This page serves as a repository for stuff I’ve created that I think others might find useful.  Please get in touch if you try to use something linked here and it doesn’t work.

Convergent GLA

This Python script provides a implementation of Magri’s (2012) convergent GLA.  It prints out (i) a ranking trace, (ii) the trial at which the last error occurred, and (iii) the final weights of all constraints, in descending order.  It also writes to file a log of what happens at each trial.  Input file format mostly follows the OTSoft format, with the following differences: (i) the second row must contain an initial weight for each constraint, and (ii) each input candidate may have one and only one optimal candidate, and that optimal candidate must be assigned a frequency of >= 0.  If that’s unclear, maybe this toy input file will make it clearer!

(more coming soon!)