|
Abstract
|
After a brief introduction to Euclidean Jordan algebra, we present a new
corrector–predictor path-following interior-point algorithm for convex, quadratic, and
symmetric cone optimization. In each iteration, the algorithm involves two kind of
steps: a predictor (affine-scaling) step and a full Nesterov and Todd (centring) step.
Moreover, we derive the complexity for the algorithm, and we obtain the best-known
iteration bound for the small-update method
|