ConvexDual {ebTobit}

Convex Optimization of the Kiefer-Wolfowitz NPMLE

Description

This method only works if there is a working installation of REBayes available. See the REBayes package and corresponding papers for more implementation details.

Usage

ConvexDual(A, ...)

Arguments

A

numeric matrix likelihoods

...

further arguments passed to Rmosek such as rtol

Details

The matrix A is structured as follows: A_ij = P(X_i | theta = t_j), where X_i is the i'th observation and t_j is the j'th set of parameters/grid-point.

Value

a vector containing the fitted prior