Confusion about Real space scheme for TD-ALDA

[hplan]

Dear Developers:

The technical paper for yambo stated that we need real-space kernel when we deal with inhomogeneous situations such as clusters exemplified in the paper. The tutorial for silicon surface also gave real-space kernel as default setting. However, i found a work of graphite system PRL 89 076402 didnot give any discussion on XC kernel problem but directly applied ALDA kernel. In this work, the authors even increased the interlayer spacing to about 7.0 \AA to examine the interlayer interaction's effect on EELS. I wonder why ALDA is OK for this layer system , since there are some interlayer spaces with vanishing density ?

Another problem is about EELS for R-space scheme. I just found it is really formidable to obtain accurate EELS with this scheme due to large BSK dimension resulting from a lot bands included. Is there any special consideration for EELS with R-space scheme ?

Best Wishes,

Hai-Ping

[andrea marini]

The technical paper for yambo stated that we need real-space kernel when we deal with inhomogeneous situations such as clusters exemplified in the paper. The tutorial for silicon surface also gave real-space kernel as default setting. However, i found a work of graphite system PRL 89 076402 didnot give any discussion on XC kernel problem but directly applied ALDA kernel. In this work, the authors even increased the interlayer spacing to about 7.0 \AA to examine the interlayer interaction's effect on EELS. I wonder why ALDA is OK for this layer system , since there are some interlayer spaces with vanishing density ?

The problem described in the Yambo paper is clearly understandable in the case of molecules and nanoscale materials. Graphite is not a molecule and not a nanoscale material and the electronic states are mostly localized on the plane. These states are not affected by the layer-layer distance and they contribution to the polarizability for light polarized parallel to the plane is always well defined.
In the perpendicular direction, instead, you may see some effect but I have no idea how large the layer-layer distance should be to see the effect of the vanishing density.

Another problem is about EELS for R-space scheme. I just found it is really formidable to obtain accurate EELS with this scheme due to large BSK dimension resulting from a lot bands included. Is there any special consideration for EELS with R-space scheme ?

Why are you using the BS equation to calculate the EELS ? The Tamm-Dancoff approximation used by Yambo to solve the BS equation makes the calculation of the EEL quite tricky.

Why do not you use the reciprocal space method (yambo -o c) ?