[CP2K-user] [CP2K:20350] Re: Lower LUMO energy than occupied molecular orbitals in OT calculations

[Marcella Iannuzzi]

Hi ..

If the system has no or very small gap, OT can fail and end up into a wrong 
state.
The calculation of the virtual states is done a posteriori, using the 
Hamiltonian obtained by OT,  this is the reason of the weird order. 

For such a system the diagonalization is a safer optimisation method, but 
proper mixing in G-space, with small mixing coefficient, and smearing need 
to be used. 
By convergence problems, reducing the mixing parameter and increasing the 
electronic temperature helps.

Other remarks: the pw-cutoff is far too low, the number of grids is 
unnecessary large (4-5 should be enough), if the material has magnetic 
properties it might be necessary to initialise the magnetisation of 
specific elements and the corresponding multiplicity. 

Regards
Marcella


On Wednesday, June 19, 2024 at 1:49:00 AM UTC+2 mshakiba.k... at gmail.com 
wrote:

> Dear CP2K developers,
>
> Hi, We are running a calculations for Cd33Se33 with Ferrocine ligands 
> (containing Fe). The input and the structure are attached. However, when 
> running a single-point calculations we face multiple problems. The first 
> thing is that the &DIAGONALIZATIO approach does not converge and it gives 
> large convergence values (like 100, 200, etc). Changing the basis set to 
> DZVP or even TZVP did not have any effect on this procedure. We also tried 
> adding smearing, increasing the cutoff, lowering the EPS_DEFAULT etc but no 
> effects on the convergence. 
>
> So, we started using the OT procedure and it converged the results in a 
> large number of steps (about 500). However, when printing out the energy 
> levels, we see that the LUMO energy is lower than the occupied orbitals 
> including HOMO, HOMO-1, HOMO-2, and HOMO-3. The results seem to be no 
> physical at all. So, even trying with unrestricted case, we get a similar 
> thing. Is this a problem of the OT procedure? Because the diagonalization 
> should print the eigenvalues in ascending order but in this case they are 
> not. We appreciate any suggestions to helps us understand the problem. 
>
> Thank you in advance.
>
> The attached files are the input, the coordinates, and one sample pdos 
> file showing the energies of the molecular orbitals.
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/23297321-6cb3-4aad-9962-ca6001ce4351n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240618/3fd1627e/attachment.htm>