Simon Boothroyd shared a script to analyze how the molecule populations distribute among (or move) the gradients for each of the parameters based on the first iteration, to find out whether we see multimodal distributions indicating a split of parameters.
Script is:
Here are the plots for fit7 for the TIG parameters, k1_bondorder1 and k1_bondorder2 for each parameter
Fit7 FF file:
TIG0:
.png?version=1&modificationDate=1612285260513&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285260698&cacheVersion=1&api=v2)
TIG1c
.png?version=1&modificationDate=1612285291609&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285292421&cacheVersion=1&api=v2)
TIG1d
.png?version=1&modificationDate=1612285308602&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285309102&cacheVersion=1&api=v2)
TIG2
.png?version=1&modificationDate=1612285324996&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285325699&cacheVersion=1&api=v2)
TIG3
.png?version=1&modificationDate=1612285341004&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285341153&cacheVersion=1&api=v2)
TIG4
.png?version=1&modificationDate=1612285356606&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285357266&cacheVersion=1&api=v2)
TIG5a
.png?version=1&modificationDate=1612285371430&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285371551&cacheVersion=1&api=v2)
TIG5b
.png?version=1&modificationDate=1612285388244&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285388806&cacheVersion=1&api=v2)
TIG6
.png?version=1&modificationDate=1612285429134&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285429368&cacheVersion=1&api=v2)
TIG7
.png?version=1&modificationDate=1612285445052&cacheVersion=1&api=v2)
.png?version=1&modificationDate=1612285445455&cacheVersion=1&api=v2)
TIG8
TIG5a, TIG6, TIG7, TIG8 have gradients that vary on a large scale here.