Schemes of involving dual variables in inverse barrier functions for problems of linear and convex programming

A new scheme of the method of inverse barrier functions is proposed for problems of linear and convex programming. The scheme is based on the idea of a parametric shifting of the constraints of the original problem, similarly to what was done in the method of modified Lagrange function for the usual quadratic penalty function. The description of the method, the proof of its convergence, and the results of numerical experiments are presented.