[CP2K:175] Re: NEB output

[Teodoro Laino]

SCF is an iterative procedure.. It would converge in 1 step if the  
guess of the wavefunction (that you use to start solving
the SCF equations) would be exactly the solution of the SCF.. Well..  
Let's say it's challenging at the moment.. ;-))
Depending on the nature of your system and the preconditioner you use  
you may expect a wide range of scf cycles..
In the best case  ~<10.. in others also of the order of 100 or more..

Teo

On 12 Jul 2007, at 15:38, monica wrote:

>
> But, I would expect, for each geometry, SCF to converge once. Instead,
> for 5 geometry opitmization steps there are 22 SCF opitmization steps.
>
>
>
> Teodoro Laino wrote:
>> Then that one is the output of the SCF calculations (energy+forces)
>> for that specific replica..
>> Output for updated geometries are appended... so you have the full
>> history..
>>
>> Teo
>>
>> On 12 Jul 2007, at 15:26, monica wrote:
>>
>>>
>>> I was talking about the output of one of the replicas.
>>>
>>>
>>> Teodoro Laino wrote:
>>>> On 12 Jul 2007, at 14:26, monica wrote:
>>>>
>>>>>
>>>>> Hi,
>>>>>
>>>>> I would like to ask about the output of the NEB calculation.
>>>>> In the output, of each replica, several SCF convergence cycles are
>>>> There are many output files (as many as the number of parallel
>>>> replica you're using) in a NEB calculations.. Which one are you
>>>> talking about?
>>>>
>>>> teo
>>>>> printed. The number of those SCF cycles is larger than number of
>>>>> geometry opimization steps printed in the corresponding *.xyz  
>>>>> file.
>>>>> For example, for 5 geometry opitmization steps (printed in the  
>>>>> *.xyz
>>>>> file) I could find 22 SCF convergence cycles.
>>>>> I would like to know to what stages of calculation those SCF
>>>>> convergence cycles refer to.
>>>>>
>>>>> Thank you in advance,
>>>>> Monica
>>>>>
>>>>>
>>>>>>
>>>
>>>
>>>>
>
>
> >