PyRAMD Scheme: A Protocol for Computing the Infrared Spectra of Polyatomic Molecules Using ab Initio Molecular Dynamics
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe paper proposed a method to simulate the vibrational spectra of polyatomic molecule. It should attract the interest from the physics, chemists and even the biologists. The paper is well organized and written. There are some concerns before it is accepted.
1. It is more useful for readers if the program is given in the revised version.
2. I don’t get the point in Fig.2. Why are the spectra of CH4 deferent at different time scale (0.5-1.5fs)?
3. Did you consider the Franck-Conden factor and transition dipole moment in your simulation?
4. The comparisons between your calculations and Gaussian and/or other software should be made.
Author Response
Comment 1:
It is more useful for readers if the program is given in the revised version.
Answer 1:
The availability of the software from the GitLab repository was stated in the Methods section of the manuscript, and the version used in this work from Zenodo was stated in the Data Availability statement. To make it less spread out in the text, the Data Availability was modified to the following text:
All software at the versions used in this manuscript, simulation and data analysis scripts, and the results are available in the Zenodo repository at DOI: 10.5281/zenodo.12744805. The latest version of the PyRAMD software can be obtained from the corresponding repository (Ref. [29]).
Comment 2:
I don’t get the point in Fig.2. Why are the spectra of CH4 deferent at different time scale (0.5-1.5fs)?
Answer 2:
This is because of the artificial blue shift appearing from the numerical integration error of the Verlet integration and of its equivalents. The origin of this blue shift and the correction are described in detail in Section 3.2 of the manuscript and also in Appendix A, alongside the papers cited in this manuscript.
Comment 3:
Did you consider the Franck-Conden factor and transition dipole moment in your simulation?
Answer 3:
The calculation of the IR spectra from MD relies not on the transition dipole moments but on the fluctuation-dissipation theorem, as described in Section 3 of this manuscript. The Franck–Condon principle applies to situations where the transition between electronic states occurs, such as UV excitation or photoionization. Thus, in this case of IR spectra, the Franck–Condon factors are not applicable.
Comment 4:
The comparisons between your calculations and Gaussian and/or other software should be made.
Answer 4:
The two examples of IR spectra of acetic acid and indene were added to the end of Section 4 (Discussion). A comparison with the scaled harmonic calculations done with ORCA is provided there.
Reviewer 2 Report
Comments and Suggestions for Authors1/ The model is practically applicable to protonated methane (where a considerable deviation from the experimental IR is observed), why not do the same with other species that have well-defined IR spectra?
2/ Can you tell us what are the reasonable limitations and scope of the proposed scheme? You mention that, despite the PyRAMD scheme being inferior to the more direct approaches... So, Is there any other more compelling or interesting reason other than computational time that you hold?
3/ What advantages does PyRAMD-scheme have over other known schemes or methodologies?
Author Response
Comment 1:
The model is practically applicable to protonated methane (where a considerable deviation from the experimental IR is observed), why not do the same with other species that have well-defined IR spectra?
Answer 1:
In the new version of the manuscript, a comparison with the IR spectra of acetic acid and indene was added to the end of Section 4 (Discussion).
Comments 2 and 3:
2) Can you tell us what are the reasonable limitations and scope of the proposed scheme? You mention that, despite the PyRAMD scheme being inferior to the more direct approaches... So, Is there any other more compelling or interesting reason other than computational time that you hold?
3) What advantages does PyRAMD-scheme have over other known schemes or methodologies?
Answer 2:
The main idea of this approach is to be an intermediate stage between the classical MD and fully-quantum PIMD simulations, taking the simplicity and computational cost of the classical MD with the inclusion of the nuclear quantum effects. The scheme combines the unnoticed frequency correction with a fresh idea of SWS and an old idea of tabulated scaling factors adopted from the harmonic calculations. To highlight this, the last paragraph of the conclusions of the manuscript was changed to the following:
We have presented a general approach (a "PyRAMD scheme") for computing the IR spectra of polyatomic systems using the AIMD approach implemented in the PyRAMD software. The approach invokes the following essential components: 1) inclusion of the NQEs by the usage of the SWS approach coupled to the thermostats; 2) frequency correction, allowing for larger time steps in the MD simulations; 3) pre-parametrized scaling factors, similar to those used for the harmonic frequency calculations. The synergistic effect of these components was demonstrated using the test cases of the protonated methane, acetic acid, and indene.
Despite this scheme being theoretically inferior to the more direct and expensive approaches, such as a combination of path integral MD for inclusion of the NQEs and of the smaller time steps in combination with highly accurate methods to obtain the PES, it was demonstrated that the approach proposed here could produce reasonable results. In this sense, the "PyRAMD scheme" can serve as an intermediate stage between classical MD and path integral MD, combining the simplicity and cost efficiency of the first with the inclusion of the NQEs of the second one. Therefore, the presented algorithm can be recommended to provide theoretical counterparts to the experimental measurements.
