D-Optimal Designs for the Mitscherlich Non-Linear Regression Function

Mitscherlich’s function is a well-known three-parameter non-linear regression that quantifies the relation between stimulus or time variable and response. It has many applications, in particular field of measurement reliability. Optimal designs for estimation this have been constructed only normally distributed responses with homoscedastic variances. In paper we generalize literature to D-optimal discrete continuous having their distribution exponential family. We also demonstrate our can be identical different from optimal variance weighted linear regression.