This PR adds an extra step to the mixture data training set construction - the common mixture data is now filtered to a more limited range before selecting the state and compositions of the data to include.
This was found to improve the selection algorithms ability to select data closer to the states / compositions of interest.
Description
This PR adds an extra step to the mixture data training set construction - the common mixture data is now filtered to a more limited range before selecting the state and compositions of the data to include.
This was found to improve the selection algorithms ability to select data closer to the states / compositions of interest.
Status