Properties of an interior embedding for solving nonlinear optimization problems

The paper presents an interior embedding of nonlinear optimization problems. This embedding satisfies a sufficient condition for the success of pathfollowing algorithms with jumps being applied to one-parametric optimization problems. The one-parametric problem obtained by the embedding is supposed to be regular in the sense of Jongen, Jonker and Twilt. This assumption is analyzed, and its genericity is proved in the space of the original optimization problems.