Anyone know about LibXC TB09 implementation?

[Nicholas Winner]

Does anybody happen to know how to properly use the LIBXC implementation of the TB09 meta GGA exchange functional? I've applied it to GaAs with tight k-point sampling and the resultant DOS still has a band gap of ~0.7 eV when the literature shows TB09 should give ~1.5eV

I've accessed it using 

&XC_FUNCTIONAL

  &MGGA_X_TB09

  &END

  &LDA_C_PW

  &END

&END

The TB09 MGGA has a self consistent solution for the parameter "c" in the original implementation, is that not being done here?

Not too familiar with how libxc works under the hood but maybe someone else does.

-Nick

[Frederick Stein]

Dear Nick,

I am not too familiar with that functional. If I understand the documentation of LibXC correctly, then your setup should be correct. In the literatur, people apparently use the same combination of functionals.

Can you share your input file? With the scarce amount of information you provide, it is difficult to estimate what causes the deviation. It can be a too low cutoff, the wrong pseudo potential, the basis set etc.

Best,

Frederick

[Nicholas Winner]

Here is my input file. I have changed 1 thing here, which is that I manually adjusted the tb09 c parameter, which does open the gap. My working theory is that TB09 in libxc does not solve self consistently, and must be tuned like a +U or hybrid parameter.

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/3d6d3fb2-ae70-462a-912e-b4bdab040ebfn%40googlegroups.com.