[CP2K-user] [CP2K:16168] Re: Problems with metadynamics postprocessing

[Marcella Iannuzzi]

Dear  JHu

if you understand how the reconstruction  of the FES works, you know that 
the Gaussian contributions to the metadynamics potential have a certain 
extension, hence they also contribute to parts of the CVs meatspace that do 
not correspond to points visited by the metatrajectory, which is the one 
reported in *-COLVAR.metadynLog. Moreover these Gaussian contributions sum 
up, and if the matadynamics settings are not properly chosen they might 
even produce misleading FES topographies. I cannot judge whether this is 
your case, but for sure it is to be expected that  the minima on the FES 
are not exact points of the metatrajectory in *-COLVAR.metadynLog. If they 
are in completely unexpected regions, far from those most frequently 
visited by the trajectory, the metadynamics settings are most probably not 
properly chosen. 

The -opt-path option of graph requires as input a path file, not the Hills 
file. 

Regards
Marcella




On Wednesday, November 3, 2021 at 7:06:56 AM UTC+1 hujin... at gmail.com wrote:

>
> Dear Marcella
>
> Thanks and sorry to bother you again. I understand the fact that the FES 
> generated by graph is reconstructed but the minima I got from graph which 
> contain the CVs' values cannot be found in *-COLVAR.metadynLog. Is this 
> situation normal?
>
> And I tried another command line "graph.popt -ndim 2 -ndw 1 2   -opt-path 
>   *HILLS.metadynLog  -orac".  And it still went wrong as follows:
>
>  ABORT in metadyn_tools/graph.F:198 CPASSERT failed
> forrtl: error (76): Abort trap signal
>
> Image              PC                Routine            Line        
> Source             
> graph.popt         0000000000500274  Unknown               Unknown  Unknown
> libpthread-2.17.s  00002B6FF116F630  Unknown               Unknown  Unknown
> libc-2.17.so       00002B6FF5641387  gsignal               Unknown  
> Unknown
> libc-2.17.so       00002B6FF5642A78  abort                 Unknown  
> Unknown
> graph.popt         000000000044E345  base_hooks_mp_cp_          78  
> base_hooks.F
> graph.popt         000000000040D97E  MAIN__                    198  graph.F
>
> graph.popt         000000000040CA62  Unknown               Unknown  Unknown
> libc-2.17.so       00002B6FF562D555  __libc_start_main     Unknown  
> Unknown
> graph.popt         000000000040C969  Unknown               Unknown  Unknown
> Aborted (core dumped)
>
> Maybe you can give me some advice about the graph usage. Thanks in advance!
>
> Regards
> JHu
>
> On Tuesday, November 2, 2021 at 10:05:10 PM UTC+8 Marcella Iannuzzi wrote:
>
>>
>> Dear JHu
>>
>> Each minimum on the FES corresponds to specific values of the CVs, as you 
>> defined them. Hence, any structure for which the CVs take those values 
>>  belongs to the minimum. There is no one specific set of coordinates. 
>>
>> I expect that the points among which the path is calculated should be 
>> taken from the generated FES and not from the restart file. 
>>
>> Kind regards
>> Marcella
>>
>>
>>  
>> On Tuesday, November 2, 2021 at 2:13:34 PM UTC+1 hujin... at gmail.com 
>> wrote:
>>
>>>
>>> Dear Marcella
>>>
>>> Really thanks for your reply. So, as for the first problem, do you know 
>>> how to locate the exact minimum of FE and find the corresponding structure? 
>>> Sorry that I have no idea about it.
>>>
>>> As for the second, since the FES is reconstructed, the data of 
>>> "-point-a" should belong to the *.restart file rather than the values I got 
>>> from the first step 'graph.popt -ndim 2 -ndw 1 2 -file *.restart 
>>> -find-minima -cp2k' , right?
>>>
>>> Regards
>>> JHu
>>>
>>> On Tuesday, November 2, 2021 at 4:35:49 PM UTC+8 Marcella Iannuzzi wrote:
>>>
>>>>
>>>> Dear JHu
>>>>
>>>> It is not surprising that you do not find exactly the same CV values in 
>>>> the *-COLVAR.metadynLog.
>>>> The FES generated by graph is reconstructed from the metadynamics 
>>>> potential, made of Gaussian contributions centred on the points where the 
>>>> trajectory is found when the Gaussian is spawn. 
>>>>
>>>> The second issue is related to an error in reading the input.
>>>>
>>>> Regards
>>>> Marcella
>>>>
>>>>
>>>>
>>>> On Tuesday, November 2, 2021 at 3:48:01 AM UTC+1 hujin... at gmail.com 
>>>> wrote:
>>>>
>>>>> Dear cp2k users and developers
>>>>>
>>>>> I'm learning the metadynamic recently and met the following problems 
>>>>> about postprocessing.
>>>>>
>>>>> The first one is that in the postprocessing of constructing the free 
>>>>> energy surface, I used 'graph.popt -ndim 2 -ndw 1 2 -file *.restart 
>>>>> -find-minima -cp2k' to get the minimum point on the free energy surface. I 
>>>>> got  three sets of data which include two CV values ​​and corresponding 
>>>>> free energy values, respectively. But awkwardly, I cannot find  the 
>>>>> corresponding CV value in the original *-COLVAR.metadynLog file and thus 
>>>>> fail to find the specific structural configuration . So, I want to know how 
>>>>> to find the structural configuration  corresponding to the minimum point on 
>>>>> the free energy surface.
>>>>>
>>>>> The second question is that I tried to use 'graph.popt -ndim 2 -ndw 1 
>>>>> 2 -file *.restart -find path -point-a CVs -point-b CVs -cp2k' to get the 
>>>>> path between the minimums, but the command lines went wrong as follows:
>>>>>
>>>>> forrtl: severe (59): list-directed I/O syntax error, unit -5, file 
>>>>> Internal List-Directed Read
>>>>> Image              PC                Routine            Line     
>>>>>    Source             
>>>>> graph.popt         00000000004F55FB  Unknown              
>>>>>  Unknown  Unknown
>>>>> graph.popt         000000000051C5B1  Unknown              
>>>>>  Unknown  Unknown
>>>>> graph.popt         000000000051B2D5  Unknown              
>>>>>  Unknown  Unknown
>>>>> graph.popt         000000000040E935  MAIN__                 
>>>>>    330  graph.F
>>>>> graph.popt         000000000040CA62  Unknown              
>>>>>  Unknown  Unknown
>>>>> libc-2.17.so       00002B42666F4555  __libc_start_main  
>>>>>    Unknown  Unknown
>>>>> graph.popt         000000000040C969  Unknown              
>>>>>  Unknown  Unknow
>>>>>
>>>>> I have no idea about this problem. If someone can offer some help on 
>>>>> any question between these two, I would be really appreciate!
>>>>>
>>>>> Regards!
>>>>> JHu
>>>>>
>>>>

-- 
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/9a3e57be-15f9-497e-9534-334ee3af5edan%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20211103/600b266b/attachment.htm>