Entropic approach to interior point solution of linear programs

Abstract

We present a maxentropic approach for obtaining interior, suboptimal solutions to a linear programming problem (not necessarily in Karmarkar’s canonical form). Each suboptimal solution lies in the relative interior set of feasible solutions. We present an application to the problem of obtaining reconstructions of minimal weighted L 1-norm, of a function from the knowledge of a few of its Fourier coefficients.