Page 1 of 1
r2SCAN-L electronic convergence
Posted: Sat May 10, 2025 2:23 pm
by chamara_somarathna
Hi all,
I am investigating the r2SCAN-L functional, which is said to have numerical stability over SCAN. I set METAGGA = R2SCANL and LASPH = TRUE, as it is advised to do so for meta-GGA functionals. However, I am still experiencing numerical instabilities even at the 600 eV plane wave cutoff. meta-GGA functionals are expected to lead to numerical instabilities at high cutoffs. However, at 600 eV cutoff, it seems a bit extreme (it took 100 SCF iterations to reach 1E-6 convergence).
Is there any other setting I have to set or be useful for improving the stability of METAGGA = R2SCANL?
Appreciate your valuable advice. Thank you very much.
Best,
Chamara
Re: r2SCAN-L electronic convergence
Posted: Sun May 11, 2025 1:08 pm
by fabien_tran1
Hi,
This is r2SCAN that is supposed to be more stable than SCAN according to the literature. Concerning the deorbitalized versions SCAN-L and r2SCAN-L, one could suppose that r2SCAN-L is more stable than SCAN-L, but tests should be done to confirm that.
Regarding tau-MGGAs versus Laplacian-MGGAs, a few tests gave me the impression that tau-MGGAs are more stable.
As you mentioned, high ENCUT could lead to numerical instability, but this concerns the Laplacian-MGGAs, while the tau-MGGAs do not seem to suffer from this problem. If you want help for your convergence problems, you need to upload the input files of your calculations.