lpsolve_over_eta_log: Solve LP over polytope described by log(eta)

In pierrejacob/montecarlodsm: Gibbs sampler for Dempster-Shafer inference in Categorical distributions

Description

Minimize linear program of the form t(objective_vector) * log(theta) over theta in the convex polytope of log(theta) such that log(theta_l) - log(theta_k) <= log(etak,l) and such that sum_k log (theta_k) = 0. Employs the lpSolveAPI package.

Usage

lpsolve_over_eta_log(eta, objective_vector)

Arguments

eta	A matrix of dimension K x K describing a convex polytope.
objective_vector	a vector giving coefficient of linear function (of log theta) to minimize.

Value

Evaluation of the objective function at its minimum.

Examples

## Not run: 
eta <- rejection_sampler(c(1,0,3,2))$etas
lpsolve_over_eta_log(eta, c(1,0,0,0))

## End(Not run)

pierrejacob/montecarlodsm documentation built on June 16, 2021, 1:06 p.m.