2

We submitted a research paper in physics. In this paper, we take structural parameters from the experimental work and then perform simulations using density functional theory and present our results. No one has performed similar calculations or experiments before through which I could validate my results. One of the reviewer’s comments is:

Main problem of this work is a lack of the experimental data that could validate relevant calculations and which would be very important especially in actual case, at least from my point of view. Why?

My question is how should I respond to this comment?

7
  • 3
    I agree that the question is mostly for Academia SE. However, part of it has some overlap with a genuine Physics question: "How to validate predictions from computational Physics?", which is very close to the question "How can we trust experimental results in Physics?". I think that everybody would agree that this last question is perfectly inside the perimeter of Physics SE. So, also the former. In your case, I think that the reviewer is not asking you to perform experiments. He/she is simply asking why you have not used experimental data and, if the reason is they do not exist, why.
    – GiorgioP
    Sep 23 '20 at 15:44
  • What validation work have you done? I understand that the novel parts of your paper can't be directly compared to current experiments, but are there other results of your calculations that can be checked?
    – Anyon
    Sep 23 '20 at 16:09
  • Is simulation possible? Can you simulate based on your theory and show similarity to what has been measured previously?
    – Bryan Krause
    Sep 23 '20 at 16:38
  • we take the structural parameters from the experimental work and then perform simulations and present our results. Sep 23 '20 at 16:40
  • 1
    If you have clarifying information, please do not post it in a comment, but edit your question. I edited in some information (please check whether this is correct), but I fail to make sense of: “The structural parameters are present there but not the rest of the calculations like electronic, optical properties etc” What does there refer to? How can you contrast parameters with calculations?
    – Wrzlprmft
    Sep 23 '20 at 17:38
8

Essentially, the reviewer is saying, you cannot/should not propose a new theory in a vacuum. One of two approaches apply:

  • Do you have a new theoretical approach that will explain observations that existing theoretical approaches cannot yet explain? Demonstrate this categorically by taking the experimental results modeled with the existing theories to model with your new theory. It seems that you argue that this is not the case for your theory. So ...

  • Do you have a new theoretical approach that points to new insights not yet measured by any existing experiments? State concisely what experimental measurements will be needed to validate your theory. You infer that this is your case.

Being somewhat aware of DFT and its applications as well as being experimentally inclined, I must say tangentially that I cannot trust that absolutely no experimental results (e.g. from spectroscopy) exist anywhere that can in some way validate your new theory, even in part. Alternatively, I cannot trust that your theory has absolutely no relevance to direct new experimental approaches in order to validate it (e.g. again spectroscopy).

In summary ... Einstein's Nobel prize was for a new theory that explained oddities in experimental observations (the photoelectric effect). Einstein's theories otherwise were put forward with proposals for experiments that might test their validity. You are being asked to do one or the other as well.

4

Density functional theory has its shortcomings, most of which are well known (bandgap issue, delocalization, etc.) and thoroughly discussed in the literature. If you stay clear of these traps, plus show that your calculations are converged (k points, basis set size, etc.), then there is no reason why you could not make your case: the theory has been extensively validated for many materials. However, if you touch one of the soft spots of the method (dispersion, charged defects, whatever) then you need to provide higher level of theory calculations (from simple corrections all the way to diagrammatic methods and quantum Monte Carlo). An expert reviewer would be aware of the limitations and point you in the right direction. However, if the reviewer is an expert regarding the material class in question but has no experience with density functional theory in particular (e.g., because s/he is an experimentalist) then it is your responsibility to argue why the level of theory is enough, based on examples and evidence from the very extensive literature. And remember that if the referee raised the question, it is likely that a reader of your paper will too, so you need to consider whether these justifications should also make it to the manuscript, not only the response letter.

0

I would probably respond that, while the note is a good idea for future work, the current work rests on foundations. Some validation of theoretical work only comes years after first proposed. A lot of Einstein's work comes easily to mind. Also Stephen Hawking and anything to do with the Grand Unified Theory. Or even gravity, I suppose.

2
  • 4
    It's probably right in principle, but not a good idea to respond with Einstein/Hawking etc., or OP could end up on the "crackpot index". Better to explain why it is ok to have a purely computational model at this stage (e.g. formulate hypothesis for an actual experiment, simple model for a complicated phenomenon etc. - note it's tricky to argue for computational models if they are very complicated, so much of this argument relies on the fact that the model should have few parameters) Sep 23 '20 at 17:37
  • 4
    @CaptainEmacs, I just use those as examples to explain the context. I agree it would probably be a mistake to say that in a reply. "Good idea for future work" is enough.
    – Buffy
    Sep 23 '20 at 17:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.