Masks refers to the temporary fixing of one or more parameters and optimizing over the rest. This can be very helpful when one parameter is a setting or control that we later wish to optimize. It can also be useful to fix a parameter to which the objective function is particularly sensitive. In this chapter, we explore this idea, noting that some tools allow such options.
We return to our Hobbs weeds example with maximum likelihood estimation of the three logistic parameters plus the dispersion. That is, given a set of values of the growth of some quantity , for example, density of weeds, at times , we wish to maximize the product of terms of the form
(see Section 1.2)
Our parameters to adjust are again the , , and , which here are ...