QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Dipankar Saha on November 25, 2015, 20:25

Title: TB09 meta-GGA
Post by: Dipankar Saha on November 25, 2015, 20:25
WARNING  :-                                                                                                                             
The computed TB09 meta-GGA exchange-correlation potential                 
diverged in   0.03 % of the simulation volume, and was                       
 truncated to be in the range [-10.0000,  10.0000] Hartree
___________________
What does it signify ??

Thanks in advance..... :)

Best_
Dipankar
Title: Re: TB09 meta-GGA
Post by: Jess Wellendorff on November 26, 2015, 10:23
It signifies a very minor numerical inaccuracy in evaluating the TB09 exchange potential in some particular and very small region of the simulation cell. No need to worry. This can happen in regions of small density/high density gradients, which is usually in vacuum regions. But again, 0.03% does not signify a real problem, so just keep going :)
Title: Re: TB09 meta-GGA
Post by: Dipankar Saha on November 26, 2015, 13:30
Thanks a lot for the reply...!! :)

Regards_
Dipankar
Title: Re: TB09 meta-GGA
Post by: Anders Blom on November 26, 2015, 23:13
So, you have vacuum in the cell... In general, it will not work to determine "c" self-consistently then; instead, you will need to fix "c", and perhaps fit its value based on a reference band gap or whatever.

The warning comes from the fact that one term in MGGA diverges if the density (or rather, it's gradient) goes to zero. So while 0.03% is not a lot, it's actually more complicated than that. Note that it says the potential "diverged in 0.03 % of the simulation volume" - yes, in a small volume, but it DIVERGED! So the potential now is infinite (or, 10 H which we set it to) somewhere in space where it probably should be zero, or 1 eV, or some other value. Thus it may still cause problems.
Title: Re: TB09 meta-GGA
Post by: Dipankar Saha on November 27, 2015, 15:06

Okay...I understand..../ Thank you  Anders for the reply...!!