[CP2K-user] [CP2K:19631] normal mode analysis on an ab initio trajectory
Luca
lcbllcc at gmail.com
Tue Dec 5 12:30:17 UTC 2023
I am currently developing a procedure to resolve complex vibrational
spectra. Although the procedure is not yet freely available, and the
related paper is currently under review, I can provide you with the
preprint of the article. Additionally, I am willing to apply the software
to resolve your computed spectra. I am looking for some applications.
The procedure requires Cartesian coordinates and velocities from a cp2k
molecular dynamics (MD) simulation. The vibrational density of states
(VDOS) is calculated from the Fourier transform of the velocity
autocorrelation function, obtained by processing ab initio (DFT) molecular
dynamics simulations.
Within this framework, I have developed a computational tool designed to
assign the vibrational mode associated with a specific frequency. The
method involves projecting velocities along a specific set of internal
coordinates, such as stretching or bending, to calculate a partial VDOS
(pVDOS). This partial VDOS accounts only for the vibrations associated with
the selected set of internal coordinates. It aids in the interpretation of
the total VDOS and the experimental spectra in a meaningful way.
Cheers,
Luca
Il giorno martedì 5 dicembre 2023 alle 12:41:00 UTC+1 Matthew Graneri ha
scritto:
> Hi Lucas,
>
> Sorry for resurrecting this issue, but I'm in a similar situation to Emma,
> trying to run the normal coordinate analysis through TRAVIS and am
> absolutely stuck. I get up to the question asking for the reference
> structures and, when I specify the file containing my reference geometry, I
> get told: 'Some atoms are not connected to this molecule // Molecule
> recognition failed'.
>
> My AIMD simulation contains several different molecular species, and I am
> trying to work out which peak corresponds to which vibration for each
> molecule in the system. I've tried using a normal .xyz file containing only
> the molecule I want TRAVIS to analyse, and I've tried using an .xyz file
> containing the entire system from the simulation, but nothing has worked.
>
> The only thing I can think of that might be an issue is that I'm using the
> .emp files from the IR spectrum simulation as the input for the analysis. I
> can't think of what other file TRAVIS would like, though...
>
> If you (or anyone else) has any idea why the analysis might not be
> working, I would very much appreciate any advice you can provide!
>
> Regards,
>
> Matthew
>
> On Monday, April 10, 2023 at 4:26:03 AM UTC+8 Francisco Gámez wrote:
>
>> Dear all
>> What I frequently found with the NMA is that I the "normal modes
>> coordinates" in the molden file after TRAVIS evaluation might be Ok and the
>> IR spectra is remarkably good, but the visualization of the NM in MOLDEN
>> shows that the assignments are disparate and make no sense from a chemical
>> point of view and, obviously, do not match with DFT calculations. Does
>> somebody has some tips to get correct mode assignment from the BOMD
>> trajectory?
>> Kind regards
>>
>> El sáb., 8 abr. 2023 0:43, Lucas Lodeiro <eluni... at gmail.com> escribió:
>>
>>> Dear Emma,
>>>
>>> AFAIK compute the IR spectra through dipole (Berry phase), Wannier or
>>> Voronoi integration must result in almost the same spectra, I remember Dr.
>>> Brehm shows the unique difference between them is the relative intensity
>>> between peaks.and Wannier could be problematic sometimes when there is high
>>> symmetry. So there is no problem. BTW, you can compute the properties of
>>> your trajectory using the REFTRAJ option in the MD.
>>>
>>> About the IR spectrum on TRAVIS, it can be computed with Voronoi
>>> integrated electron density (which currently can be done directly in CP2K
>>> on the fly), with fixed charges or fluctuating charges calculated along the
>>> AIMD. I used the last one with self-consistent Hirschfeld charges and it
>>> worked very well.
>>>
>>> But the point is, with just your trajectory you can compute the power
>>> spectrum which can show you some "missing" signals in IR (maybe), and the
>>> NMA without problem. The other spectral properties need electron densities
>>> and more.
>>> In the program, the NMA is called NC (normal coordinate analysis) . When
>>> the "List of functions" is displayed, in the Spectroscopic Functions
>>> section there is "nc". This is what you need. Then you need give some
>>> information and feed with the conrformer's structures (previously optimized
>>> at the same level).
>>> I suggest computing first the power spectra with TRAVIS and
>>> "optimize/play" with the variables and become familiar with them (the
>>> defaults are good, but sometimes a little change can upgrade the spectrum),
>>> because for the "nc" the same procedure is done + the reference structures.
>>>
>>> Finally you will have something like this:
>>> Mode Integral (K) Center (cm^-1)
>>> ----------------------------------------------
>>> 1 -0.112446 2735.82
>>> 2 -0.0757895 1377.51
>>> 3 -0.0508332 529.47
>>> 4 -0.00656595 615.81
>>> 5 0.0120252 542.87
>>> 6 0.177628 1415.45
>>> ----------------------------------------------
>>> 7 327.5 528.37
>>> 8 335.364 535.27
>>> 9 338.557 561.62
>>> ...
>>> 23 319.248 2958.93
>>> 24 330.046 3163.23
>>> ----------------------------------------------
>>>
>>> Which list the modes frequencies and their thermalization... The mode
>>> temperature must be similar to the AIMD mean temperature (the first six
>>> must be near zero)... and a molden file is printed with the normal modes
>>> for each reference structure where you can see them. The normal mode
>>> frequencies must be similar to the peaks in the power spectrum.
