Just wanted to jot down an idea for speeding up RMG runs using QMTP (implementation and testing of this is currently low on my priority list). For certain problematic molecules where QMTP does not succeed using existing parameters, a large amount of time may be spent going through the ~46 QMTP attempts repeatedly as the molecule is re-encountered during the mechanism generation run (or a separate mechanism generation run, for that matter). In order to prevent these repeated attempts from happening, we could create a (empty) file [InChIKey].failure (or [InChIKey].mopacfailure and [InChIKey.g03failure]) when all existing attempts fail. Once this file is encountered in subsequent runs, it could be serve as a cue to skip the QMTP attempts and go straight to the group-additivity fallback. Granted, if the library of attempts is updated, this could prove to be problematic (by using group additivity unnecessarily) unless all the .failure files in the QMfiles folder are deleted.
Just wanted to jot down an idea for speeding up RMG runs using QMTP (implementation and testing of this is currently low on my priority list). For certain problematic molecules where QMTP does not succeed using existing parameters, a large amount of time may be spent going through the ~46 QMTP attempts repeatedly as the molecule is re-encountered during the mechanism generation run (or a separate mechanism generation run, for that matter). In order to prevent these repeated attempts from happening, we could create a (empty) file [InChIKey].failure (or [InChIKey].mopacfailure and [InChIKey.g03failure]) when all existing attempts fail. Once this file is encountered in subsequent runs, it could be serve as a cue to skip the QMTP attempts and go straight to the group-additivity fallback. Granted, if the library of attempts is updated, this could prove to be problematic (by using group additivity unnecessarily) unless all the .failure files in the QMfiles folder are deleted.