Uniform sampling s.t. linear constraints

Implementations of the Hit-And-Run and Gibbs sampling algorithms for sampling constraints A*x<=b.

Implementations of the Hit-And-Run and Gibbs sampling algorithms for the generation of uniformly distributed random vector x satisfying the system of linear inequalities A.x<=b.

The generated random vectors may be used to initialize a locally searching non-linear optimization algorithm in order to locate a global optimium through a multi-start optimization strategy.

It is assumed that A*x<=b is not unbounded. In order to use the samplers an initially feasible point x0 has to be specified (which can be found, e.g., by using a LP-solver).