[CP2K-user] [CP2K:22101] Re: Energy calculation for charged system LSD/UKS ?

[Andreas Döll]

Thank You for Your answer. 
I know about the new "tutorial paper" and would recommend anyone to read 
it. But why would EPS be the cause of my problem? This cannot be the cause 
for density at the edges? These values were for testing anyways, for 
production I make my convergence criterion more tight.

Changing the following things does nothing:

CUTOFF 1000 
EPS_DEFAULT 1.0E-9
EPS_SCF 1.0E-8

I played around and still, immediately at step 1, I get density at the edge 
error message. I dont think EPS values are the source of the problem
Any input is appreciated!

Cheers, 
Andreas
bmk schrieb am Donnerstag, 12. Februar 2026 um 14:02:34 UTC+1:

> It could be related to your EPS_SCF and EPS_DEFAULT, which are both quite 
> high. For DFT-MD I usually set EPS_SCF to 1.0E-6 and EPS_DEFAULT to 
> 1.0E-12, and for accurate single points I set EPS_SCF to 1.0E-8 and 
> EPS_DEFAULT to 1.0E-16. Check 
> https://pubs.acs.org/doi/full/10.1021/acs.jpcb.5c05851 section 2.1.4 for 
> more details.
>
> The amount of OT inner SCF cycles also seems quite high to me.
> Op donderdag 5 februari 2026 om 17:07:27 UTC+1 schreef Andreas Döll:
>
>> Hello, 
>> I just want to update people who may have the same question in the 
>> future. After further research, my problems could be partially solved with 
>> the following changes to my input script: 
>>
>> ...
>> CHARGE +1
>>     LSD
>>     MULTIPLICITY 1
>> ...
>> POISSON_SOLVER
>> PERIODIC NONE
>> ...
>> &CENTER_COORDINATES
>>         &END
>> ...
>>
>> All of this, and a sufficiently large cutoff (my research shows 5+ A) 
>> worked for me for case a) where I have a multiplicity of 1. Centering is 
>> there to prevent problems at the edge of the box with the solver.
>> Sadly for system b) where the multiplicity is higher, I always get the 
>> usual wavelet solver issue, even though I 
>>
>> a) Center my atoms
>> b) Have a large box (more than 10 A from edges even)
>> c) have a large cutoff
>>
>> I keep getting problems with the Wavelet solver at the edge: Usually, the 
>> error message
>>
>>  *** WARNING in ps_wavelet_methods.F:238 :: Density non-zero on the edges 
>> ***
>>  *** of the unit cell: wrong results in WAVELET solver                   
>>  ***
>>
>> can be treated with a larger box, it seems to me from looking around the 
>> Cp2kgroup that higher multiplicity might be difficult, but even when 
>> increasing the box to have *20+ A vacuum* in each direction I still get 
>> a problem at the edge. I attach my input script for this issue, maybe 
>> somebody can spot some obvious mistake quickly! Is this maybe a chemistry 
>> issue (getting multiplicity wrong? I just cut out a molecule off a larger 
>> structure, so at each branch of the Triazine, a Hydrogen is missing)
>>
>> Thank you for your patience
>> Andreas Döll schrieb am Mittwoch, 4. Februar 2026 um 11:46:43 UTC+1:
>>
>>> Dear cp2k community,
>>>
>>> I have two charged systems I want to perform a single energy calculation.
>>>
>>> System a) Neutral molecule with a Lithium at some distance
>>> System b) Charged molecule (cut out of a larger molecular structure, so 
>>> 3 Hydrogens are missing) 
>>>
>>> For both calculations, my energy either diverges or makes jumps. 
>>> Usually I have performed calculations with neutral molecules, not 
>>> charged systems, so I am not familiar with the proper procedure. I followed 
>>> the example give here 
>>> https://www.cp2k.org/exercises:2014_ethz_mmm:hfx_h2ion 
>>> I defined my Charge in the DFT section, as well as using LSD option and 
>>> switched to a Nonperiodic system - as far as I understand this is 
>>> necessary? Do I need to define the multiplicity of my system as well ?
>>>
>>> Cheers, 
>>> Andreas
>>>
>>>
>>>

-- 
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 visit https://groups.google.com/d/msgid/cp2k/50535024-62b5-406e-8137-6b446a045177n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20260213/b133735c/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: output.dat
Type: application/octet-stream
Size: 63507 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20260213/b133735c/attachment-0001.obj>