Question about TDDDH

[hong_tang1]

Dear Alexey Tal or other VASP experts,

I am wondering how TDDDH ( Tal et al. PHYSICAL REVIEW RESEARCH2,032019(R)(2020)) is actually calculated in vasp.5.4.4.

As I understand, in TDDDH, both the kernel fxc and the screened Coulomb W effects are included.
For fxc (LFXC = .TRUE.) , I suppose it is included by solving Casida Equation, right?

Then, for the screened Coulomb W, it is by using a model dielectric function for the screening. We add
"ALGO = TDHF
LFXC = .TRUE.
HFSCREEN = 1.70
AEXX = 0.175
LMODELHF = .TRUE." So, the LADDER=.True. is also activated. My question is how the LADDER effect is calculated? Is it in the Casida equation? Or you modified the Casida Equation for the so-called mBSE to be cable to include the screened Coulomb W related contribution? Or the screened Coulomb W related contribution is calculated by a separate subroutine? Then just add to the result of solving Casida equation?

Please help me to understand, I appreciate your help very much!


Best,
Hong

[alexey.tal]

Dear Hong,

As I understand, in TDDDH, both the kernel fxc and the screened Coulomb W effects are included.
For fxc (LFXC = .TRUE.) , I suppose it is included by solving Casida Equation, right?

That's correct. LFXC = .TRUE. includes the local part of the xc kernel in the Hamiltonian.

My question is how the LADDER effect is calculated? Is it in the Casida equation? Or you modified the Casida Equation for the so-called mBSE to be cable to include the screened Coulomb W related contribution? Or the screened Coulomb W related contribution is calculated by a separate subroutine? Then just add to the result of solving Casida equation?

The "ladder" diagrams are taken into account in TDDDH via the screened potential W or the nonlocal part of the exchange kernel in the Hamiltonian, so yes this term is included in the Casida equation.