[CP2K-user] [CP2K:19043] Re: Plotting Metadynamics Colvar with graph.psmp

[Liam H]

Hi Marcella,

Thank you that makes sense, I see the number of CVs do matter, I've 
replotted using only 2 ndim=ndw and the data looks much better. I can find 
the minima as well using -find-minima. 

I also figured out how to make use of the -find-path option by pasting the 
minima floats obtained from the -find-minima output for -point-a and 
-point-b, so I'll post it here incase anyone else is wondering.

input:

FES|  Mapping Gaussians :: 100 %
FES|  Dumping FES structure in file: < fes.dat >
 Iteration nr.           1  1.504066111949192E-002

graph.psmp -cp2k -file input.restart -ndim 2 -ndw 1 2 -out fes.dat 
-find-minima

output:

FES|  Dumping FES structure in file: < fes.dat >
FES| Trust hyper-radius     10    10 which is equivalent to:     0.191530   
 0.483382
FES| Minimum found (    2.727613    2.186932 ). FES value =    -0.001262 
Hartree. Number of Iter:    102
FES| Minimum found (    3.017643    2.627451 ). FES value =    -0.019981 
Hartree. Number of Iter:    101


then taking 2 relevant minima and pasting one after another with exact same 
digits

graph.psmp -cp2k -file input.restart -ndim 2 -ndw 1 2 -out fes.dat 
-find-path -point-a  2.727613    2.186932 -point-b 3.017643    2.627451

FES|  Mapping Gaussians :: 100 %
FES|  Dumping FES structure in file: < fes.dat >
 Iteration nr.           1  1.504066111949192E-002
...
..
 MEP saved on <mep.data> file.

FES| NORMAL FES TERMINATION.

Best regards,
Liam

On Wednesday, June 28, 2023 at 6:02:03 AM UTC-2:30 Marcella Iannuzzi wrote:

