Force-Noise of SCAN and r2SCAN

[mwuensch]

Hey everyone,
quite a while ago I got the tip to use small EDIFFGs of up to a minimum of -0.001 when using the SCAN+rVV10, since it has quite noisy forces. I am now switching to the r2SCAN+rVV10 functional and am wondering whether this advice still applies. Does anyone have experience with this?

Cheers
Max

[martin.schlipf]

In general, metaGGAs are more noisy than GGAs so the advice still applies. That being said r2SCAN is more robust than its precdessors so you may need to be a bit less strict than for SCAN.

However the important point is to always test these kind of parameters for the system of interest because the absolute value will be e.g. quite different for MD runs or phonon calculations.

[mwuensch]

Thank you!