[CP2K-user] [CP2K:10785] Converging b3lyp/6-311++G**

Matthias Krack matthia... at psi.ch
Mon Oct 1 21:26:53 CEST 2018


Dear Kristjan

EPS_FIT is too large. Try 1.0E-6 instead of 1.0E-4. The soft basis set 
should include at least all exponents smaller than 6.0. Check the hard and 
soft basis set, e.g. by selecting the PRINT_LEVEL medium which will print 
the basis set information. In addition reduce EPS_SCHWARZ to 1.0E-12.

HTH

Matthias

On Monday, 1 October 2018 20:02:03 UTC+2, Kristjan Eimre wrote:
>
> Thanks for the suggestion, Frederick.
>
> I exchanged the XC section with the one from 
> `regtest-hybrid-3/H2O-hybrid-b3lyp_shortcut.inp`, so
>
>     &XC
>       &XC_FUNCTIONAL B3LYP
>       &END XC_FUNCTIONAL
>       &HF
>         &SCREENING
>           EPS_SCHWARZ 1.0E-10
>         &END
>         &MEMORY
>           MAX_MEMORY  5
>         &END
>       &END
>     &END XC
>
> But unfortunately it's still not converging. I attached the input & output 
> of this run.
>
> (By the way, I was previously using B3LYP which is also used in the test 
> `regtest-libxc/H2O-hybrid-b3lyp_libxc.inp`)
>
>
> On Monday, October 1, 2018 at 6:46:44 PM UTC+2, Frederick Stein wrote:
>>
>> Dear Kristjan,
>>
>> You have to delete the line with the keyword FRACTION in the HF-section. 
>> And I would use the B3LYP shortcut. There is an example file in 
>> regtest-hybrids-3, I think.
>>
>> Cheers.
>> Frederick
>>
>> Kristjan Eimre <kris... at gmail.com> schrieb am Mo., 01.10.2018, 17:41:
>>
>>> Dear CP2K community,
>>>
>>> I am having trouble converging the SCF of a b3lyp/6-311++G** calculation 
>>> for a fairly small molecule consisting of C and H.
>>>
>>> Without too much trouble, I was able to converge the b3lyp/6-311++G** 
>>> for benzene and also b3lyp/6-311G** for my molecule (so without the diffuse 
>>> basis functions).
>>>
>>> I optimized the geometries also with b3lyp/6-311G**.
>>>
>>> Additionally, for my molecule, I converged blyp/6-311++G**, which I was 
>>> using as the initial guess to b3lyp calculation.
>>>
>>> I attached one of the inputs scripts that I tried and the corresponding 
>>> output.
>>>
>>> Here are the different things that I tried:
>>> * Played with the parameters in QS (most notably EPSFIT, ALPHA0_H)
>>> * OT with CG minimizer
>>> * Diagonalization
>>> * MT poisson solver
>>> * UKS calculation
>>> * smaller basis set, 6-311+G* (from EMSL)
>>>
>>> But unfortunately everything failed.
>>>
>>> Which knobs should I turn to possibly reach convergence? Any hints are 
>>> very much appreciated.
>>>
>>> best regards,
>>> Kristjan
>>>
>>> -- 
>>> 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+... at googlegroups.com.
>>> To post to this group, send email to cp... at googlegroups.com.
>>> Visit this group at https://groups.google.com/group/cp2k.
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20181001/947cdd46/attachment.html>


More information about the CP2K-user mailing list