Open mcduffie opened 6 years ago
i believe my documentation is a little ambiguous. with run_LBLRTM_ABSCO.py
, the input .ini file can contain multiple species, and the driver script loops over each molecule. so by "should only be one molecule per run," i mean that one config file can have as many species as necessary, but they are run through the ABSCO computation in series
What I meant was that in practice people would not run all species in a single executable run. In order to speed up computation they would probably spread out each molecule to a different computational node.
ah, ok. looks like this touches upon the parallelization we discuss in https://github.com/pernak18/ABSCO/issues/4#issuecomment-423240712. i don't think it'd be difficult to write a similar driver script but with the ability to farm out the molecules to different nodes instead of looping over the species list
In the README file it mentions under the "molnames" field that there "should only be one molecule per run". If I consider how people might run this in practice then that means people would have a separate config file for each molecule. There would be a lot of duplicated contents between each of these configs. With that context, it would be desirable if config files could include others to reduce the duplication and potential for errors when needing to update multiple files.