroup$ Added some more information that I became recently aware of. Apologies for the delay. $\endgroup$– gogoJul 21, 2020 at 22:16
1 Answer
Sorry for the delay. The most recent capabilities I am aware of using VASP version 6 (not open source unfortunately) includes calculation of single point energies, geometry optimizations (some caveats here exist in terms of how strongly variational principle is captured), and I think also phonon calculations. Interestingly, VASP6 has an O(N^3) scaling. This means depending upon the computational resources available, one can model several hundred atoms in the unit cell. I am sure VASP manual has more technical details on this.
If one only wants to do single point energy calculations here, it would be important to benchmark or understand how much the RPA energies will depend unit cell parameters that was obtained from a particular DFT flavor.