Relative file paths caused difficulty debugging Arkane issue

[sevyharris]

Bug Description

I ran an Arkane kinetics job that kept giving me this error:
The species in reaction HO2(16) + C[CH]CC(23) <=> O2(2) + butane(1) do not have adequate information for TST, using default kinetics values.

I kept getting zero for the reaction rate, and it turned out that the problem was actually how I was structuring/naming my input files. It was using the transition state's energy file in place of the species' energy files because I had wrongly assumed that Arkane would check for the species's energy file relative to the directory in which that species's input file is defined.

To give more of a visual, I had this directory structure:

my_reaction
    input.py
    TS.py <-------------------------- defines TS energy file as 'energy.out'
    energy.out
    species_1
        species_1.py <--------------- defines species 1 energy file as 'energy.out' but ends up grabbing the TS energy.out
        energy.out
    species_2
        species_2.py <--------------- defines species 2 energy file as 'energy.out' but ends up grabbing the TS energy.out
        energy.out

When it needed to be like this:

my_reaction
    input.py
    TS
        TS.py <-------------------------- defines TS energy file as 'energy.out'
        energy.out
    species_1
        species_1.py <--------------- defines species 1 energy file as 'energy.out' and reads the correct energy file
        energy.out
    species_2
        species_2.py <--------------- defines species 2 energy file as 'energy.out' and reads the correct energy file
        energy.out

How To Reproduce

This one's a little tricky to reproduce exactly. You get different error messages depending on how much replacing the species e_elec with the TS e_elec changes the reaction e0. For example, I tried using the reactions/rate_demo example rearranged to the first directory structure shown above and it actually produces no official errors, but gives you absurd kinetics with an A factor of 10^34 cm^3/mol/s. You can find that example here: https://github.com/sevyharris/rmg_tools/tree/main/arkane/relative_paths_error

Expected Behavior

I don't know if this strictly counts as a bug. I think I was particularly creative with my file structure and hopefully most users will have more sense than I did and realize that the TS deserves its own directory.

But also I do think it makes sense for paths to be read in as relative to the file in which they are defined. I'll see if I can find an appropriate place in the code to suggest a change. The answer may be a warning in the documentation not to structure your directories the way I did, or to not name files in the same folder as input.py using the same basename as something else in a deeper folder.

Or maybe just making this issue is enough of a papertrail that the next person to have this problem (most likely me again in 6 months' time) will have a much easier time debugging.

Installation Information

Describe your installation method and system information.

• OS (include version if known): WSL: Ubuntu 20.04 LTS
• Installation method: source
• RMG version information: 3.2.0-515-g2a1f91003
  • RMG-Py: 2a1f910
  • RMG-database: 0add9cdee2dd2e0677c765086dbcd471d1377cb4