[QE-users] Computing Charge Density Differences

[Victor Bermudez]

Hello,
	I'm trying to compute the charge density difference for a molecule on a substrate, modelled as a two-dimensionally-periodic slab. The relationship I'm using is 

	Diff(n) = w1*n(mol.+slab) - w2*n(slab) - w3*n(mol.) 

where "n" is a density computed using Postprocessing (pp.x, plot_num=7) and w's are weighting factors. 
	I'm aware that all densities have to be obtained with the same atom positions, ecutwfc, ecutrho, etc., and I've done that.  I'm having trouble getting results that are not obviously wrong (e.g., Diff(n) that is orders of magnitude too large). I've read the available posts that I can find, but none addresses my questions, which are: 

(1) The SCF calculations used to obtain the wavefunctions were all done with 'vdw_corr=TS'. When I run pp.x to get the densities, the output file includes the message 
	*** vdW-TS term will be missing in potential *** 
Does this mean that I shouldn't have used TS when computing the wavefunctions ? Can this message be ignored, or is there a fatal error ? 

(2)  The molcule is paramagnetic, so the SCF calculations for n(mol.+slab) and n(mol.) were spin-polarized, but that for n(slab) was not. In obtaining Diff(n) I'm using the highest majority-spin bands for n(mol.+slab) and n(mol.). 
	Can I then use the highest doubly-occupied band for n(slab) if I set w2/w1 = 0.5 ? 
	Do I have to do a spin-polarized calculation for n(slab) even though it's diamagnetic ? 
	Is further adjustment of w2/w1 need to account for the fact that mol.+slab and slab involve a different total number of electrons ? 

(3)	The slab comprises a 4x4 supercell with 1 molecule. Should I then set w3/w1 = 1/16 ? Again, is any further adjustment of w3/w1 needed ? 

(4)	In the SCF calculations of n(mol.+slab) and n(slab) I used a 3x3 MP grid (because of the large supercell). For n(mol.), on the other hand, I used just the gamma point (because the isolated molecules are far apart).
	Is that OK, or should I have used the same MP grid in obtaining n(mol.) ? 

Thanks in advance for your help. 

Best Wishes,
Vic Bermudez
U.S. Naval Research Laboratory (retired)