[CP2K-user] [CP2K:17859] Re: Vibrational analisys
Marcella Iannuzzi
marci.akira at gmail.com
Thu Oct 13 16:19:38 UTC 2022
Dear Lucas
In the main output you should find how the calculations are distributed
among the processors' groups.
It is possible that some calculations are done more than once to avoid that
processors remain idle.
What happens if you run the 48 calculations with 24 cores?
Kind regards
Marcella
On Thursday, October 13, 2022 at 5:38:28 PM UTC+2 Lucas Lodeiro wrote:
> Hi Marcella,
>
> I expressed it badly, I want to say that each displaced atomic
> configuration needs an energy+gradient calculation, this is 6N
> energy+gradient evaluations, plus the minimum energy geometry. Of course
> the number of output files is equal to the replica layout, but my problem
> is that the number of energy+gradient evaluations into all these replica
> outputs is not equal to 6N.
> A practical example, I have an 8 atom isolated molecule, so it needs 48
> energy+gradient evaluations which are printed in the main output correctly.
> I launched this calculation in a node with 20 cores, so the replica layout
> is 20 and there are 20 output files. When I see these output files I check
> there are 5 energy computations into each output, so there are 5*20=100
> energy evaluations. This is the point which makes me worried. I do not
> understand why there are that huge amount of energy evaluations when 6N=48.
>
> Regards - Lucas
>
> El mié, 12 oct 2022 a las 4:51, Marcella Iannuzzi (<marci... at gmail.com>)
> escribió:
>
>>
>> Hi Lucas
>>
>> The number of output files of the type ${projectname}-r-#.out
>> should be equal to REPLICA| layout of the replica grid, number of groups
>> and not to 6N
>>
>> Kind regards
>> Marcella
>>
>> On Wednesday, October 12, 2022 at 12:35:08 AM UTC+2 Lucas Lodeiro wrote:
>>
>>> Hello all,
>>>
>>> I have a detailed doubt about the files and data printed for a
>>> vibrational analysis. AFAIK, the calculation is done by numerical
>>> differentiation of gradients (differentiation in both directions and
>>> average them) so for a molecule with N atoms it needs 6N calculations,
>>> which is clearly printed in the main output. Also, each
>>> replica/configurationn (6N) is computed and a particular output is printed
>>> for each one.
>>> Up to this point everything looks fine, but when I see the particular
>>> replica output, there are 5 energy computations into the output... I would
>>> expect just one.
>>> Why is this happening? Am I missing something?
>>>
>>> Regards - Lucas
>>>
>> --
>> 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+uns... at googlegroups.com.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/cp2k/90776d8b-086e-4772-8683-8dd97b5afc8dn%40googlegroups.com
>> <https://groups.google.com/d/msgid/cp2k/90776d8b-086e-4772-8683-8dd97b5afc8dn%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>
>
--
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 on the web visit https://groups.google.com/d/msgid/cp2k/a685a265-6fd7-4970-83d4-050344a60db9n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20221013/97ed0b72/attachment.htm>
More information about the CP2K-user
mailing list