[CP2K-user] [CP2K:12439] Divergence of Broyden mixing method without OT

[Mohammad Shakiba]

Dear Juerg Hutter ,

Thanks for your useful information. You're right. I checked it and it seems 
it has converged to an excited state. I think the problem is because of my 
molecular structure since smearing also does not converge.

Thanks again and Best.

On Thursday, October 31, 2019 at 3:12:52 PM UTC+3:30, jgh wrote:
>
> Hi 
>
> if the restart with Diagonalization is not converged, this means 
> that your OT has converged to an excited state (assuming no other trivial 
> mistakes of handling files etc.). 
>
> You can check this with this procedure: 
>
> 1) converge using OT 
> 2) restart using Diagonalization, do only 1 SCF step 
> 3) restart using OT, you should now get a lower energy. 
>
> This usually is possible for systems with a very small gap. 
> Try Diagonalization using smearing in that case. 
>
> regards 
>
> Juerg Hutter 
> -------------------------------------------------------------- 
> Juerg Hutter                         Phone : ++41 44 635 4491 
> Institut für Chemie C                FAX   : ++41 44 635 6838 
> Universität Zürich                   E-mail: h... at chem.uzh.ch 
> <javascript:> 
> Winterthurerstrasse 190 
> CH-8057 Zürich, Switzerland 
> --------------------------------------------------------------- 
>
> -----c... at googlegroups.com <javascript:> wrote: ----- 
> To: "cp2k" <c... at googlegroups.com <javascript:>> 
> From: "Mohammad Shakiba" 
> Sent by: c... at googlegroups.com <javascript:> 
> Date: 10/31/2019 12:27PM 
> Subject: Re: [CP2K:12439] Divergence of Broyden mixing method without OT 
>
> Dear Juerg Hutter, 
>
> Thanks for you reply. I tried this but it does not converge. I used this 
> method for other systems (for example for CdSe or PbSe quantum dots) but 
> when I use it for perovskite, it does not work. I thought it probably is 
> because of iodine, so changed the iodine to Br or Cl but did not change 
> anything. Now I wonder that it is because of my molecular structure since I 
> have optimized the structure using OT method. Also, I have run MD at 300K 
> and the MD results are good. Do you think that the structure is my problem 
> and I have to change it? The same structure image is shown below: 
>
> Thanks and Best. 
>
>
> On Thursday, October 31, 2019 at 2:00:19 PM UTC+3:30, jgh wrote:Hi 
>   
> this should work for you (as far as I get from your information). 
>   
> 1) Optimize wfn with OT 
> 2) Restart using Diagonalization and ADDED_MOS 
>    This should be converged in 1) so no SCF iterations 
>    Calculate (print) MOs 
>   
> regards 
>   
> Juerg Hutter 
>     
> -------------------------------------------------------------- 
> Juerg Hutter                         Phone : ++41 44 635 4491 
> Institut für Chemie C                FAX   : ++41 44 635 6838 
> Universität Zürich                   E-mail: h... at chem.uzh.ch 
> Winterthurerstrasse 190 
> CH-8057 Zürich, Switzerland 
> --------------------------------------------------------------- 
>   
> -----c... at googlegroups.com wrote: ----- 
> To: "cp2k" <c... at googlegroups.com> 
> From: "Mohammad Shakiba"   
> Sent by: c... at googlegroups.com 
> Date: 10/31/2019 11:19AM 
> Subject: [CP2K:12437] Divergence of Broyden mixing method without OT 
>   
> Hello everyone, 
>   
> I wanted to print the MOs and I used the Broyden mixing method instead of 
> DIIS while running MD for an optimized perovskite system with OT method. 
> The DIIS mixing diverges quickly and the Broyden mixing does not change 
> significantly and the convergence does not go further than 1.0E-2. My 
> system is a MAPbI3 perovskite consisting of Pb and Iiodine atoms. I use the 
> DZVP-MOLOPT basis set and GTH pseudopotential. 
>   
> I tried reducing EPS_DEFAULT and change other parameters but did not work. 
> When I use it with the OT method by reducing the EPS_DEFAULT I obtain the 
> convergence but not able to print the MOs containing LUMOs. I even used the 
> wfn file obtained from the OT calculation with Broyden but again it does 
> not work. Also, I used SZV basis set and different types of solvers for 
> POISSON. None of them worked and I am completely confused. Can anyone help 
> me to obtain the convergence? Or any hint so I can print the MOs with LUMOs 
> (since with OT I can print MOs but they do not contain the LUMOs)? 
>   
> Thanks in advance. 
>   
> Here is my input:   
>   --   
>  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 c... at googlegroups.com. 
>  To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cp2k/273264c3-b84c-4dbd-8736-74df084aa9c4%40googlegroups.com. 
>
>   
>   
> [attachment "B.inp" removed by Jürg Hutter/at/UZH] 
>   
>   -- 
>  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 c... at googlegroups.com <javascript:>. 
>  To view this discussion on the web visit 
> https://groups.google.com/d/msgid/cp2k/d0a88369-6d46-4d04-980e-6e6cb0759b39%40googlegroups.com. 
>
>   
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191031/52dbe1ee/attachment.htm>