>
> Dear ...
>
> The search of minima using "graph" is possible only considering the whole 
> metadynamics space, no for projections of it to lower dimension.
> ndw=ndim  and ndim should be the actual dimension of the space, otherwise 
> whatever you get is wrong.
> Regards
> Marcella
>
>
> On Wednesday, June 28, 2023 at 1:44:11 AM UTC+2 Liam H wrote:
>
>> More importantly though, since I do have the minima, I'd like to figure 
>> how to use the -find-path option.
>>
>> graph.psmp -cp2k -file input.restart -find-path -ndim 2 -ndw 1  2  -out 
>> fes.dat
>>
>> gives the output:
>>  ABORT in metadyn_tools/graph.F:362 -find-path requires the specification 
>> of -po
>>  int-a and -point-b !
>>
>> So i've attempted a few combinations
>> graph.psmp -cp2k -file input.restart -find-path -point-a -point-b -ndim 2 
>> -ndw 1  2  -out fes.dat
>>
>> But I haven't been able to find any examples of the particular format for 
>> specifying -point-a and -point-b
>>
>> ive tried in the format of the COLVAR , specifying XY -point-a  2.070, 
>>  6.416 -point-b 0.248,  3.076
>> ive tried XYZ, ive tried without comma, ive tried line number
>>
>> If anyone has any information on the format of the -find-path -point-a 
>> -point-b function it would be very much appreciated!
>>
>>
>> On Tuesday, June 27, 2023 at 4:15:45 PM UTC-2:30 Liam H wrote:
>>
>>> Yes that's right although it seems to work even though I sampled 6 it 
>>> will work with  -ndim 2 or  -ndim 3,
>>>
>>> However, the issue i'm having now.. when using the -find-minima command:
>>>
>>> Also this example is with only 3 CV to simplify, still the same issue 
>>> with 6.
>>>
>>> it works perfectly when using 
>>>
>>> graph.psmp -cp2k -file input.restart -find-minima -ndim 2 -ndw 1  2  
>>> -out fes.dat
>>>
>>> however if I try the same thing even with the same two CV with -ndim 3, 
>>> it fails for both -ndw 1 2 and -ndw 1 3
>>> Here is an example with the output.
>>>
>>> The data gets plotted but the -find-minima fails, it works when i use 
>>> -ndim 2 , it finds 9 minima, but in this case then i cannot access CV3 -ndw 
>>> 1 3
>>>
>>>
>>> graph.psmp -cp2k -file input.restart -find-minima -ndim 3 -ndw 1  2  
>>> -out fes.dat
>>>
>>> FES|  Parsing file:   <input.restart>
>>>
>>> **********************************************************************
>>> FES|  Parameters for FES:
>>>
>>> FES|  NDIM         ::           3
>>> FES|  NWD          ::           2
>>> FES|  HILLS        ::         120
>>> FES|  COLVAR #   1 ::     (  2.070 ,  6.416)               (   MAPPED)
>>> FES|  COLVAR #   3 ::     (  0.248 ,  3.076)               (   MAPPED)
>>> FES|  COLVAR #   2 ::     (  1.784 ,  3.636)               (NO MAPPED)
>>> FES|  NGRID        ::         100    100
>>> FES|  DX           ::       0.044  0.029
>>> FES|  CUTOFF       ::     .10000E-05
>>> **********************************************************************
>>>
>>>
>>> FES|  Computing Free Energy Surface
>>> FES|  Mapping Gaussians ::  10 %
>>> FES|  Mapping Gaussians ::  20 %
>>> FES|  Mapping Gaussians ::  30 %
>>> FES|  Mapping Gaussians ::  40 %
>>> FES|  Mapping Gaussians ::  50 %
>>> FES|  Mapping Gaussians ::  70 %
>>> FES|  Mapping Gaussians ::  80 %
>>> FES|  Mapping Gaussians ::  90 %
>>> FES|  Mapping Gaussians :: 100 %
>>> FES|  Mapping Gaussians :: 100 %
>>> FES|  Dumping FES structure in file: < FES10.dat >
>>>  ABORT in metadyn_tools/graph_methods.F:279 Not implemented for 
>>> projected FES!
>>> forrtl: error (76): Abort trap signal
>>>
>>> On Tuesday, June 27, 2023 at 1:55:32 PM UTC-2:30 Marcella Iannuzzi wrote:
>>>
>>>>
>>>> Hello,
>>>>
>>>> Both are wrong, the dimension of the sampled space is 6 therefore -ndim 
>>>> 6 is the correct input.
>>>> Regards
>>>> Marcella
>>>>
>>>> On Tuesday, June 27, 2023 at 6:01:50 PM UTC+2 Liam H wrote:
>>>>
>>>>> My concern now is that the plot I get using
>>>>>
>>>>> graph.psmp -cp2k -file input.restart -ndim 2 -ndw 1  2  -out fes.dat
>>>>>
>>>>> Is not identical when doing -ndim 3 and only mapping the first two CV
>>>>>
>>>>> graph.psmp -cp2k -file input.restart -ndim 3 -ndw 1  2  -out fes.dat
>>>>>
>>>>> I am not sure how this could be, raw data itself looks correct but 
>>>>> when plotting the surface it isn't the same the minimums and saddlepoint 
>>>>> are completely different when plotting the same CV
>>>>>
>>>>> On Tuesday, June 27, 2023 at 1:23:39 PM UTC-2:30 Liam H wrote:
>>>>>
>>>>>> Nevermind I figured it out, by changing the -ndim to a higher number 
>>>>>> , it still allows you to map COLVAR -ndw 1 3
>>>>>>
>>>>>> graph.psmp -cp2k -file input.restart -ndim 2 -ndw 1  3  -out fes.dat 
>>>>>> Doesnt work
>>>>>>
>>>>>> graph.psmp -cp2k -file input.restart -ndim 3 -ndw 1  3  -out fes.dat
>>>>>> works
>>>>>> On Tuesday, June 27, 2023 at 12:21:45 PM UTC-2:30 Liam H wrote:
>>>>>>
>>>>>>> Hello.
>>>>>>>
>>>>>>> I have been running metadynamics simulations with multiple colvar, 6 
>>>>>>> in this case.
>>>>>>>
>>>>>>> This is because I want to be able to plot different surfaces with 
>>>>>>> different colvar, because I can't visualize a 6 dimensional surface.
>>>>>>>
>>>>>>> graph.psmp -cp2k -file input.restart -ndim 2 -ndw 1 2 -out fes.dat
>>>>>>>
>>>>>>> This works fine to give me the 3D surface, (cv1,cv2,energy)
>>>>>>> I would like to plot (cv1,cv3,energy) 
>>>>>>>
>>>>>>> but running the command:
>>>>>>> graph.psmp -cp2k -file input.restart -ndim 2 -ndw 1 3 -out fes.dat
>>>>>>>
>>>>>>> Does not work.
>>>>>>>
>>>>>>> graph.psmp -cp2k -file input.restart -ndim 6 -ndw 1 2 3 4 5 6 -out 
>>>>>>> fes.dat
>>>>>>>
>>>>>>> This works but produces such a massive file its not practical to try 
>>>>>>> and plot the 2d surface..
>>>>>>> Is there any way to specify which cvs to be used in the plot, 
>>>>>>> without plotting them all at once?
>>>>>>>
>>>>>>>
>>>>>>>

-- 
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/e620cc8d-b1c1-40ee-9ace-7dae882b8236n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20230630/b475b4c8/attachment.htm>