In many practical statistical situations, it is desirable to restrict the flexibility of nonparametric regression models to accommodate prior information. We propose an estimator for regression models with a smoothness penalty and constraints imposed by the nature of the problem. Our estimator...