Wavenumber bug for certain ranges, and smaller issues and suggestions

[gabrielastro]

First of all, a big thank you for providing this excellent tool, so well made! This is excellent even for a casual user who would like some opacity information but without having to download huge amounts of data and/or go through installation processes (often with incompatibilities, issues with paths in set-up files,etc.). And thanks for the high resolution!

I have come across a few issues using https://marfa.app:

• If inputting integers for the minimum and maximum wavenumber that are not multiples of 10 cm^-1, the output spectrum in Output.dat is shifted in wavelength (i.e. the shape is correct but the nu values are wrong, so of course the curve is effectively), which is a big problem because everything looks fine otherwise. It took me some time to figure out what was going on.
• For some ranges (e.g.: from 2445 to 2475 cm^-1), one gets again Error occurred while running MARFA executable.
• Selecting SO2 or HNO3 leads to Error occurred while running MARFA executable with the following settings:

2440 — 2490
125
100 or 999 K
0.1 or 1 bar
absorption cross-section

• Choosing T = 1000 K leads to ****** in info.txt but 999 K is ok :)
• For NO, the absorption is nan at least in the range 2465 to 2470 cm^-1. I would understand 0 but nan seems strange.

Here are some suggestions for web program:

• Allow searching with numbers (e.g. "co2" or "CO2"); currently, the drop-down menu matches only on the letters (typing "co" is enough to restrict the choices) but not numbers. (It is very pretty and helpful that you have subscripts, though!) The list is not so long currently so this is not major, though.
• This is small but is there some rounding issue in the wavenumbers? In output.dat, one finds e.g.:

     2440.00000     4.3218765e-25
     2440.00049     4.3467498e-25
     2440.00098     4.3716216e-25
     2440.00146     4.3964939e-25
…

instead of 2440.0005, 2440.0010, etc..

• In info.txt, add the units of the cut-off. By the way, what do you think of the recommendation of Gharib-Nehzad et al. 2024 to use 25 cm^-1 as a default value for the cut-off?
• When looking for molecule not available, the message that the option is not choosable is given на русском языке 😉.
• In info.txt and on the download summary page, the units of "Absorption cross-section" are given as cm² but it should be cm²/mol
• Related to #9, it would be good even to provide the option of not downloading the binary file (especially while that issue is not resolved and the range there is too long). If one is downloading a small region in the IR (e.g. at 4 µm), the Output.dat file is ~600 kB but the .ptbin file is ~20 MB, which takes up too much space.
• It would be convenient for quick plotting or reading-in if the header part of Output.dat were commented out with #.
• Especially e.g. for water, I am looking forward to seeing other datasets to get a sense of the “theoretical errorbars”.

About your paper (https://arxiv.org/abs/2411.03418)[https://arxiv.org/abs/2411.03418]:

• Have you thought about publishing your paper in JOSS? (Tiny note: 5*10E-4 →5e-4 cm^-1 in the abstract.) I do not know the journal well but it looks as though it could be a good venue. Also, ASCL would be good.
• You could also compare MARFA to other on-line services, e.g.: DACE, ExoMolHR, etc. to help users. (Sorry for my ignorance if this is already done in another paper!)

But once again, congratulations on an excellent service and a really well-made website (lightweight, easily legible, numbers stay in the fields, etc.)!

[Razumovskyy]

@gabrielastro Thank you so much for your feedback and sorry for the late reply (Github didn't notify me) !
I am very grateful for your testing, I will carefully look into all points you have made and address them either with fixes or commentaries.

Currently the manuscript was submitted to JQSRT and returned for fixing minor corrections. I am making a small comparative analysis with HAPI.

Also, if you want to contribute to solving any of the issues now or in future, you are very welcome ! But of course, I will look close into your points, and get back to you very soon !

[gabrielastro]

@Razumovskyy Thanks for your message! No problem. Good to know that Github does not always notify the authors! I will keep this in mind in the future.

Ah, JQSRT sounds even better; good for you.

I looked again briefly at the list and I think that all these points can be modified only on the server side, but thanks for the offer to contribute!