Query Regarding Error Message in Pewald3D Calculator

[paulalex...@gmail.com]

Hi

Just a follow up to my previous question regarding the use of the pewald3D calculator.
I think I've got everything set up correctly, since the calculator runs, and is giving me sensible numbers.

The problem I'm encountering is that for some of my molecules, the calculation fails with the message: "Converged R-sum = true, converged K-sum = false, converged induction = true". From Carl's thesis, I take it that this message pertains to the interaction energy of the polarisation cloud. 

My question is whether there are any common pitfalls I should double-check, and/or any parameters I can change in an informed way to help these calculations complete successfully?

Thanks so much,

Paul

[Denis Andrienko]

Hi Paul,

This sort of error normally means that there is a deeper issue - either in your molecular configurations, or in the mps files. Can you have a look at the distribution of site energies for those molecules that converged (charged minus neutral energies). If this is too broad (more than 0.5 eV or so) then I would first check the forcefields, both polarizable and static.

If this is not the case, you can play a bit with the convergence criteria, though this can become computationally demanding.

you can try to reduce wSOR_N to 0.2

increase the number of iterations to 1000

decrease a bit the tolerance (not recommended)

 

I cc this email to Carl who coded this part, he might give you more hints.

Best,

Denis

--
You received this message because you are subscribed to the Google Groups "votca-ctp" group.
To unsubscribe from this group and stop receiving emails from it, send an email to votca-ctp+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/votca-ctp/e3911287-fb07-49d8-b589-9f12f4b42d8fn%40googlegroups.com.

[Paul Hume]

Hi Carl and Denis

Thank you both very much for your responses, I really do appreciate your time!
I'll start working through your suggestions now, but will briefly address Carl's questions:

*Yes, I ran the background polarisation calculator first
*I've been using an 8 nm cutoff for Coulomb interactions, and 6 nm for polarisation (just taken from the example in the manual)

While I have you, I also have a practical question regarding the set-up for shape terms. If I wan't to simulate IE/EA in a thin-film context, is the correct way to do that to construct a supercell with the desired film thickness (i.e. repeat the XRD unit cell in one direction), and then run the calculator with the xyslab shape term? Whereas for the non-film case, I just use the cube shape term and input the XRD unit cell?

Kindest regards,

Paul

On Wed, Feb 3, 2021 at 9:45 AM Carl Poelking <carlpo...@googlemail.com> wrote:

Hi again

I just remembered that there should be a parameter "kfactor" that sets an upper limit for the k-space cutoff ~ kfactor/real-space cutoff. You can try increasing that.

The convergence criteria are relatively tight, so you could also try to relax the threshold via "convergence.energy".

Internally the code does not use a hard cutoff, but sums over consecutive shells in k-space until some criterion is satisfied. Running the code in debug mode would give you more information where things are going wrong.

Note that changing the polarization parameters is not going to fix the convergence issues.

My feeling is that you're probably ok, and that the error has mostly to do with overly strict convergence bounds.

Carl

[Paul Hume]

Hi Carl and Denis

Thanks again for your help - I'm still trying to figure the convergence issue out, but I think it may be a problem with my system geometry.

The reason that I think this is that when I check the mapping with tdump, the extract.trajectory_qm.pdb file has some quite severe clashes between alkyl groups.

My understanding was that this file essentially takes the qm geometries and maps them onto the centre-of-mass positions of the md system, so I hadn't initially worried about it. But it would be great if you could let me know whether my interpretation is correct, or not?

Kindest regards,

Paul

[Denis Andrienko]

Hi Paul,

Overlap of sidechains is a nnown issue ;)

Every CH3 or CH2 group has to be a bead, this normally helps. Or chop the sidechains off - either by hand or with a .map file.

If it does not work, I will forward you to someone who recently had this problem.

cheers,

Denis