Reviewer 3 Report
Comments and Suggestions for AuthorsIn the present mns Tikhonov present a general framework for computing the IR anharmonic vibrational spectra of polyatomic molecules using the Born-Oppenheimer molecular dynamics using the software PyRAMD.
The mns is interesting and well written. I suggest that it should be published after a major revision.
Examples of experimental and theoretical MD IR action spectra are given only for CH4 (Fig 4) and CH5+ (Fig 5). I think that additional Figures should be plotted for other more complicated molecules than these two species, so as to show that the PyRAMD scheme is successfully applied for polyatomic molecules more complicated that the simple CH4 (Fig 4) and CH5+ (Fig 5).
The author has calculated the ethane (C2H6), methylamine (CH3NH2), and methanol (CH3OH) (lines 218-219). At least the experimental and theoretical MD IR spectra of these three molecules should be added to the mns.
Minor errors/corrections
1. Before the references a space is needed. See for instances lines 44, 65, 70, 74 etc. This typo occurs in many lines.
2. Given that there is only one author, I suggest that the text maybe rephrased, i.e.,
“we present..” -> it is presented…
“We propose a new criterion…” -> It is proposed a new criterion…
Author Response
Comment 1:
Examples of experimental and theoretical MD IR action spectra are given only for CH4 (Fig 4) and CH5+ (Fig 5). I think that additional Figures should be plotted for other more complicated molecules than these two species, so as to show that the PyRAMD scheme is successfully applied for polyatomic molecules more complicated that the simple CH4 (Fig 4) and CH5+ (Fig 5).
Answer 1:
The end of the Discussion (Section #4) now includes two more illustrative cases, acetic acid and indene.
Comment 2:
The author has calculated the ethane (C2H6), methylamine (CH3NH2), and methanol (CH3OH) (lines 218-219). At least the experimental and theoretical MD IR spectra of these three molecules should be added to the mns.
Answer 2:
These three molecules were also added to Figure 4 of the manuscript at the PBEh-3c level of theory. All the rest of the calculations are available in the accompanying dataset at Zenodo (DOI: 10.5281/zenodo.12744805).
Comment 3:
A space is needed before the references. For instance, see lines 44, 65, 70, 74, etc. This typo occurs in many lines.
Answer 3:
This typo was fixed.
Comment 4:
Given that there is only one author, I suggest that the text maybe rephrased, i.e.,
“we present..” -> it is presented…
“We propose a new criterion…” -> It is proposed a new criterion…
Answer 4:
I deliberately chose the style of "we present" and so on, as I feel that manuscripts written in this manner are easier to read and comprehend. If it is ok, I would prefer to keep this style.
Reviewer 4 Report
Comments and Suggestions for AuthorsThe author describes a procedure to obtain infrared spectra of polyatomic molecules by means of ab initio molecular dynamics. Some applications to illustrate specific issues of the method are made for molecules of up to 6 atoms. I find the manuscript very well written providing a dense and rich description full of numerical details which surely will be of interest for those readers familiar with computational calculations of molecular spectra from first principle techniques. In particular, I have enjoyed both the introduction and the explanation of the general basis for such theoretical calculations.
I openly recommend the publication of the manuscript in its present status, but leaving to minor things for the consideration of the author:
- Figure 1 constitutes a fine example of spectra resulting for the usual FFT method and the here discussed rLSSA approach. As suggested in pg. 3 after Eq. (7) the latter constitutes a smoother representation of the actual dataset under consideration. I wonder if such smoothing procedures are not also possible somehow in the context of the FFT method, thus yielding a probably better comparison as that shown in Figure 1.
- I agree the corrected molecular dynamics spectra shown in Figure 5 are closer to experiment that those uncorrected. However I would suggest that still there is room for further improvement in terms of a comparison between theory and experiment. What would be in the authors' opinion the possible aspects to modify / improve of the present method in this sense?
Author Response
Comment 1:
Figure 1 constitutes a fine example of spectra resulting for the usual FFT method and the here discussed rLSSA approach. As suggested in pg. 3 after Eq. (7) the latter constitutes a smoother representation of the actual dataset under consideration. I wonder if such smoothing procedures are not also possible somehow in the context of the FFT method, thus yielding a probably better comparison as that shown in Figure 1.
Answer 1:
Following the suggestion, a zero-padded FFT was also added to the same figure. The discussion in Section 3.1 was also minorly modified to accompany this change.
Comment 2:
- I agree the corrected molecular dynamics spectra shown in Figure 5 are closer to experiment that those uncorrected. However I would suggest that still there is room for further improvement in terms of a comparison between theory and experiment. What would be in the authors' opinion the possible aspects to modify / improve of the present method in this sense?
Answer 2:
This probably has to do with the quality of the quantum chemical method and the quality of the tabulated scale factor. The latter issue was pointed out in the new version of the manuscript upon the discussion of newly added IR spectra of acetic acid and indene and a comparison of scaled MD-based spectra with scaled harmonic spectra, where the latter seems to perform slightly better.
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsAccepted as it is.
