The correlation between resistance change and vcore should indeed be linear, as there are only resistances in the feedback loop, AFAIK, and resistances are linear parts.
The only thing non-linear is the resistance/turn curve of a potentiometer, which gets highly non-linear in the lower reistance regions.