Abstract
In this paper we propose a family of affine scaling interior point algorithms, called galpv4, using a primal approach, based on a large class of differential barrier functions. We show that these algorithms are in fact an extension and generalization of the classical affine scaling algorithm based on the well-known log barrier function. After carrying out a complete convergence analysis, we select some of these algorithms for comparison with the classical affine scaling algorithm, performed with the help of the familiar Netlib test set.
Acknowledgments
The anonymous referees are kindly acknowledged. Their comments greatly helped to improve the paper.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Self-concordant functions were introduced by Yu. E. Nesterov and A. S. Nemirowskii in [Citation18]. These functions are particularly useful in the analysis of Newton's method.
2 We recall that the idea to approximate a linear program by a nonlinear optimization problem is due originally to Courant [Citation19] in 1941 with a penalty function of exterior type and later to Frisch [Citation20], in 1955, when he introduced the logarithmic barrier function which is an interior penalty one. We recall also that the notion of interior penalty operators were introduced by Auslender [Citation21] in 1976 to generalize the concept of barrier functions.
3 An affine scaling algorithm was originally proposed by Dikin in 1967 [Citation22]. It was rediscovered by several researchers such as Barnes [Citation5] and Vanderbei et al. [Citation6] after Karmarkar [Citation23] proposed his famous projective scaling algorithm.
4 A good description can be found in for example the book of Roos, Terlaky and Vial [Citation24].
5 Note that the primal-dual affine scaling approach, proposed for the first time by Monteiro, Adler and Resende [Citation25], has also been the subject of many works such as Jansen, Roos and Terlaky [Citation26] and Potra [Citation27].