Bug: Prism can freeze when fitting a model with shared parameters which don't have initial values defined.

This page explains a bug in  5.0b which can lead to Prism freezing during a nonlinear regression curve fit. 

It happens when you create a user-defined model, and don't define rules for initial values for all parameters. You then fit the model, constraining one or more parameters to a constant value. Since the value is constant, there is no need for an initial value. Assuming the model and constants are reasonable, the fit works fine. 

Now you decide not to hold the parameter to a constant value, but instead share it, so Prism uses global nonlinear regression to fit one value for all data sets. Since there is no initial value defined, Prism can't even begin to fit the model. It tries anyway, and can freeze. It should, of course, give a useful message and we'll fix this in 5.0c. 

To work around this problem, make sure that you have created a rule for initial values for the parameter before deciding to share its value. 

To assign the rule for initial values:

1. Select custom equation on "Fit" tab of "Parameters:Nonlinear Regression" dialog.
2. Press "View/Edit equation" button ("Gear" button on Mac) to edit equation. 
3. Go to the "Rules for Initial Values" tab.
4. Specify a rule for initial values. The rule can be just to set a specified value.