Fit model parameters may optionally be box constrained for nonlinear minimization (see e.g. NonlinearMinimizerBase.cs). However, a current limitation is that one has to either provide a full set of lower and/or upper bounds, or leave all parameters unbounded (in any direction).

Sometimes it makes sense for some parameters to be bounded, e.g. due to physical constraints, while it doesn't make sense for others. It would be great if it was possible to mix bounded and unbounded parameters. This could, for instance, be implemented by allowing individual bounds to be nullable, and then checking/catching the individual "null-bounds" in the ProjectToInternalParameters method.