Abstract
Many efficient interior-point methods (IPMs) are based on the use of a self-concordant barrier function for the domain of the problem that has to be solved. Recently, a wide class of new barrier functions has been introduced in which the functions are not self-concordant, but despite this fact give rise to efficient IPMs. Here, we introduce the notion of locally self-concordant barrier functions and we prove that the new barrier functions are locally self-concordant. In many cases, the (local) complexity numbers of the new barrier functions along the central path are better than the complexity number of the logarithmic barrier function by a factor between 0.5 and 1.
Original language | American English |
---|---|
Journal | Iranian Journal of Operations Research |
Volume | 3 |
State | Published - Jan 1 2012 |
Keywords
- Kernel function
- Linear optimization
- Polynomial complexity
- Primal-dual interior-point method
- Self-concordance
- Self-dual embedding
DC Disciplines
- Education
- Mathematics