|
Abstract
|
We present an adaptive full Nesterov-Todd step infeasible
interior-point method for semidefinite optimization. The proposed
algorithm requires two types of full Nesterov-Todd steps are called,
feasibility steps and centering steps, respectively. At each
iteration both feasibility and optimality are reduced exactly at the
same rate. In each iteration of the algorithm we use the largest
possible barrier parameter value $\theta$. The value $\theta$ varies
from iteration to iteration and it lies between the two values
$\frac{1}{4n}$ and $\frac{1}{5n}$,which results a faster algorithm.
|