cell_opt issues

[Luca Dau]

Dear Matt

Thank you for your help. 
The following three keywords lead to wrong cell optimizations:

         CALCULATE_C9_TERM .TRUE.
         REFERENCE_C9_TERM .TRUE.
         LONG_RANGE_CORRECTION TRUE


Now everything is fine.
Regards 

Luca  Dau.



Il giorno giovedì 27 aprile 2017 10:16:10 UTC+2, Matt W ha scritto:
>
> I could only guess that maybe something is weird with the dispersion 
> corrections.
>
> Your cell (previous one) was just expanding along the c axis.
>
> My only suggestion is to try turning off the C9 terms and see if that 
> changes things.
>
> Matt
>
> On Thursday, April 27, 2017 at 9:06:01 AM UTC+1, Luca Dau wrote:
>>
>> Dear Matt
>>
>> thank you for your answer. Unfortunately the energy increases during the 
>> optimization even when following your advice.
>>
>> Regards.
>>
>> Luca Dau
>>
>> Il giorno mercoledì 26 aprile 2017 10:09:54 UTC+2, Matt W ha scritto:
>>>
>>> You could try using the direct cell optimization routines (TYPE 
>>> DIRECT_CELL_OPT), and default settings for the optimizers, they should be 
>>> much more efficient.
>>>
>>> Matt
>>>
>>> On Monday, April 24, 2017 at 1:48:02 PM UTC+1, Luca Dau wrote:
>>>>
>>>> please find attached all the files you need.
>>>>
>>>> Luca 
>>>>
>>>> Il giorno lunedì 24 aprile 2017 14:22:06 UTC+2, Matt W ha scritto:
>>>>>
>>>>> Hi,
>>>>>
>>>>> I can't see anything obviously wrong with your input deck.
>>>>>
>>>>> You'd better post a sample of the output, just first couple of 
>>>>> geometry steps so we can see forces/gradients.
>>>>>
>>>>> Matt
>>>>>
>>>>> On Monday, April 24, 2017 at 10:57:10 AM UTC+1, Luca Dau wrote:
>>>>>>
>>>>>> Dear Developers 
>>>>>>
>>>>>> I am trying to optimize the cell of a 6x6x4 h-BN supercell. My 
>>>>>> problem is that the energy increases instead to decrease. 
>>>>>> I made many tests but I am not able to find the solution to this 
>>>>>> issue. I have attached input and geo. 
>>>>>> Furthermore I have a similar problem with a supercell of metallic Ti.
>>>>>> Could you please help me?
>>>>>>
>>>>>> Thank you in advance.
>>>>>> Luca
>>>>>>
>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20170428/127712df/attachment.htm>