>>> In my experience when a time step over 1fs is used the nc starts to
>>> fail... also if you have the velocities "trajectory" you can use it (with
>>> -vel option) to obtain a better velocity autocorrelation function when the
>>> time step is equal or bigger than 0.5fs.
>>>
>>> If you have doubts or need help do not hesitate to write to me, I am not
>>> an expert but I can help or learn on the way :P.
>>>
>>> Regards - Lucas
>>>
>>> El vie, 7 abr 2023 a las 6:59, Emma Rossi (<emma.r... at studenti.unipd.it>)
>>> escribió:
>>>
>>>> Dear Lucas,
>>>>
>>>> Thank you very much for your suggestion. I obtained the IR spectrum
>>>> through the dipole moment of the whole box computed at each step of the
>>>> trajectory by CP2K. Unfortunately, I couldn't use TRAVIS since I had not
>>>> printed the electron density during the simulation.
>>>>
>>>> However, it would be great if I could use TRAVIS as post processing
>>>> tool to perform NMA on the trajectory. As far as I have seen in the
>>>> documentation, the online tutorials/presentations deal only with the
>>>> simulation of the spectra, which, indeed, require electron density. I
>>>> couldn't find any instruction about how to perform the NMA (I've found just
>>>> the original paper).
>>>>
>>>> However, if you have expertise in TRAVIS usage, I would be very
>>>> grateful if you could answer a doubt about TRAVIS NMA tool.
>>>>
>>>> Can I perform NMA on my trajectory without performing as a preliminary
>>>> step the calculation of the IR spectrum? As I mentioned above, I have not
>>>> the electron density.
>>>>
>>>> Thank you very much for your kind support.
>>>>
>>>> Best regards,
>>>> Emma Rossi
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Il giorno giovedì 6 aprile 2023 alle 23:59:45 UTC+2 Lucas Lodeiro ha
>>>> scritto:
>>>>
>>>>> Hi Emma,
>>>>>
>>>>> I do not know how you managed to obtain the IR spectrum from your
>>>>> trajectory, but I suggest the TRAVIS program to do it:
>>>>> http://www.travis-analyzer.de/
>>>>> There is many ways to accomplish it, and the results are very good.
>>>>> Also, there is the possibility to obtain the "normal modes" of the
>>>>> trajectory which are consistent with the trajectory IR/Power spectrum.
>>>>> There are some tutorials, and the program is self explaining:
>>>>> https://brehm-research.de/spectroscopy.php
>>>>>
>>>>> In the normal modes section you can feed the program with one or more
>>>>> minimum energy conformer where the trajectory is projected, and
>>>>> approximated normal modes are obtained.
>>>>>
>>>>> Regards - Lucas
>>>>>
>>>>> El jue, 6 abr 2023 a las 14:21, Emma Rossi (<
>>>>> emma.r... at studenti.unipd.it>) escribió:
>>>>>
>>>>>> Dear all,
>>>>>>
>>>>>> I would need to assign the bands of an IR spectrum, obtained through
>>>>>> AIMD simulation.
>>>>>>
>>>>>> I would like to perform a normal mode analysis (NMA) on the
>>>>>> trajectory I have already produced to properly take into account solvation
>>>>>> effects as well as the correct weight of the sampled conformers.
>>>>>>
>>>>>> I guess I could do so retracing the traj.xyz and using the
>>>>>> vibrational analysis section in the input, but I would be very greatful if
>>>>>> you could clarify a few issues about this strategy:
>>>>>>
>>>>>> 1) is it the most efficient and correct way to do NMA over a
>>>>>> trajectory or do you suggest any other ways?
>>>>>>
>>>>>> 2) how the output with eigenfrequencies and eigenvectors should look
>>>>>> like? I can figure the typical output for NMA on single configurations, but
>>>>>> I have no ideas of what should I expect for NMA on trajectories (I suppose
>>>>>> that the hessian will be diagonalized at each step, but how these info will
>>>>>> be put together at the end of the process? )
>>>>>>
>>>>>> I have read the original paper of CP2K and the pdf presentations
>>>>>> available online, but I couldn't fix my problem.
>>>>>>
>>>>>> I am very willing to understand and learn from anyone who will help.
>>>>>>
>>>>>> Thank you very much in advance for your support.
>>>>>>
>>>>>> Best regards,
>>>>>> Emma Rossi
>>>>>>
>>>>>> --
>>>>>>
>>>>> 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/6edca95f-7fb7-4a27-97cd-5868c9ae611en%40googlegroups.com
>>>>>> <https://groups.google.com/d/msgid/cp2k/6edca95f-7fb7-4a27-97cd-5868c9ae611en%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+uns... at googlegroups.com.
>>>> To view this discussion on the web visit
>>>> https://groups.google.com/d/msgid/cp2k/6bc02dbb-0824-46a7-8edc-8d98c5c87f21n%40googlegroups.com
>>>> <https://groups.google.com/d/msgid/cp2k/6bc02dbb-0824-46a7-8edc-8d98c5c87f21n%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+uns... at googlegroups.com.
>>>
>> To view this discussion on the web visit
>>> https://groups.google.com/d/msgid/cp2k/CAOFT4PLFUJ3KjuE%2BstwXiCONooLJx%3DCBBYA_1UuXM7cFL%2B6V%2BA%40mail.gmail.com
>>> <https://groups.google.com/d/msgid/cp2k/CAOFT4PLFUJ3KjuE%2BstwXiCONooLJx%3DCBBYA_1UuXM7cFL%2B6V%2BA%40mail.gmail.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/fc079dcd-703c-4831-930c-747cfc998e02n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20231205/2366758b/attachment-0001.htm>
More information about the CP2K-user
mailing list