[QE-users] Convergence testing for charge density

[Léon Luntadila Lufungula]

Dear QE users,

I'm currently performing convergence tests for my system by looking at the energies, forces and stresses for different values of ecutwfc, ecutrho and k-points to get the right optimized parameters for my calculations. However, during my research I will be mainly focused on properties of the electron density as I'll be performing QTAIM, ELF and NCI calculations, so I would like to converge my parameters with respect to the electron density distribution of my system but I'm unsure about the correct procedure in this situation. I have a couple of things I'm uncertain about:


  1.  I use the pp.x code to create a .cube file containing the electron density of my system, which uses the values of nr1,nr2,nr3 to construct the .cube file. The values of nr1,nr2,nr3 are themselves dependent on the values of ecutwfc and ecutrho, so by varying these values , I would obtain different grids in consecutive calculations, which makes it hard to compare the results. Would manually assigning values to nr1,nr2,nr3 during the SCF calculations be a valid option to ensure the grids of the obtained cube files are the same? Or could this hinder the convergence as the FFT-grids are now sub-optimal in some cases?
  2.  Once the .cube files for the different calculations are generated, how would one choose which values to compare to get an idea of the convergence? Would it be sufficient to choose a given point inside the .cube file and see how this value converges with the given parameters, or would you have to (in some way) look at all the values inside the .cube file to assess convergence? Perhaps you could do this by calculating the deviation in the electron density at each point in the .cube file w.r.t. some reference calculation with very high values for ecutwfc, ecutrho and k-points and then taking the average of the deviation at all points inside the .cube file to get an actual number to compare between calculations.

If anyone has some experience with convergence testing w.r.t the electron density I would be very happy to get some help on the subject!

Thanks in advance and happy holidays,
Léon Luntadila Lufungula
Structural Chemistry Group
University of Antwerp
Belgium
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20221222/01a14cb2/attachment.html>