Is there a way to fix the lower asymptote (parameter c) at 0 in the Logistic 4p Hill model in Fit Curve? Basically, I want the 3-parameter version of the Logistic hill model.